42986 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

DEPARTMENT OF TRANSPORTATION half a million barrels per day (2–3 Æ EPA: Docket Center (EPA/DC), EPA percent of total daily consumption, West, Room B102, 1301 Constitution National Highway Traffic Safety according to the Energy Information Avenue NW, Washington, DC, Attention Administration Administration) and would impact the Docket ID No. EPA–HQ–OAR–2018– global climate by 3/1000th of one degree 0283. Such deliveries are only accepted 49 CFR Parts 523, 531, 533, 536, and Celsius by 2100, also when compared to during the Docket’s normal hours of 537 the standards set forth in 2012. operation, and special arrangements DATES: Comments: Comments are should be made for deliveries of boxed ENVIRONMENTAL PROTECTION requested on or before October 23, 2018. information. Æ AGENCY Under the Paperwork Reduction Act, NHTSA: West Building, Ground comments on the information collection Floor, Rm. W12–140, 1200 New Jersey 40 CFR Parts 85 and 86 provisions must be received by the Avenue SE, Washington, DC 20590, [NHTSA–2018–0067; EPA–HQ–OAR–2018– Office of Management and Budget between 9 a.m. and 4 p.m. Eastern Time, 0283; FRL–9981–74–OAR] (OMB) on or before October 23, 2018. Monday through Friday, except Federal See the SUPPLEMENTARY INFORMATION holidays. RIN 2127–AL76; RIN 2060–AU09 section on ‘‘Public Participation,’’ Instructions: All submissions received below, for more information about must include the agency name and The Safer Affordable Fuel-Efficient docket number or Regulatory (SAFE) Vehicles Rule for Model Years written comments. Public Hearings: NHTSA and EPA Information Number (RIN) for this 2021–2026 Passenger Cars and Light rulemaking. All comments received will Trucks will jointly hold three public hearings in Washington, DC; the Detroit, MI area; be posted without change to http:// AGENCY: Environmental Protection and in the Los Angeles, CA area. The www.regulations.gov, including any Agency and National Highway Traffic agencies will announce the specific personal information provided. For Safety Administration. dates and addresses for each hearing detailed instructions on sending comments and additional information ACTION: Notice of proposed rulemaking. location in a supplemental Federal Register notice. The agencies will on the rulemaking process, see the SUMMARY: The National Highway Traffic accept oral and written comments to the ‘‘Public Participation’’ heading of the Safety Administration (NHTSA) and the rulemaking documents, and NHTSA SUPPLEMENTARY INFORMATION section of Environmental Protection Agency (EPA) will also accept comments to the Draft this document. are proposing the ‘‘Safer Affordable Environmental Impact Statement (DEIS) Docket: For access to the dockets to Fuel-Efficient (SAFE) Vehicles Rule for at these hearings. The hearings will start read background documents or Model Years 2021–2026 Passenger Cars at 10 a.m. local time and continue until comments received, go to http:// and Light Trucks’’ (SAFE Vehicles www.regulations.gov, and/or: everyone has had a chance to speak. See • Rule). The SAFE Vehicles Rule, if the SUPPLEMENTARY INFORMATION section For EPA: EPA Docket Center (EPA/ finalized, would amend certain existing on ‘‘Public Participation,’’ below, for DC), EPA West, Room 3334, 1301 Corporate Average Fuel Economy more information about the public Constitution Avenue NW, Washington, (CAFE) and tailpipe carbon dioxide hearings. DC 20460. The Public Reading Room is emissions standards for passenger cars open from 8:30 a.m. to 4:30 p.m., and light trucks and establish new ADDRESSES: You may send comments, Monday through Friday, excluding legal standards, all covering model years identified by Docket No. EPA–HQ– holidays. The telephone number for the 2021 through 2026. More specifically, OAR–2018–0283 and/or NHTSA–2018– Public Reading Room is (202) 566–1744. NHTSA is proposing new CAFE 0067, by any of the following methods: • For NHTSA: Docket Management • standards for model years 2022 through Federal eRulemaking Portal: http:// Facility, M–30, U.S. Department of 2026 and amending its 2021 model year www.regulations.gov. Follow the Transportation, West Building, Ground CAFE standards because they are no instructions for sending comments. Floor, Rm. W12–140, 1200 New Jersey • longer maximum feasible standards, and Fax: EPA: (202) 566–9744; NHTSA: Avenue SE, Washington, DC 20590. The EPA is proposing to amend its carbon (202) 493–2251. Docket Management Facility is open dioxide emissions standards for model • Mail: between 9 a.m. and 4 p.m. Eastern Time, years 2021 through 2025 because they Æ EPA: Environmental Protection Monday through Friday, except Federal are no longer appropriate and Agency, EPA Docket Center (EPA/DC), holidays. reasonable in addition to establishing Air and Radiation Docket, Mail Code FOR FURTHER INFORMATION CONTACT: new standards for model year 2026. The 28221T, 1200 Pennsylvania Avenue EPA: Christopher Lieske, Office of preferred alternative is to retain the NW, Washington, DC 20460, Attention Transportation and Air Quality, model year 2020 standards (specifically, Docket ID No. EPA–HQ–OAR–2018– Assessment and Standards Division, the footprint target curves for passenger 0283. In addition, please mail a copy of Environmental Protection Agency, 2000 cars and light trucks) for both programs your comments on the information Traverwood Drive, Ann Arbor, MI through model year 2026, but comment collection provisions for the EPA 48105; telephone number: (734) 214– is sought on a range of alternatives proposal to the Office of Information 4584; fax number: (734) 214–4816; discussed throughout this document. and Regulatory Affairs, Office of email address: lieske.christopher@ Compared to maintaining the post-2020 Management and Budget (OMB), Attn: epa.gov, or contact the Assessment and standards set forth in 2012, current Desk Officer for EPA, 725 17th St. NW, Standards Division, email address: estimates indicate that the proposed Washington, DC 20503. [email protected]. NHTSA: James SAFE Vehicles Rule would save over Æ NHTSA: Docket Management Tamm, Office of Rulemaking, Fuel 500 billion dollars in societal costs and Facility, M–30, U.S. Department of Economy Division, National Highway reduce highway fatalities by 12,700 Transportation, West Building, Ground Traffic Safety Administration, 1200 New lives (over the lifetimes of vehicles Floor, Rm. W12–140, 1200 New Jersey Jersey Avenue SE, Washington, DC through MY 2029). U.S. fuel Avenue SE, Washington, DC 20590. 20590; telephone number: (202) 493– consumption would increase by about • Hand Delivery: 0515.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00002 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42987

5 SUPPLEMENTARY INFORMATION: CO2—‘‘cause[s] or contribute[s] to air section 202(a) of the Clean Air Act. The I. Overview of Joint NHTSA/EPA Proposal pollution which may reasonably be regulations also required the issuance of II. Technical Foundation for NPRM Analysis anticipated to endanger public health or a draft Technical Assessment Report 3 III. Proposed CAFE and CO2 Standards for welfare.’’ NHTSA and EPA are (TAR) by November 15, 2017, an MYs 2021–2026 proposing these standards concurrently opportunity for public comment on the IV. Alternative CAFE and GHG Standards because tailpipe CO2 emissions draft TAR, and, before making a Final Considered for MYs 2021/22–2026 standards are directly and inherently Determination, an opportunity for V. Proposed Standards, the Agencies’ related to fuel economy standards,4 and public comment on whether the GHG Statutory Obligations, and Why the if finalized, these rules would apply Agencies Propose To Choose Them Over standards for MY 2022–2025 remain the Alternatives concurrently to the same fleet of appropriate. In July 2016, the draft TAR VI. Preemption of State and Local Laws vehicles. By working together to was issued for public comment jointly VII. Impacts of the Proposed CAFE and CO2 develop these proposals, the agencies by the EPA, NHTSA, and the California Standards reduce regulatory burden on industry Air Resources Board (CARB).6 VIII. Impacts of Alternative CAFE and CO2 and improve administrative efficiency. Following the draft TAR, EPA published Standards Considered for MYs 2021/22– Consistent with both agencies’ a Proposed Determination for public 2026 statutes, this proposal is entirely de IX. Vehicle Classification comment on December 6, 2016 and X. Compliance and Enforcement novo, based on an entirely new analysis provided less than 30 days for public XI. Public Participation reflecting the best and most up-to-date comments over major holidays.7 EPA XII. Regulatory Notices and Analyses information available to the agencies at published the January 2017 the time of this rulemaking. The Determination on EPA’s website and I. Overview of Joint NHTSA/EPA agencies worked together in 2012 to regulations.gov finding that the MY Proposal develop CAFE and CO2 standards for 2022–2025 standards remained MYs 2017 and beyond; in that A. Executive Summary appropriate.8 rulemaking action, EPA set CO2 In this notice, the National Highway standards for MYs 2017–2025, while On March 15, 2017, President Trump Traffic Safety Administration (NHTSA) NHTSA set final CAFE standards for announced a restoration of the original and the Environmental Protection MYs 2017–2021 and also put forth mid-term review timeline. The Agency (EPA) (collectively, ‘‘the ‘‘augural’’ CAFE standards for MYs President made clear in his remarks, agencies’’) are proposing the ‘‘Safer 2022–2025, consistent with EPA’s CO2 ‘‘[i]f the standards threatened auto jobs, Affordable Fuel-Efficient (SAFE) standards for those model years. EPA’s then commonsense changes’’ would be Vehicles Rule for Model Years 2021– CO2 standards for MYs 2022–2025 were made in order to protect the economic 2026 Passenger Cars and Light Trucks’’ subject to a ‘‘mid-term evaluation,’’ by viability of the U.S. automotive (SAFE Vehicles Rule). The proposed which EPA bound itself through industry.’’ 9 In response to the SAFE Vehicles Rule would set regulation to re-evaluate the CO2 President’s direction, EPA announced in Corporate Average Fuel Economy standards for those model years and to a March 22, 2017, Federal Register (CAFE) and carbon dioxide (CO2) undertake to develop new CO2 notice, its intention to reconsider the emissions standards, respectively, for standards through a regulatory process Final Determination of the mid-term passenger cars and light trucks if it concluded that the previously evaluation of GHGs emissions standards manufactured for sale in the United finalized standards were no longer for MY 2022–2025 light-duty vehicles.10 States in model years (MYs) 2021 appropriate. EPA regulations on the The Administrator stated that EPA 1 through 2026. CAFE and CO2 standards mid-term evaluation process required would coordinate its reconsideration have the power to transform the vehicle EPA to issue a Final Determination no with the rulemaking process to be fleet and affect Americans’ lives in later than April 1, 2018 on whether the undertaken by NHTSA regarding CAFE significant, if not always immediately GHG standards for MY 2022–2025 light- standards for cars and light trucks for obvious, ways. The proposed SAFE duty vehicles remain appropriate under the same model years. Vehicles Rule seeks to ensure that On August 21, 2017, EPA published a government action on these standards is 3 42 U.S.C. 7521, see also 74 FR 66495 (Dec. 15, notice in the Federal Register appropriate, reasonable, consistent with 2009) (‘‘Endangerment and Cause or Contribute announcing the opening of a 45-day law, consistent with current and Findings for Greenhouse Gases under Section 202(a) of the Clean Air Act’’). public comment period and inviting foreseeable future economic realities, 4 See, e.g., 75 FR 25324, at 25327 (May 7, 2010) stakeholders to submit any additional and supported by a transparent (‘‘The National Program is both needed and comments, data, and information they assessment of current facts and data. possible because the relationship between believed were relevant to the The agencies must act to propose and improving fuel economy and reducing tailpipe CO2 emissions is a very direct and close one. The Administrator’s reconsideration of the finalize these standards and do not have amount of those CO emissions is essentially discretion to decline to regulate. 2 constant per gallon combusted of a given type of 5 40 CFR 86.1818–12(h)(1); see also 77 FR 62624 fuel. Thus, the more fuel efficient a vehicle is, the Congress requires NHTSA to set CAFE (Oct. 15, 2012). 2 less fuel it burns to travel a given distance. The less standards for each model year. 6 81 FR 49217 (Jul. 27, 2016). fuel it burns, the less CO2 it emits in traveling that Congress also requires EPA to set distance. [citation omitted] While there are 7 81 FR 87927 (Dec. 6, 2016). emissions standards for light-duty emission control technologies that reduce the 8 Docket item EPA–HQ–OAR–2015–0827–6270 vehicles if EPA has made an pollutants (e.g., carbon monoxide) produced by (EPA–420–R–17–001). This conclusion generated a ‘‘endangerment finding’’ that the imperfect combustion of fuel by capturing or significant amount of public concern. See, e.g., converting them to other compounds, there is no Letter from Auto Alliance to Scott Pruitt, pollutant in question—in this case, such technology for CO . Further, while some of 2 Administrator, Environmental Protection Agency those pollutants can also be reduced by achieving 1 NHTSA sets CAFE standards under the Energy a more complete combustion of fuel, doing so only (Feb. 21, 2017); Letter from Global Automakers to Policy and Conservation Act of 1975 (EPCA), as increases the tailpipe emissions of CO2. Thus, there Scott Pruitt, Administrator, Environmental amended by the Energy Independence and Security is a single pool of technologies for addressing these Protection Agency (Feb. 21, 2017). Act of 2007 (EISA). EPA sets CO2 standards under twin problems, i.e., those that reduce fuel 9 See https://www.whitehouse.gov/briefings- the Clean Air Act (CAA). consumption and thereby reduce CO2 emissions as statements/remarks-president-trump-american- 2 49 U.S.C. 32902. well.’’) center-mobility-detroit-mi/. 10 82 FR 14671 (Mar. 22, 2017).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00003 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 42988 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

January 2017 Determination.11 EPA held suggest that the CAFE standards below presents those alternatives. We a public hearing in Washington DC on previously set for MY 2021 are no note further that prior to MY 2021, CO2 12 September 6, 2017. EPA received longer maximum feasible, and the CO2 targets include adjustments reflecting more than 290,000 comments in standards previously set for MY 2021 the use of automotive refrigerants with response to the August 21, 2017 are no longer appropriate. Agencies reduced global warming potential notice.13 always have authority under the (GWP) and/or the use of technologies EPA has since concluded, based on Administrative Procedure Act to revisit that reduce the refrigerant leaks, and more recent information, that those previous decisions in light of new facts, optionally offsets for nitrous oxide and standards are no longer appropriate.14 as long as they provide notice and an methane emissions. In the interests of NHTSA’s ‘‘augural’’ CAFE standards for opportunity for comment, and it is harmonizing with the CAFE program, MYs 2022–2025 were not final in 2012 plainly the best practice to do so when EPA is proposing to exclude air because Congress prohibits NHTSA changed circumstances so warrant.16 conditioning refrigerants and leakage, from finalizing new CAFE standards for Thus, the proposed SAFE Vehicles and nitrous oxide and methane more than five model years in a single Rule would maintain the CAFE and CO2 emissions for compliance with CO2 rulemaking.15 NHTSA was therefore standards applicable in MY 2020 for standards after model year 2020 but obligated from the beginning to MYs 2021–2026, while taking comment seeks comment on whether to retain undertake a new rulemaking to set on a wide range of alternatives, these element, and reinsert A/C leakage CAFE standards for MYs 2022–2025. including different stringencies and offsets, and remain disharmonized with The proposed SAFE Vehicles Rule retaining existing CO2 standards and the the CAFE program. EPA also seeks 17 begins the rulemaking process for both augural CAFE standards. Table I–4 comment on whether to change existing agencies to establish new standards for methane and nitrous oxide standards MYs 2022–2025 passenger cars and light 16 See FCC v. Fox Television, 556 U.S. 502 (2009). that were finalized in the 2012 rule. 17 Note: This does not mean that the miles per trucks. Standards are concurrently being gallon and grams per mile levels that were Specifically, EPA seeks information proposed for MY 2026 in order to estimated for the MY 2020 fleet in 2012 would be from the public on whether those provide regulatory stability for as many the ‘‘standards’’ going forward into MYs 2021–2026. existing standards are appropriate, or years as is legally permissible for both Both NHTSA and EPA set CAFE and CO2 standards, whether they should be revised to be respectively, as mathematical functions based on agencies together. vehicle footprint. These mathematical functions less stringent or more stringent based on Separately, the proposed SAFE that are the actual standards are defined as ‘‘curves’’ any updated data. Vehicles Rule includes revised that are separate for passenger cars and light trucks, While actual requirements will standards for MY 2021 passenger cars under which each vehicle manufacturer’s ultimately vary for automakers compliance obligation varies depending on the and light trucks. The information now footprints of the cars and trucks that it ultimately depending upon their individual fleet available and the current analysis produces for sale in a given model year. It is the mix of vehicles, many stakeholders will MY 2020 CAFE and CO2 curves which we propose likely be interested in the current would continue to apply to the passenger car and 11 82 FR 39551 (Aug. 21, 2017). estimate of what the MY 2020 CAFE and light truck fleets for MYs 2021–2026. The mpg and 12 82 FR 39976 (Aug. 23, 2017). g/mi values which those curves would eventually CO2 curves would translate to, in terms 13 The public comments, public hearing require of the fleets in those model years would be of miles per gallon (mpg) and grams per transcript, and other information relevant to the known for certain only at the ends of each of those mile (g/mi), in MYs 2021–2026. These Mid-term Evaluation are available in docket EPA– model years. While it is convenient to discuss estimates are shown in the following HQ–OAR–2015–0827. CAFE and CO2 standards as a set ‘‘mpg,’’ ‘‘g/mi,’’ 14 83 FR 16077 (Apr. 2, 2018). or ‘‘mpg-e’’ number, attempting to define those tables. 15 49 U.S.C. 32902. values today will end up being inaccurate. BILLING CODE 4910–59–P

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00004 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42989

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00005 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.000 EP24AU18.001 42990 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

BILLING CODE 4910–59–C air conditioning refrigerants and leakage alternatives and have specifically In the tables above, estimated (and, optionally, offsets for nitrous modeled eight alternatives (including required CO2 increases between MY oxide and methane emissions) after the proposed alternative) and the 2020 and MY 2021 because, again, EPA model year 2020. current requirements (i.e., baseline/no- is proposing to exclude CO2-equivalent As explained above, the agencies are action). The modeled alternatives are emission improvements associated with taking comment on a wide range of provided below:

Summary of Rationale Light-Duty Vehicle Greenhouse Gas evaluation’’ process, the agencies have Since finalizing the agencies’ previous Emission and Corporate Average Fuel gathered new information, and have joint rulemaking in 2012 titled ‘‘Final Economy Standards,’’ and even since performed new analysis. That new Rule for Model Year 2017 and Later EPA’s 2016 and early 2017 ‘‘mid-term information and analysis has led the

18 Carbon dioxide equivalent of air conditioning calculated using the Global Warming Potential refrigerant leakage, nitrous oxide and methane refrigerant leakage, nitrous oxide and methane (GWP) of each of the emissions. emissions would no longer be able to be included emissions are included for compliance with the 19 Beginning in MY 2021, the proposal provides with the tailpipe CO2 for compliance with tailpipe EPA standards for all MYs under the baseline/no that the GWP equivalents of air conditioning CO2 standards. action alternative. Carbon dioxide equivalent is

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00006 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.002 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42991

agencies to the tentative conclusion that phenomenon. First, as discussed, there that precluded observation in the holding standards constant at MY 2020 is a known pool of technologies for known (MYs 2008 and 2010) fleets, levels through MY 2026 is maximum improving fuel economy and reducing likely leading to double counting in feasible, for CAFE purposes, and CO2 emissions. Many of these cases where the known vehicles already appropriate, for CO2 purposes. technologies, when actually reflected the assumed efficiency Technologies have played out implemented on vehicles, can be used improvement. For example, the agencies differently in the fleet from what the to improve other vehicle attributes such assumed that transmission ‘‘shift agencies assumed in 2012. as ‘‘zero to 60’’ performance, towing, optimizers’’ would be available and The technology to improve fuel and hauling, etc., either instead of or in fairly widely used in MYs 2017–2025, economy and reduce CO2 emissions has addition to improving fuel economy and but involving software controls, a not changed dramatically since prior reducing CO2 emissions. As one ‘‘technology’’ not defined in a way that analyses were conducted: A wide example, a V6 engine can be would be observed in the fleet (unlike, variety of technologies are still available turbocharged and downsized so that it for example, a dual clutch to accomplish the goals of the programs, consumes only as much fuel as an inline transmission). and a wide variety of technologies 4-cylinder engine, or it can be would likely be used by industry to turbocharged and downsized so that it To be clear, this is no one’s ‘‘fault’’— accomplish these goals. There remains consumes less fuel than it would the CAFE and CO2 standards do not no single technology that the majority of originally have consumed (but more require manufacturers to use particular vehicles made by the majority of than the inline 4-cylinder would) while technologies in particular ways, and manufacturers can implement at low also providing more low-end torque. As both agencies’ past analyses generally cost without affecting other vehicle another example, a vehicle can be sought to illustrate technology paths to attributes that consumers value more lightweighted so that it consumes less compliance that were assumed to be as than fuel economy and CO2 emissions. fuel than it would originally have cost-effective as possible. If Even when used in combination, consumed, or so that it consumes the manufacturers choose different paths for technologies that can improve fuel same amount of fuel it would originally reasons not accounted for in regulatory economy and reduce CO2 emissions still have consumed but can carry more analysis, or choose to use technologies need to (1) actually work together and content, like additional safety or differently from what the agencies (2) be acceptable to consumers and infotainment equipment. Manufacturers previously assumed, it does not avoid sacrificing other vehicle attributes employing ‘‘fuel-saving/emissions- necessarily mean that the analyses were while also avoiding undue increases in reducing’’ technologies in the real world unreasonable when performed. It does vehicle cost. Optimism about the costs make decisions regarding how to mean, however, that the fleet ought to and effectiveness of many individual employ that technology such that fewer be reflected as it stands today, with the technologies, as compared to recent than 100% of the possible fuel-saving/ technology it has and as that technology prior rounds of rulemaking, is emissions-reducing benefits result. They has been used, and consider what somewhat tempered; a clearer do this because this is what consumers technology remains on the table at this understanding of what technologies are want, and more so than exclusively fuel point, whether and when it can already on vehicles in the fleet and how economy improvements. realistically be available for widespread they are being used, again as compared This makes actual fuel economy gains use in production, and how much it to recent prior rounds of rulemaking, more expensive. would cost to implement. means that technologies that previously Thus, even though the technologies Incremental additional fuel economy appeared to offer significant ‘‘bang for may be largely the same, previous benefits are subject to diminishing the buck’’ may no longer do so. assumptions about how much fuel can returns. Additionally, in light of the reality that be saved or how much emissions can be vehicle manufacturers may choose the reduced by employing various As fleet-wide fuel efficiency improves relatively cost-effective technology technologies may not have played out as and CO2 emissions are reduced, the option of vehicle lightweighting for a prior analyses suggested, meaning that incremental benefit of continuing to wide array of vehicles and not just the previous assumptions about how much improve/reduce inevitably decreases. largest and heaviest, it is now it would cost to save that much fuel or This is because, as the base level of fuel recognized that as the stringency of reduce that much in emissions fall economy improves, fewer gallons are standards increases, so does the correspondingly short. For example, the saved from subsequent incremental likelihood that higher stringency will agencies assumed in the 2010 final rule improvements. Put simply, a one mpg increase on-road fatalities. As it turns that dual clutch transmissions would be increase for vehicles with low fuel out, there is no such thing as a free widely used to improve fuel economy economy will result in far greater lunch.20 due to expectations of strong savings than an identical 1 mpg increase Technology that can improve both effectiveness and very low cost: In for vehicles with higher fuel economy, fuel economy and/or performance may practice, dual clutch transmissions had and the cost for achieving a one-mpg not be dedicated solely to fuel economy. significant customer acceptance issues, increase for low fuel economy vehicles As fleet-wide fuel efficiency has and few manufacturers employ them in is far less than for higher fuel economy improved over time, additional the U.S. market today.21 The agencies vehicles. This means that improving improvements have become both more included some ‘‘technologies’’ in the fuel economy is subject to diminishing complicated and more costly. There are 2012 final rule analysis that were returns. Annual fuel consumption can two primary reasons for this defined ambiguously and/or in ways be calculated as follows:

20 Mankiw, N. Gregory, Principles of e.g., Steve Lehto, ‘‘What you need to know about what-you-need-to-know-about-the-proposed- Macroeconomics, Sixth Edition, 2012, at 4. the settlement for Ford Powershift owners,’’ Road settlement-for-ford-powershift-owners/ (last 21 In fact, one manufacturer saw enough customer and Track, Oct. 19, 2017. Available at https:// accessed Jul. 2, 2018). pushback that it launched a buyback program. See, www.roadandtrack.com/car-culture/a10316276/

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00007 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 42992 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

For purposes of illustration, assume a as large. Going from 40 to 50 mpg would reference case projections in the Energy vehicle owner who drives a light vehicle save only 75 gallons/year. Yet, each Information Administration’s (EIA’s) 15,000 miles per year (a typical additional fuel economy improvement Annual Energy Outlook (AEO 2017 and assumption for analytical purposes).22 If becomes much more expensive as the AEO 2018) do not indicate particularly that owner trades in a vehicle with fuel low-hanging fruit of low-cost high fuel prices in the foreseeable economy of 15 mpg for one with fuel technological improvement options are future, given underlying assumptions,24 economy of 20 mpg, the owner’s annual picked.23 Automakers, who must and (2) as the baseline level of fuel fuel consumption would drop from nonetheless continue adding technology economy continues to increase, the 1,000 gallons to 750 gallons—saving 250 to improve fuel economy and reduce marginal cost of the next gallon saved gallons annually. If, however, that CO2 emissions, will either sacrifice similarly increases with the cost of the owner were to trade in a vehicle with other performance attributes or raise the technologies required to meet the fuel economy of 30 mpg for one with price of vehicles—neither of which is savings. The following figure illustrates fuel economy of 40 mpg, the owner’s attractive to most consumers. the fact that fuel savings and annual gasoline consumption would If fuel prices are high, the value of corresponding avoided costs diminish drop from 500 gallons/year to 375 those gallons may be enough to offset with increasing fuel economy, showing gallons/year—only 125 gallons even the cost of further fuel economy the same basic pattern as a 2014 though the mpg improvement is twice improvements, but (1) the most recent illustration developed by EIA.25

22 A different vehicle-miles-traveled (VMT) percent improvement is 60 g/mi, so that the vehicle every joint CAFE and CO2 rulemaking since 2009) assumption would change the absolute numbers in would emit 240 g/mi. At 180 g/mi, a 20% has applied fuel price projections from EIA’s the example, but would not change the improvement is 36 g/mi, so the vehicle would get Annual Energy Outlook (AEO). AEO projections, mathematical principles. Today’s analysis uses 144 g/mi. In order to continue achieving similarly documentation, and underlying data and estimates mileage accumulation schedules that average about large (on an absolute basis) emissions reductions, are available at https://www.eia.gov/outlooks/aeo/. 15,000 miles annually over the first six years of mathematics require the percentage reduction to vehicle operation. continue increasing. 25 Today in Energy: Fuel economy improvements 23 The examples in the text above are presented 24 The U.S. Energy Information Administration show diminishing returns in fuel savings, U.S. in mpg because that is a metric which should be (EIA) is the statistical and analytical agency within Energy Information Administration (Jul. 11, 2014), readily understandable to most readers, but the the U.S. Department of Energy (DOE). EIA is the https://www.eia.gov/todayinenergy/detail.php?id= example would hold true for grams of CO2 per mile nation’s premiere source of energy information, and 17071. as well. If a vehicle emits 300 g/mi CO2, a 20 every fuel economy rulemaking since 2002 (and

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00008 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.003 EP24AU18.004 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42993

This effect is mathematical in nature value fuel economy and low CO2 increase the overall average CO2 and long-established, but when emissions above other attributes, and emission rates of the new vehicle fleet. combined with relatively low fuel prices thanks in part to CAFE and CO2 Consumers are also demonstrating a potentially through 2050, and the standards, they have a plentiful preference for more powerful engines likelihood that a large majority of selection of high-fuel economy and low and vehicles with higher seating American consumers could CO2-emitting vehicles to choose from, positions and ride height (and consequently continue to place a higher but those consumers represent a accompanying mass increase relative to value on vehicle attributes other than relatively small percentage of buyers. footprint) 30—all of which present fuel economy, it makes manufacturers’ Changed petroleum market has challenges for achieving increased fuel ability to sell light vehicles with ever- supported a shift in consumer economy levels and lower CO2 emission higher fuel economy and ever-lower preferences. rates. carbon dioxide emissions increasingly In 2012, the agencies projected fuel The Consequence of Unreasonable difficult. Put more simply, if gas is prices would rise significantly, and the Fuel Economy and CO2 Standards: cheap and each additional improvement United States would continue to rely Increased vehicle prices keep consumers saves less gas anyway, most consumers heavily upon imports of oil, subjecting in older, dirtier, and less safe vehicles. would rather spend their money on the country to heightened risk of price Consumers tend to avoid purchasing attributes other than fuel economy when shocks.28 Things have changed things that they neither want or need. they are considering a new vehicle significantly since 2012, with fuel prices The analysis in today’s proposal moves purchase, whether that is more safety significantly lower than anticipated, and closer to being able to represent this fact technology, a better infotainment projected to remain low through 2050. through an improved model for vehicle package, a more powerful powertrain, or Furthermore, the global petroleum scrappage rates. While neither this nor other features (or, indeed, they may market has shifted dramatically with the a sales response model, also included in prefer to spend the savings on United States taking advantage of its today’s analysis, nor the combination of something other than automobiles). own oil supplies through technological the two, are consumer choice models, Manufacturers trying to sell consumers advances that allow for cost-effective today’s analysis illustrates market-wide more fuel economy in such extraction of shale oil. The U.S. is now impacts on the sale of new vehicles and circumstances may convince consumers the world’s largest oil producer and the retention of used vehicles. Higher who place weight on efficiency and expected to become a net petroleum vehicle prices, which result from more- reduced carbon emissions, but exporter in the next decade.29 stringent fuel economy standards, have consumers decide for themselves what At least partially in response to lower an effect on consumer purchasing attributes are worth to them. And while fuel prices, consumers have moved decisions. As prices increase, the some contend that consumers do not more heavily into crossovers, sport market-wide incentive to extract sufficiently consider or value future fuel utility vehicles and pickup trucks, than additional travel from used vehicles savings when making vehicle anticipated at the time of the last increases. The average age of the in- purchasing decisions,26 information rulemaking. Because standards are service fleet has been increasing, and regarding the benefits of higher fuel based on footprint and specified when fleet turnover slows, not only economy has never been made more separately for passenger cars and light does it take longer for fleet-wide fuel readily available than today, with a host trucks, these shifts do not necessarily economy and CO2 emissions to improve, of online tools and mandatory pose compliance challenges by but also safety improvements, criteria prominent disclosures on new vehicles themselves, but they tend to reduce the pollutant emissions improvements, on the Monroney label showing fuel overall average fuel economy rates and many other vehicle attributes that also savings compared to average vehicles. provide societal benefits take longer to This is not a question of ‘‘if you build increasing vehicle footprint size in order to get be reflected in the overall U.S. fleet as ‘‘easier’’ CAFE and CO2 standards. This well because of reduced turnover. it, they will come.’’ Despite the misunderstands, somewhat, how the footprint- widespread availability of fuel economy based standards work. While it is correct that larger- Raising vehicle prices too far, too fast, information, and despite manufacturers footprint vehicles have less stringent ‘‘targets,’’ the such as through very stringent fuel difficulty of compliance rests in how far above or economy and CO2 emissions standards building and marketing vehicles with below those vehicles are as compared to their higher fuel economy and increasing (especially considering that, on a fleet- targets, and more specifically, whether the wide basis, new vehicle sales and their offerings of hybrid and electric manufacturer is selling so many vehicles that are far vehicles, in the past several years as gas short of their targets that they cannot average out turnover do not appear strongly to compliant levels through other vehicles sold that responsive to fuel economy), has effects prices have remained low, consumer beat their targets. For example, under the CAFE preferences have shifted markedly away beyond simply a slowdown in sales. program, a manufacturer building a fleet of larger- Improvements over time have better from higher-fuel-economy smaller and footprint vehicles may have an objectively lower longer-term effects simply by not midsize passenger vehicles toward mpg-value compliance obligation than a alienating consumers, as compared to crossovers and truck-based utility manufacturer building a more mixed fleet, but it may still be more challenging for the first great leaps forward that drive people out vehicles.27 Some consumers plainly manufacturer to reach its compliance obligation if of the new car market or into vehicles it is selling only very-low-mpg variants at any given 26 In docket numbers EPA–HQ–OAR–2015–0827 footprint. There is only so much that increasing that do not meet their needs. The and NHTSA–2016–0068, see comments submitted footprint makes it ‘‘easier’’ for a manufacturer to industry has achieved tremendous gains by, e.g., Consumer Federation of America (NHTSA– reach compliance. in fuel economy over the past decade, 2016–0068–0054, at p. 57, et seq.) and the 28 The 2012 final rule analysis relied on the and these increases will continue at Environmental Defense Fund (EPA–HQ–OAR– Energy Information Administration’s Annual 2015–0827–4086, at p. 18, et seq.). Energy Outlook 2012 Early Release, which assumed least through 2020. 27 Carey, N. Lured by rising SUV sales, significantly higher fuel prices than the AEO 2017 Along with these gains, there have automakers flood market with models, Reuters (or AEO 2018) currently available. See 77 FR 62624, also been tremendous increases in (Mar. 29, 2018), available at https:// 62715 (Oct. 15, 2012) for the 2012 final rule’s vehicle prices, as new vehicles become www.reuters.com/article/us-autoshow-new-york- description of the fuel price estimates used. increasingly unaffordable—with the suvs/lured-by-rising-suv-sales-automakers-flood- 29 Annual Energy Outlook 2018, U.S. Energy market-with-models-idUSKBN1H50KI (last accessed Information Administration, at 53 (Feb. 6, 2018), average new vehicle transaction price Jun. 13, 2018). Many commentators have recently https://www.eia.gov/outlooks/aeo/pdf/ argued that manufacturers are deliberately AEO2018.pdf. 30 See id.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00009 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 42994 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

recently exceeding $36,000—up by allow a consumer to keep their monthly changed, the analytical methods and more than $3,000 since 2014 alone.31 In payment affordable but can have serious inputs have been updated (including fact, a recent independent study potential financial consequences. updates to address issues still present in indicated that the average new car price Longer-term financing leads (generally) analyses published in 2016, 2017, and is unaffordable to median-income to higher interest rates, larger finance early 2018), and today, the analysis families in every metropolitan region in charges and total consumer costs, and a suggests that, compared to the proposed the United States except one: longer period of time with negative standards today, the previously-issued 32 Washington, DC. That analysis used equity. In 2012, the agencies expected standards would increase average the historically accepted approach that prices to increase under the standards vehicle prices by about $2,100. While consumers should make a down- announced at that time. The agencies today’s estimate is similar in magnitude payment of at least 20% of a vehicle’s estimated that, compared to a to the 2012 estimate, it is relative to a purchase price, finance for no longer continuation of the model year 2016 than four years, and make payments of standards, the standards issued through baseline that includes increases in 10% or less of the consumer’s annual model year 2025 would eventually stringency between MY 2016 and MY income to car payments and insurance. increase average prices by about $1,500– 2020. Compared to leaving vehicle But the market looks nothing like that $1,800.34 35 36 Circumstances have technology at MY 2016 levels, today’s these days, with average financing terms analysis shows the previously-issued of 68 months, and an increasing www.valuepenguin.com/auto-loans/average-auto- standards through model year 2025 proportion exceeding 72 or even 84 loan-interest-rates (last accessed Jun. 15, 2018). could eventually increase average 34 months.33 Longer financing terms may 77 FR 62624, 62666 (Oct. 15, 2012). vehicle prices by approximately $2,700. 35 The $1,500 figure reported in 2012 by NHTSA reflected application of carried-forward credits in A pause in continued increases in fuel 31 See, e.g., Average New-Car Prices Rise Nearly model year 2025, rather than an achieved CAFE economy standards, and cost increases 4 Percent for January 2018 On Shifting Sales Mix, level that could be sustainably compliant beyond attributable thereto, is appropriate. According To Kelley Blue Book, Kelley Blue Book, 2025 (with standards remaining at 2025 levels). As https://mediaroom.kbb.com/2018-02-01-Average- for the 2016 draft TAR, NHTSA has since updated New-Car-Prices-Rise-Nearly-4-Percent-For-January- its modeling approach to extend far enough into the 2018-On-Shifting-Sales-Mix-According-To-Kelley- future that any unsustainable credit deficits are explicitly for multiyear planning and credit Blue-Book (last accessed Jun. 15, 2018). eliminated. Like analyses published by EPA in banking. 32 Bell, C. What’s an ‘affordable’ car where you 2016, 2017, and early 2018, the $1,800 figure 36 live? The answer may surprise you, Bankrate.com reported in 2012 by EPA did not reflect either While EPA did not refer to the reported $1,800 (Jun. 28, 2017), available at https:// simulation of manufacturers’ multiyear plans to as an estimate of the increase in average prices, www.bankrate.com/auto/new-car-affordability- progress from the initial MY 2008 fleet to the MY because EPA did not assume that manufacturers survey/ (last accessed Jun. 15, 2018). 2025 fleet or any accounting for manufacturers’ would reduce profit margins, the $1,800 estimate is 33 Average Auto Loan Interest Rates: 2018 Facts potential application of banked credits. Today’s appropriately interpreted as an estimate of the and Figures, ValuePenguin, available at https:// analysis of both CAFE and CO2 standards accounts average increase in vehicle prices.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00010 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42995

Preferred Alternative technological feasibility and economic improved fleet turnover as more For all of these reasons, the agencies practicability. The analysis suggests consumers will be able to afford newer are proposing to maintain the MY 2020 that, compared to the standards issued and safer vehicles. previously for MYs 2021–2025, today’s Recent NHTSA analysis shows that fuel economy and CO2 emissions standards for MYs 2021–2026. Our goal proposed rule will eventually (by the the proportion of passengers killed in a is to establish standards that promote early 2030s) increase U.S. petroleum vehicle 18 or more model years old is consumption by about 0.5 million nearly double that of a vehicle three both energy conservation and safety, in 39 light of what is technologically feasible barrels per day—about two to three model years old or newer. As the and economically practicable, as percent of projected total U.S. average car on the road is approaching directed by Congress. consumption. While significant, this 12 years old, apparently the oldest in additional petroleum consumption is, our history,40 major safety benefits will Energy Conservation from an economic perspective, dwarfed occur by reducing fleet age. Other safety EPCA requires that NHTSA, when by the cost savings also projected to benefits will occur from other areas determining the maximum feasible result from today’s proposal, as such as avoiding the increased driving levels of CAFE standards, consider the indicated by the consideration of net need of the Nation to conserve energy. benefits appearing below. 39 Passenger Vehicle Occupant Injury Severity by However, EPCA also requires that Vehicle Age and Model Year in Fatal Crashes, Safety Benefits From Preferred Traffic Safety Facts Research Note, DOT HS 812 NHTSA consider other factors, such as Alternative 528. Washington, DC: National Highway Traffic Safety Administration. April 2018. 37 Data on new vehicle prices are from U.S. Today’s proposed rule is anticipated 40 See, e.g., IHS Markit, Vehicles Getting Older: Bureau of Economic Analysis, National Income and Average Age of Light Cars and Trucks in U.S. Rises Product Accounts, Supplemental Table 7.2.5S, Auto to prevent more than 12,700 on-road Again in 2016 to 11.5 years, IHS Markit Says, IHS and Truck Unit Sales, Production, Inventories, fatalities 38 and significantly more Markit (Nov. 22, 2016), http://news.ihsmarkit.com/ Expenditures, and Price (https://www.bea.gov/ injuries as compared to the standards press-release/automotive/vehicles-getting-older- iTable/iTable.cfm?reqid=19&step=2#reqid= set forth in the 2012 final rule over the average-age-light-cars-and-trucks-us-rises-again-201 19&step=3&isuri=1&1921=underlying&1903=2055, (‘‘. . . consumers are continuing the trend of last accessed Jul. 20, 2018). Median Household lifetimes of vehicles as more new, safer holding onto their vehicles longer than ever. As of Income data are from U.S. Census Bureau, Table A– vehicles are purchased than the current the end of 2015, the average length of ownership 1, Households by Total Money Income, Race, and (and augural) standards. A large portion measured a record 79.3 months, more than 1.5 Hispanic Origin of Householder: 1967 to 2016 of these safety benefits will come from months longer than reported in the previous year. (https://www.census.gov/data/tables/2017/demo/ For used vehicles, it is nearly 66 months. Both are income-poverty/p60-259.html, last accessed Jul. 20, significantly longer lengths of ownership since the 2018). 38 Over the lifetime of vehicles through MY 2029. same measure a decade ago.’’).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00011 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.005 42996 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

that would otherwise result from higher The Preferred Alternative Would Have reducing corresponding emissions. On fuel efficiency (known as the rebound Negligible Environmental Impacts on the other hand, increasing fuel effect) and avoiding the mass reductions Air Quality consumption would increase emissions in passenger cars that might otherwise Improving fleet turnover will result in resulting from petroleum refining and be required to meet the standards consumers getting into newer and related ‘‘upstream’’ processes. Our established in 2012.41 Together these cleaner vehicles, accelerating the rate at analysis shows that none of the and other factors lead to estimated which older, more-polluting vehicles regulatory alternatives considered in annual fatalities under the proposed are removed from the roadways. Also, this proposal would noticeably impact standards that are significantly reducing fuel economy (relative to net emissions of smog-forming or other reduced 42 relative to those that would levels that would occur under ‘‘criteria’’ or toxic air pollutants, as occur under current (and augural) previously-issued standards) would illustrated by the following graph. That standards. increase the marginal cost of driving said, the resultant tailpipe emissions newer vehicles, reducing mileage reductions should be especially accumulated by those vehicles, and beneficial to highly trafficked corridors.

Climate Change Impacts From Preferred NHTSA’s Environmental Impact minimal. The following graph compares Alternative Statement performed for this the estimated atmospheric CO2 rulemaking shows that the preferred concentration (789.76 ppm) in 2100 The estimated effects of this proposal alternative would result in 3/1,000ths of under the proposed standards to the in terms of fuel savings and CO2 a degree Celsius increase in global estimated level (789.11 ppm) under the emissions, again perhaps somewhat average temperatures by 2100, relative standards set forth in 2012—or an 8/ counter-intuitively, is relatively small as to the standards finalized in 2012. On a 100ths of a percentage increase: 43 compared to the 2012 final rule. net CO2 basis, the results are similarly

41 The agencies are specifically requesting necessarily choose to avoid mass reductions in the annually for the CAFE program and 1,150 fewer comment on the appropriateness and level of the ways that the agencies assumed. See, 77 FR 623624, fatalities for the CO2 program over calendar years effects of the rebound effect. The agencies also seek 62763 (Oct. 15, 2012). By choosing where and how 2036–2045. comment on changes as compared to the 2012 to limit assumed mass reduction, the 2012 rule’s 43 Counter-intuitiveness is relative, however. The modeling relating to mass reduction assumptions. safety analysis reduced the projected apparent risk During that rulemaking, the analysis limited the to safety associated with aggressive fuel economy estimated effects of the 2012 final rule on climate amount of mass reduction assumed for certain and CO2 targets. That specific assumption has been were similarly small in magnitude, as shown in the vehicles, which impacted the results regarding removed for today’s analysis. Final EIS accompanying that rule and available on potential for adverse safety effects, even while 42 The reduction in annual fatalities varies each NHTSA’s website. acknowledging that manufacturers would not calendar year, averaging 894 fewer fatalities

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00012 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.006 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42997

Net Benefits From Preferred Alternative as compared to if EPA retained the other alternatives analyzed, recognizing previously-set CO2 standards and the statutory considerations for both Maintaining the MY 2020 curves for NHTSA finalized the augural standards. agencies. Comment is sought on MYs 2021–2026 will save American This was identified as the preferred whether this is an appropriate basis for consumers, the auto industry, and the alternative, in part, because it selection. public a considerable amount of money maximizes net benefits compared to the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00013 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.007 42998 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

These estimates, reported as changes Compliance Flexibilities adjustment- and incentive-type relative to impacts under the standards flexibilities, where there is not always issued in 2012, account for impacts on This proposal also seeks comment on consumer interest in the technologies to a variety of changes to NHTSA’s and vehicles produced during model years be incentivized nor is there necessarily EPA’s compliance programs for CAFE 2016–2029, as well as (through changes clear fuel-saving and emissions- and CO2 as well as related programs. reducing benefit to be derived from that in utilization) vehicles produced in Compliance flexibilities can generally earlier model years, throughout those incentivization. The agencies seek be grouped into two categories. The first comment on all of those requests as part vehicles’ useful lives. Reported values category are those compliance are in 2016 dollars, and reflect three- of this proposal. flexibilities that reduce unnecessary Over-compliance credits, which can percent and seven-percent discount compliance costs and provide for a more be built up in part through use of the rates. Under CAFE standards, costs are efficient program. The second category above-described per-vehicle estimated to decrease by $502 billion of compliance flexibilities are those that adjustments and incentives, can be overall at a three-percent discount rate distort the market—such as by saved and either applied retroactively to ($335 billion at a seven-percent incentivizing the implementation of one accounts for previous non-compliance, discount rate); benefits are estimated to type of technology by providing credit or carried forward to mitigate future decrease by $326 billion at a three- for compliance in excess of real-world non-compliance. Such credits can also percent discount rate ($204 billion at a fuel savings. be traded to other automakers for cash seven-percent discount rate). Thus, net Both programs provide for the or for other credits for different fleets. benefits are estimated to increase by generation of credits based upon fleet- But such trading is not pursued openly. $176 billion at a three-percent discount wide over-compliance, provide for Under the CAFE program, the public is rate and $132 billion at a seven-percent adjustments to the test measured value not made aware of inter-automaker discount rate. The estimated impacts of each individual vehicle based upon trades, nor are shareholders. And even under CO2 standards are similar, with the implementation of certain fuel the agencies are not informed of the net benefits estimated to increase by saving technologies, and provide price of credits. With the exception of $201 billion at a three-percent discount additional incentives for the statutorily-mandated credits, the rate and $141 billion at a seven-percent implementation of certain preferred agencies seek comment on all aspects of discount rate. technologies (regardless of actual fuel the current system. The agencies are savings). Auto manufacturers and others particularly interested in comments on have petitioned for a host of additional flexibilities that may distort the market.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00014 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.008 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 42999

The agencies seek comment as to standards are, in the aggregate, at least related to fuel economy. Tailpipe CO2 whether some adjustments and non- as protective of public health and standards, whether in the form of fleet- statutory incentives and other welfare as applicable Federal standards, wide CO2 limits or in the form of provisions should be eliminated and is arbitrary and capricious; that requirements that manufacturers selling stringency levels adjusted accordingly. California does not need its own vehicles in California sell a certain In general, well-functioning banking standards to meet compelling or number of low- and no-tailpipe-CO2 and trading provisions increase market extraordinary conditions; or that such emissions vehicles as part of their efficiency and reduce the overall costs California standards and accompanying overall sales, are unquestionably related of compliance with regulatory enforcement procedures are not to fuel economy standards. Standards objectives. The agencies request consistent with Section 202(a) of the that control tailpipe CO2 emissions are comment on whether the current system CAA. In this proposal, EPA is proposing de facto fuel economy standards as implemented might need to withdraw the waiver granted to because CO2 is a direct and inevitable improvements to achieve greater California in 2013 for the GHG and ZEV byproduct of the combustion of carbon- efficiencies. We seek comment on requirements of its Advanced Clean based fuels to make energy, and the vast specific programmatic changes that Cars program, in light of all of these majority of the energy that powers could improve compliance with current factors. passenger cars and light trucks comes standards in the most efficient way, Attempting to solve climate change, from carbon-based fuels. ranging from requiring public disclosure even in part, through the Section 209 Improving fuel economy means of some or all aspects of credit trades, waiver provision is fundamentally getting the vehicle to go farther on a to potentially eliminating credit trading different from that section’s original gallon of gas; a vehicle that goes farther in the CAFE program. We request purpose of addressing smog-related air on a gallon of gas produces less CO2 per commenters to provide any data, quality problems. When California was unit of distance; therefore, improving evidence, or existing literature to help merely trying to solve its air quality fuel economy necessarily reduces agency decision-making. issues, there was a relatively- tailpipe CO2 emissions, and reducing straightforward technology solution to CO2 emissions necessarily improves fuel One National Standard the problems, implementation of which economy. EPCA therefore necessarily Setting appropriate and maximum did not affect how consumers lived and preempts California’s Advanced Clean feasible fuel economy and tailpipe CO2 drove. Section 209 allowed California to Cars program to the extent that it emissions standards requires regulatory pursue additional reductions to address regulates or prohibits tailpipe CO2 efficiency. This proposal addresses a its notorious smog problems by emissions. Section VI of this proposal, fundamental and unnecessary requiring more stringent standards, and below, discusses the CAA waiver and complication in the currently-existing allowed California and other States that EPCA preemption in more detail. regulatory framework, which is the failed to comply with Federal air quality Eliminating California’s regulation of regulation of GHG emissions from standards to make progress toward fuel economy pursuant to Congressional passenger cars and light trucks by the compliance. Trying to reduce carbon direction will provide benefits to the State of California through its GHG emissions from motor vehicles in any American public. The automotive standards and Zero Emission Vehicle significant way involves changes to the industry will, appropriately, deal with (ZEV) mandate and subsequent entire vehicle, not simply the addition fuel economy standards on a national adoption of these standards by other of a single or a handful of control basis—eliminating duplicative States. Both EPCA and the CAA technologies. The greater the emissions regulatory requirements. Further, preempt State regulation of motor reductions are sought, the greater the elimination of California’s ZEV program vehicle emissions (in EPCA’s case, likelihood that the characteristics and will allow automakers to develop such standards that are related to fuel capabilities of the vehicle currently vehicles in response to consumer economy standards). The CAA gives sought by most American consumers demand instead of regulatory mandate. EPA the authority to waive preemption will have to change significantly. Yet, This regulatory mandate has required for California under certain even decades later, California continues automakers to spend tens of billions of circumstances. EPCA does not provide to be in widespread non-attainment dollars to develop products that a for a waiver of preemption under any with Federal air quality standards.46 In significant majority of consumers have circumstances. In short, the agencies the past decade, California has not adopted, and consequently to sell propose to maintain one national disproportionately focused on GHG such products at a loss. All of this is standard—a standard that is set emissions. Parts of California have a real paid for through cross subsidization by exclusively by the Federal government. and significant local air pollution increasing prices of other vehicles not problem, but CO2 is not part of that local just in California and other States that Proposed Withdrawal of California’s problem. have adopted California’s ZEV mandate, Clean Air Act Preemption Waiver but throughout the country. California’s Tailpipe CO Emissions EPA granted a waiver of preemption 2 Standards and ZEV Mandate Conflict Request for Comment to California in 2013 for its ‘‘Advanced With EPCA Clean Car’’ regulations, composed of its The agencies look forward to all GHG standards, its ‘‘Low Emission Moreover, California regulation of comments on this proposal, and wish to Vehicle (LEV)’’ program and the ZEV tailpipe CO2 emissions, both through its emphasize that obtaining public input is program,44 and, as allowed under the GHG standards and ZEV program, extremely important to us in selecting CAA, a number of other States adopted conflicts directly and indirectly with from among the alternatives in a final California’s standards.45 The CAA states EPCA and the CAFE program. EPCA rule. While the agencies and the that EPA shall not grant a waiver of expressly preempts State standards Administration met with a variety of preemption if EPA finds that stakeholders prior to issuance of this California’s determination that its 46 See California Nonattainment/Maintenance proposal, those meetings have not Status for Each County by Year for All Criteria resulted in a predetermined final rule Pollutants, current as of May 31, 2018, at https:// 44 78 FR 2112 (Jan. 9, 2013). www3.epa.gov/airquality/greenbook/anayo_ca.html outcome. The Administrative Procedure 45 CAA Section 177, 42 U.S.C. 7507. (last accessed June 15, 2018). Act requires that agencies provide the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00015 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43000 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

public with adequate notice of a and GHG emissions also used the CAFE EPA developed two models after proposed rule followed by a meaningful model for analysis.47 2009, referred to as the ‘‘ALPHA’’ and opportunity to comment on the rule’s DOT recently arranged for a formal ‘‘OMEGA’’ models, which provide some content. The agencies are committed to peer review of the model. In general, of the same capabilities as the following that directive. reviewers’ comments strongly supported Autonomie and CAFE models. EPA the model’s conceptual basis and applied the OMEGA model to conduct II. Technical Foundation for NPRM implementation, and commenters analysis of GHG standards promulgated Analysis provided several specific in 2010 and 2012, and the ALPHA and A. Basics of CAFE and CO2 Standards recommendations. DOT staff agreed OMEGA models to conduct analysis Analysis with many of these recommendations discussed in the above-mentioned 2016 and have worked to implement them Draft TAR and Proposed and Final The agencies’ analysis of CAFE and wherever practicable. Implementing Determinations regarding standards CO2 standards involves two basic some of them would require beyond 2021. In an August 2017 notice, elements: first, estimating ways each considerable further research, the agencies requested comments on, manufacturer could potentially respond development, and testing, and will be among other things, whether EPA to a given set of standards in a manner considered going forward. For a handful should use alternative methodologies that considers potential consumer of other recommendations, DOT staff and modeling, including DOE/ response; and second, estimating disagreed, often finding the Argonne’s Autonomie full-vehicle various impacts of those responses. recommendations involved simulation tool and DOT’s CAFE Estimating manufacturers’ potential considerations (e.g., other policies, such model.53 responses involves simulating as those involving fuel taxation) beyond Having reviewed comments on the manufacturers’ decision-making the model itself or were based on subject and having considered the processes regarding the year-by-year concerns with inputs rather than how matter fully, the agencies have application of fuel-saving technologies the model itself functioned. A report determined it is reasonable and to specific vehicles. Estimating impacts available in the docket for this appropriate to use DOE/Argonne’s involves calculating resultant changes rulemaking presents peer reviewers’ model for full-vehicle simulation, and to in new vehicle costs, estimating a detailed comments and use DOT’s CAFE model for analysis of variety of costs (e.g., for fuel) and effects recommendations, and provides DOT’s regulatory alternatives. EPA interprets (e.g., CO2 emissions from fuel detailed responses.48 Section 202(a) of the CAA as giving the combustion) occurring as vehicles are The agencies also use four DOE and agency broad discretion in how it driven over their lifetimes before DOE-sponsored models to develop develops and sets GHG standards for eventually being scrapped, and inputs to the CAFE model, including light-duty vehicles. Nothing in Section estimating the monetary value of these three developed and maintained by 202(a) mandates that EPA use any effects. Estimating impacts also involves DOE’s Argonne National Laboratory. specific model or set of models for The agencies use the DOE Energy consideration of the response of analysis of potential CO2 standards for consumers—e.g., whether consumers Information Administration’s (EIA’s) light-duty vehicles. EPA weighs many will purchase the vehicles and in what National Energy Modeling System factors when determining appropriate (NEMS) to estimate fuel prices,49 and quantities. Both of these basic analytical levels for CO2 standards, including the elements involve the application of used Argonne’s Greenhouse gases, cost of compliance (see Section many analytical inputs. Regulated Emissions, and Energy use in 202(a)(2)), lead time necessary for The agencies’ analysis uses the CAFE Transportation (GREET) model to compliance (also Section 202(a)(2)), model to estimate manufacturers’ estimate emissions rates from fuel 50 safety (see NRDC v. EPA, 655 F.2d 318, potential responses to new CAFE and production and distribution processes. 336 n. 31 (D.C. Cir. 1981) and other DOT also sponsored DOE/Argonne to CO2 standards and to estimate various impacts on consumers,54 and energy impacts of those responses. The model use their Autonomie full-vehicle impacts associated with use of the makes use of many inputs, values of simulation system to estimate the fuel technology.55 Using the CAFE model which are developed outside of the economy impacts for roughly a million combinations of technologies and model to estimate the battery cost associated with model and not by the model. For 51 52 example, the model applies fuel prices; vehicle types. each technology combination based on it does not estimate fuel prices. The characteristics of the simulated vehicle and its level 47 While this rulemaking employed the CAFE of electrification. Information regarding Argonne’s model does not determine the form or model for analysis, EPA and DOT used different BatPAC model is available at http:// stringency of the standards; instead, the versions of the CAFE model for establishing their www.cse.anl.gov/batpac/. model applies inputs specifying the respective standards, and EPA also used the EPA 52 Additionally, the impact of engine technologies form and stringency of standards to be MOVES model. See 81 FR 73478, 73743 (Oct. 25, on fuel consumption, torque, and other metrics was 2016). characterized using GT POWER simulation analyzed and produces outputs showing 48 Docket No. NHTSA–2018–0067. modeling in combination with other engine effects of manufacturers working to 49 See https://www.eia.gov/outlooks/aeo/info_ modeling that was conducted by IAV Automotive meet those standards, which become the nems_archive.php. Today’s notice uses fuel prices Engineering, Inc. (IAV). The engine characterization basis for comparing between different estimated using the Annual Energy Outlook (AEO) ‘‘maps’’ resulting from this analysis were used as inputs for the Autonomie full-vehicle simulation potential stringencies. 2017 version of NEMS (see https://www.eia.gov/ outlooks/archive/aeo17/ and https://www.eia.gov/ modeling. Information regarding GT Power is DOT’s Volpe National Transportation outlooks/aeo/data/browser/#/?id=3-AEO2017 available at https://www.gtisoft.com/gt-suite- Systems Center (often simply referred to &cases=ref2017&sourcekey=0). applications/propulsion-systems/gt-power-engine- 50 simulation-software. as the ‘‘Volpe Center’’) develops, Information regarding GREET is available at https://greet.es.anl.gov/index.php. Availability of 53 82 FR 39533 (Aug. 21, 2017). maintains, and applies the model for NEMS is discussed at https://www.eia.gov/ 54 Since its earliest Title II regulations, EPA has NHTSA. NHTSA has used the CAFE outlooks/aeo/info_nems_archive.php. Today’s considered the safety of pollution control model to perform analyses supporting notice uses fuel prices estimated using the AEO technologies. See 45 FR 14496, 14503 (1980). every CAFE rulemaking since 2001, and 2017 version of NEMS. 55 See George E. Warren Corp. v. EPA, 159 F.3d 51 As part of the Argonne simulation effort, 616, 623–624 (D.C. Cir. 1998) (ordinarily the 2016 rulemaking regarding heavy- individual technology combinations simulated in permissible for EPA to consider factors not duty pickup and van fuel consumption Autonomie were paired with Argonne’s BatPAC specifically enumerated in the Act).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00016 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43001

allows consideration of the following future, and manufacturers’ year-by-year business, frequency of engine on/off factors: the CAFE model explicitly plans involve some vehicles ‘‘carrying transitions). This steadily increasing evaluates the cost of compliance for forward’’ technology from prior model realism has, in turn, steadily increased each manufacturer, each fleet, and each years and some other vehicles possibly confidence in the appropriateness of model year; it accounts for lead time applying ‘‘extra’’ technology in using Autonomie to make significant necessary for compliance by directly anticipation of standards in ensuing investment decisions. Notably, DOE incorporating estimated manufacturer model years, and manufacturers’ uses Autonomie for analysis supporting production cycles for every vehicle in planning also involves applying credits budget priorities and plans for programs the fleet, ensuring that the analysis does carried forward between model years. managed by its Vehicle Technologies not assume vehicles can be redesigned Furthermore, manufacturers cannot Office (VTO). Considering the to incorporate more technology without optimize the powertrain for fuel advantages of DOE/Argonne’s regard to lead time considerations; it economy on every vehicle model Autonomie model, it is reasonable and provides information on safety effects configuration—for example, a given appropriate to use Autonomie to associated with different levels of engine shared among multiple vehicle estimate fuel economy levels and CO2 standards and information about many models cannot practicably be split into emission rates for different other impacts on consumers, and it different versions for each configuration combinations of technologies as applied calculates energy impacts (i.e., fuel of each model, each with a slightly to different types of vehicles. saved or consumed) as a primary different displacement. The CAFE Commenters have also suggested that function, besides being capable of model is designed to account for these the CAFE model’s graphical user providing information about many other real-world factors. interface (GUI) facilitates others’ ability factors within EPA’s broad CAA Considering the technological to use the model quickly—and without discretion to consider. heterogeneity of manufacturers’ current specialized knowledge or training—and Because the CAFE model simulates a product offerings, and the wide range of to comment accordingly.56 For today’s wide range of actual constraints and ways in which the many fuel economy- proposal, DOT has significantly practices related to automotive improving/CO2 emissions-reducing expanded and refined this GUI, engineering, planning, and production, technologies included in the analysis providing the ability to observe the such as common vehicle platforms, can be combined, the CAFE model has model’s real-time progress much more sharing of engines among different been designed to use inputs that provide closely as it simulates year-by-year vehicle models, and timing of major an estimate of the fuel economy compliance with either CAFE or CO2 vehicle redesigns, the analysis produced achieved for many tens of thousands of standards.57 Although the model’s by the CAFE model provides a different potential combinations of fuel- ability to produce realistic results is transparent and realistic basis to show saving technologies. Across the range of independent of the model’s GUI, it is pathways manufacturers could follow technology classes encompassed by the anticipated the CAFE model’s GUI will over time in applying new technologies, analysis fleet, today’s analysis involves facilitate stakeholders’ meaningful which helps better assess impacts of more than a million such estimates. review and comment during the potential future standards. Furthermore, While the CAFE model requires no comment period. because the CAFE model also accounts specific approach to developing these Beyond these general considerations, fully for regulatory compliance inputs, the National Academy of several specific related technical provisions (now including CO2 Sciences (NAS) has recommended, and comments and considerations underlie compliance provisions), such as stakeholders have commented, that full- the agencies’ decision in this area, as adjustments for reduced refrigerant vehicle simulation provides the best discussed, where applicable, in the leakage, production ‘‘multipliers’’ for balance between realism and remainder of this Section. some specific types of vehicles (e.g., practicality. DOE/Argonne has spent Other commenters expressed a PHEVs), and carried-forward (i.e., several years developing, applying, and number of concerns with whether banked) credits, the CAFE model expanding means to use distributed DOT’s CAFE model could be used for provides a transparent and realistic computing to exercise its Autonomie CAA analysis. Many of these concerns basis to estimate how such technologies full-vehicle simulation tool over the focused on inputs used by the CAFE might be applied over time in response scale necessary for realistic analysis of model for prior rulemaking to CAFE or CO2 standards. CAFE or average CO2 standards. This analyses.58 59 60 Because inputs are There are sound reasons for the scalability and related flexibility (in agencies to use the CAFE model going terms of expanding the set of 56 From Docket Number EPA–HQ–OAR–2015– forward in this rulemaking. First, the technologies to be simulated) makes 0827, see Comment by Global Automakers, Docket ID EPA–HQ–OAR–2015–0827–9728, at 34. CAFE and CO2 fact analyses are Autonomie well-suited for developing 57 The updated GUI provides a range of graphs inextricably linked. Furthermore, the inputs to the CAFE model. updated in real time as the model operates. These analysis produced by the CAFE model Additionally, DOE/Argonne’s graphs can be used to monitor fuel economy or CO2 and DOE/Argonne’s Autonomie Autonomie also has a long history of ratings of vehicles in manufacturers’ fleets and to addresses several analytical needs. The development and widespread monitor year-by-year CAFE (or average CO2 ratings), costs, avoided fuel outlays, and avoided CO2-related CAFE model provides an explicit year- application by a much wider range of damages for specific manufacturers and/or specific by-year simulation of manufacturers’ users in government, academia, and fleets (e.g., domestic passenger car, light truck). application of technology to their industry. Many of these users apply Because these graphs update as the model products in response to a year-by-year Autonomie to inform funding and progresses, they should greatly increase users’ progression of CAFE standards and design decisions. These real-world understanding of the model’s approach to considerations such as multiyear planning, accounts for sharing of technologies and exercises have contributed significantly payment of civil penalties, and credit use. the implications for timing, scope, and to aspects of Autonomie important to 58 For example, EDF’s recent comments (EDF at limits on the potential to optimize producing realistic estimates of fuel 12, Docket ID. EPA–HQ–OAR–2015–0827–9203) powertrains for fuel economy. In the economy levels and CO2 emission rates, stated ‘‘the data that NHTSA needs to input into its model is sensitive confidential business real world, standards actually are such as estimation and consideration of information that is not transparent and cannot be specified on a year-by-year basis, not performance, utility, and driveability independently verified . . .’’ and claimed ‘‘the simply some single year well into the metrics (e.g., towing capability, shift Continued

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00017 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43002 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

exogenous to any model, they do not decision may be informed by non-EPA- that includes the specific engines and determine whether it would be created models does not, in any way, transmissions on each model variant, reasonable and appropriate for EPA to constitute a delegation of its statutory observed sales volumes, and all fuel use DOT’s model for analysis. Other power to set standards or decision- economy improvement technology that concerns focused on characteristics of making authority.61 Arguing to the is already present on those vehicles. the CAFE model that were developed to contrary would suggest, for example, From there it adds technology, in better align the model with EPCA and that EPA’s decision would be invalid response to the standards being EISA; the model has been revised to because it relied on EIA’s Annual considered, in a way that minimizes the accommodate both EPCA/EISA and Energy Outlook for fuel prices rather cost of compliance and reflects many CAA analysis, as explained further than developing its own model for real-world constraints faced by below. Some commenters also argued estimating future trends in fuel prices. automobile manufacturers. After that use of any models other than Yet, all Federal agencies that have simulating compliance, the model ALPHA and OMEGA for CAA analysis occasion to use forecasts of future fuel calculates impacts of the simulated would constitute an arbitrary and prices regularly (and appropriately) standard: technology costs, fuel savings capricious delegation of EPA’s decision- defer to EIA’s expertise in this area and (both in gallons and dollars), CO2 making authority to DOT, if DOT rely on EIA’s NEMS-based analysis in reductions, social costs and benefits, models are used for analysis instead. the AEO, even when those same and safety impacts. These comments were made prior to the agencies are using EIA’s forecasts to Today’s analysis reflects several development of the CAA analysis inform their own decision-making. changes made to the CAFE model since function in the CAFE model, and, Moreover, DOT’s CAFE model with 2012, when NHTSA used the model to moreover, appear to conflate the inputs from DOE/Argonne’s Autonomie estimate the effects, costs, and benefits analytical tool used to inform decision- model has produced analysis supporting of final CAFE standards for light-duty making with the action of making the rulemaking under the CAA. In 2015, vehicles produced during MYs 2017– decision. As explained elsewhere in this EPA proposed new GHG standards for 2021 and augural standards for MYs document and as made repeatedly clear MY 2021–2027 heavy-duty pickups and 2022–2025. Key changes relevant to this over the past several rulemakings, the vans, finalizing standards in 2016. analysis include the following: CAFE model neither sets standards nor Supporting the NPRM and final rule, • Expansion of model inputs, dictates where and how to set standards; EPA relied on analysis implemented by procedures, and outputs to it simply informs as to the effects of DOT using DOT’s CAFE model, and accommodate technologies not included setting different levels of standards. In DOT used inputs developed by DOE/ in prior analyses, this rulemaking, EPA will be making its Argonne using DOE/Argonne’s • Updated approach to estimating the own decisions regarding what CO2 Autonomie model. combined effect of fuel-saving standards would be appropriate under The following sections provide a brief technologies using large scale the CAA. The CAA does not require technical overview of the CAFE model, simulation modeling, EPA to create a specific model or use a including changes NHTSA made to the • Modules that dynamically estimate specific model of its own creation in model since 2012, before discussing new vehicle sales and existing vehicle setting GHG standards. The fact EPA’s inputs to the model and then diving scrappage in response to changes to new more deeply into how the model works. vehicle prices that result from OMEGA model’s focus on direct technological For more information on the latter topic, manufacturers’ compliance actions, inputs and costs—as opposed to industry self- see the CAFE model documentation July • A safety module that estimates the reported data—ensures the model more accurately changes in light-duty traffic fatalities characterizes the true feasibility and cost 2018 draft, available in the docket for effectiveness of deploying greenhouse gas reducing this rulemaking and on NHTSA’s resulting from changes to vehicle technologies.’’ Neither statement is correct, as website. exposure, vehicle retirement rates, and nothing about either the CAFE or OMEGA model reductions in vehicle mass to improve either obviates or necessitates the use of CBI to 1. Brief Technical Overview of the fuel economy, develop inputs. Model • 59 In recent comments (CARB at 28, Docket ID. Disaggregation of each EPA–HQ–OAR–2015–0827–9197), CARB stated The CAFE model is designed to manufacturer’s fleet into separate ‘‘another promising technology entering the market simulate compliance with a given set of ‘‘domestic’’ passenger car and ‘‘import’’ was not even included in the NHTSA compliance CAFE or CO2 standards for each passenger car fleets to better represent modeling’’ and that EPA assumes a five-year redesign cycle, whereas NHTSA assumes a six to manufacturer selling vehicles in the the statutory requirements of the CAFE seven-year cycle.’’ Though presented as criticisms United States. The model begins with a program, of the models, these comments—at least with representation of the current (for today’s • Changes to the algorithm used to respect to the CAFE model—actually concern analysis, model year 2016) vehicle apply technologies, enabling more model inputs. NHTSA did not agree with CARB about the commercialization potential of the engine model offerings for each manufacturer explicit accounting of shared vehicle technology in question (‘‘Atkinson 2’’) and applied components (engines, transmissions, model inputs accordingly. Also, rather than 61 ‘‘[A] federal agency may turn to an outside platforms) and ‘‘inheritance’’ of major applying a one-size-fits-all assumption regarding entity for advice and policy recommendations, technology within or across powertrains redesign cadence, NHTSA developed estimates provided the agency makes the final decisions and/or platforms over time, specific to each vehicle model and applied these as itself.’’ U.S. Telecom. Ass’n v. FCC, 359 F.3d 554, • model inputs. 565–66 (D.C. Cir. 2004). To the extent commenters An industry labor quantity module, 60 NRDC’s recent comments (NRDC at 37, Docket meant to suggest outside parties have a reliance which estimates net changes in the ID. EPA–HQ–OAR–2015–0827–9826) state EPA interest in EPA using ALPHA and OMEGA to set amount of U.S. automobile labor for should not use the CAFE model because it ‘‘allows standards, EPA does not agree a reliance interest is dealerships, Tier 1 and 2 supplier manufacturers to pay civil penalties in lieu of properly placed on an analytical methodology (as meeting the standards, an alternative compliance opposed to on the standards themselves). Even if it companies, and original equipment pathway currently allowed under EISA and EPCA.’’ were, all parties that closely examined ALPHA and manufacturers (OEMs), While the CAFE model can simulate civil penalty OMEGA-based analyses in the past either also • Cost estimation of batteries for payment, NRDC’s comment appears to overlook the simultaneously closely examined CAFE and electrification technologies incorporates fact that this result depends on model inputs; the Autonomie-based analyses in the past, or were fully inputs can easily be specified such that the CAFE capable of doing so, and thus, should face no an updated version of Argonne National model will set aside civil penalty payment as an additional difficulty now they have only one set of Laboratory’s BatPAC (battery) model for alternative to compliance. models and inputs/outputs to examine. hybrid electric vehicles (HEVs), plug-in

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00018 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43003

hybrid electric vehicles (PHEVs), and alternatives the agencies are CO2 standards is the analysis fleet, battery electric vehicles (BEVs), considering. All but one involve which is a snapshot of the recent consistent with how we estimate different standards, and three involve a vehicle market. The analysis fleet effectiveness for those values, gradual discontinuation of CAFE and provides a snapshot to project what • Expanded accounting for CAFE GHG adjustments reflecting the vehicles will exist in future model years credits carried over from years prior to application of technologies that improve covered by the standards and what those included in the analysis (a.k.a. air conditioner efficiency or, in other technologies they will have, and then ‘‘banked’’ credits) and application to ways, improve fuel economy under evaluate what additional technologies future CAFE deficits to better evaluate conditions not represented by long- can feasibly be applied to those vehicles anticipated manufacturer responses to standing fuel economy test procedures. in a cost-effective way to raise their fuel 62 proposed standards, Like the baseline no action alternative, economy and lower their CO2 emission • The ability to represent a all of these alternatives are more levels.63 manufacturer’s preference for fine stringent than the preferred alternative. Part of reflecting what vehicles will payment (rather than achieving full Section III and Section IV describe the exist in future model years is knowing compliance exclusively through fuel preferred and other regulatory which vehicles are produced by which economy improvements) on a year-by- alternatives, respectively. manufacturers, how many of each are year basis, These alternatives were examined sold, and whether they are passenger • Year-by-year simulation of how because they will be considered as cars or light trucks. This is important manufacturers could comply with EPA’s options for the final rule. The agencies because it improves our understanding seek comment on these alternatives, CO2 standards, including of the overall impacts of different levels Æ seek any relevant data and information, Calculation of vehicle models’ CO2 of CAFE and CO2 standards; overall emission rates before and after and will review responses. That review impacts result from industry’s response application of fuel-saving (and, could lead to the selection of one of the to standards, and industry’s response is other regulatory alternatives for the final therefore, CO2-reducing) technologies, made up of individual manufacturer rule or some combination of the other Æ Calculation of manufacturers’ fleet responses to the standards in light of the regulatory alternatives (e.g., combining average CO2 emission rates, overall market and their individual passenger cars standards from one Æ Calculation of manufacturers’ fleet assessment of consumer acceptance. alternative with light truck standards average CO2 emission rates under Having an accurate picture of attribute-based CO standards, from a different alternative). 2 Because outputs depend on inputs manufacturers’ existing fleets (and the Æ Accounting for adjustments to vehicle models in them) that will be average CO emission rates reflecting (e.g., the results of the analysis in terms 2 of quantities and kinds of technologies subject to future standards helps us reduction of air conditioner refrigerant better understand individual leakage, required to meet different levels of standards, and the societal and private manufacturer responses to those future Æ Accounting for the treatment of standards in addition to potential alternative fuel vehicles for CO benefits associated with manufacturers 2 meeting different levels of standards changes in those standards. compliance, Another part of reflecting what Æ Accounting for production depend on input data, estimates, and assumptions), the analysis also explores vehicles will exist in future model years ‘‘multipliers’’ for PHEVs, BEVs, is knowing what technologies are compressed natural gas (CNG) vehicles, the sensitivity of results to many of these inputs. For example, the net already on those vehicles. Accounting and fuel cell vehicles (FCVs), for technologies already being on Æ Accounting for transfer of CO benefits of any regulatory alternative 2 will depend strongly on fuel prices well vehicles helps avoid ‘‘double-counting’’ credits between regulated fleets, the value of those technologies, by Æ Accounting for carried-forward beyond 2025. Fuel prices a decade and more from now are not knowable with assuming they are still available to be (a.k.a. ‘‘banked’’) CO2 credits, including applied to improve fuel economy and credits from model years earlier than certainty. The sensitivity analysis reduce CO2 emissions. It also promotes modeled explicitly. involves repeating the ‘‘central’’ or ‘‘reference case’’ analysis under more realistic determinations of what 2. Sensitivity Cases and Why We alternative inputs (e.g., higher fuel additional technologies can feasibly be Examine Them prices in one case, lower fuel prices in applied to those vehicles: if a manufacturer has already started down Today’s notice presents estimated another case), and exploring changes in analytical results, which is discussed a technological path to fuel economy or impacts of the proposed CAFE and CO2 further in the agencies’ Preliminary performance improvements, we do not standards defining the proposals, assume it will completely abandon that relative to a baseline ‘‘no action’’ Regulatory Impact Analysis (PRIA) accompanying today’s notice. path because that would be unrealistic regulatory alternative under which the and would not accurately represent standards announced in 2012 remain in B. Developing the Analysis Fleet for manufacturer responses to standards. place through MY 2025 and continue Assessing Costs, Benefits, and Effects of Each vehicle model (and configurations unchanged thereafter. Relative to this Alternative CAFE Standards of each model) in the analysis fleet, same baseline, today’s notice also The following sections describe what therefore, has a comprehensive list of its presents analysis estimating impacts the analysis fleet is and why it is used, technologies, which is important under a range of other regulatory how it was developed for this NPRM, because different configurations may have different technologies applied to 62 While EPCA/EISA precludes NHTSA from and the analysis-fleet-related topics on considering manufacturers’ potential use of credits which comment is sought. in model years for which the agency is establishing 63 The CAFE model does not generate compliance new standards, NHTSA considers credit use in 1. Purpose of Developing and Using an paths a manufacturer should, must, or will deploy. earlier model years. Also, as allowed by NEPA, Analysis Fleet It is intended as a tool to demonstrate a compliance NHTSA’s EISs present results of analysis that pathway a manufacturer could choose. It is almost considers manufacturers’ potential use of credits in The starting point for the evaluation certain all manufacturers will make compliance all model years, including those for which the of the potential feasibility of different choices differing from those projected in the CAFE agency is establishing new standards. stringency levels for future CAFE and model.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00019 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43004 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

them.64 Additionally, the analysis development, and manufacturing and trucks, and between manufacturers) accounts for platforms within equipment budgets and resources.66 endogenously as part of the analysis, manufacturers’ fleets, recognizing Manufacturers have repeatedly told the rather than using external forecasts of platforms will share technologies, and agencies that sustainable business plans future car/truck split and future the vehicles that make up that platform require careful management of resources manufacturer sales volumes. This leads should receive (or not receive) and capital spending, and that the the model to produce different estimates additional technological improvements length of time each product remains in of future production volumes under together. The specific engineering production is crucial to recouping the different standards and in response to characteristics of each model/ upfront product development and plant/ different inputs, reflecting the configuration are available in the equipment costs, as well as the capital expectation that regulatory standards aforementioned input file.65 For the needed to fund the development and and other external factors will, in fact, regulatory alternatives considered in manufacturing equipment needed for impact the market. today’s proposal, estimates of rates at future products. Because the production The input file for the CAFE model which various technologies might be volume of any given vehicle model characterizing the analysis fleet 67 expected to penetrate manufacturers’ varies within a manufacturer’s product includes a large amount of data about fleets (and the overall market) are line and also varies among different vehicle models/configurations, their summarized below in Sections VI and manufacturers, redesign schedules technological characteristics, the VII, and in Chapter 6 of the typically vary for each model and manufacturers and fleets to which they accompanying PRIA and in detailed manufacturer. Some (relatively few) belong, and initial prices and model output files available at NHTSA’s technological improvements are small production volumes which provide the website. A solid characterization of a enough they can be applied in any starting points for projection (by the recent model year as an analytical model year; others are major enough sales model) to ensuing model years. starting point helps to realistically they can only be cost-effectively applied The following sections discuss aspects estimate ways manufacturers could at a vehicle redesign, when many other of how the analysis fleet was built for potentially respond to different levels of things about the vehicle are already this proposal and seek comment on standards, and the modeling strives to changing. Ensuring the CAFE model those topics. realistically simulate how makes technological improvements to 2. Source Data for Building the Analysis manufacturers could progress from that vehicles only when it is feasible to do Fleet starting point. Nevertheless, so also helps the analysis better manufacturers can respond in many represent manufacturer responses to The source data for the vehicle ways beyond those represented in the different levels of standards. models/configurations in the analysis analysis (e.g., applying other A final important aspect of reflecting fleet and their technologies is a central technologies, shifting production what vehicles will exist in future model input for the analysis. The sections volumes, changing vehicle footprint), years and potential manufacturer below discuss pros and cons of different such that it is impossible to predict with responses to standards is estimating potential sources and what was used for any certainty exactly how each how future sales might change in this proposal. manufacturer will respond. Therefore, response to different potential (a) Use of Confidential Business recent trends in manufacturer standards. If potential future standards Information Versus Publicly-Releasable performance and technology appear likely to have major effects in Sources application, although of interest in terms of shifting production from cars to terms of understanding manufacturers’ trucks (or vice versa), or in terms of Since 2001, CAFE analysis has used current compliance positions, are not in shifting sales between manufacturers or either confidential, forward-estimating themselves innately indicative of future groups of manufacturers, that is product plans from manufacturers, or potential. important for the agencies to consider. publicly available data on vehicles Yet, another part of reflecting what For previous analyses, the CAFE model already sold, as a starting point for vehicles will exist in future model years used a static forecast contained in the determining what technologies can be is having reasonable real-world analysis fleet input file, which specified applied to what vehicles in response to assumptions about when certain changes in production volumes over potential different levels of standards. technologies can be applied to vehicles. time for each vehicle model/ These two sources present a tradeoff. Each vehicle model/configuration in the configuration. This approach yielded Confidential product plans analysis fleet also has information about results that, in terms of production comprehensively represent what its redesign schedule, i.e., the last year volumes, did not change between vehicles a manufacturer expects to it was redesigned and when the scenarios or with changes in important produce in coming years, accounting for agencies expect it to be redesigned model inputs. For example, very plans to introduce new vehicles and again. Redesign schedules are a key part stringent standards with very high fuel-saving technologies and, for of manufacturers’ business plans, as technology costs would result in the example, plans to discontinue other each new product can cost more than same estimated production volumes as vehicles and even brands. This $1.0B and involve a significant portion less stringent standards with very low information can be very thorough and of a manufacturer’s scarce research, technology costs. can improve the accuracy of the New for today’s proposal, the CAFE analysis, but for competitive reasons, 64 Considering each vehicle model/configuration model begins with the first-year most of this information must be also improves the ability to consider the differential production volumes (i.e., MY 2016 for redacted prior to publication with impacts of different levels of potential standards on today’s analysis) and adjusts ensuing rulemaking documentation. This makes different manufacturers, since all vehicle model/ it difficult for public commenters to configurations ‘‘start’’ at different places, in terms sales mix year by year (between cars and of the technologies they already have and how reproduce the analysis for themselves as those technologies are used. 66 Shea, T. Why Does It Cost So Much For 65 Available with the model and other input files Automakers To Develop New Models?, Autoblog 67 Available with the model and other input files supporting today’s announcement at https:// (Jul. 27, 2010), https://www.autoblog.com/2010/07/ supporting today’s announcement at https:// www.nhtsa.gov/corporate-average-fuel-economy/ 27/why-does-it-cost-so-much-for-automakers-to- www.nhtsa.gov/corporate-average-fuel-economy/ compliance-and-effects-modeling-system. develop-new-models/. compliance-and-effects-modeling-system.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00020 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43005

they develop their comments. Some information from manufacturers.68 the potential for change is likely not non-industry commenters have also Information from MY 2016 was chosen significant enough to merit using final expressed concern manufacturers would as the foundation for today’s analysis data from an earlier model year have an incentive in the submitted fleet because, at the time the rulemaking reflecting a more outdated fleet. plans to (deliberately or not) analysis was initiated, the 2016 fleet Moreover, even ostensibly final CAFE underestimate their future fuel economy represented the most up-to-date compliance data can sometimes be capabilities and overstate their information available in terms of subject to later revision (e.g., if errors in expectations about, for example, the individual vehicle models and fuel economy tests are discovered), and levels of performance of future vehicle configurations, production technology the purpose of today’s analysis is not to levels, and production volumes. If MY support enforcement actions but rather models in order to affect the analysis. 2017 data had been used while this to provide a realistic assessment of Since 2010, EPA and NHTSA have analysis was being developed, the manufacturers’ potential responses to based analysis fleets almost exclusively agencies would have needed to use future standards. on information from commercial and product planning information that could Manufacturers integrated a significant public sources, starting with CAFE not be made available to the public until amount of new technology in the MY compliance data and adding a later date. 2016 fleet, and this was especially true information from other sources. The analysis fleet was initially for newly-designed vehicles launched in An analysis fleet based primarily on developed with 2016 mid-model year MY 2016. While subsequent fleets will public sources can be released to the compliance data because final involve even further application of public, solving the issue of commenters compliance data was not available at technology, using available data for MY being unable to reproduce the overall that time, and the timing provided 2016 provides the most realistic detailed analysis when they want to. However, manufacturers the opportunity to review foundation for analysis that can be made industry commenters have argued such and comment on the characterization of available publicly in full detail, allowing stakeholders to independently an analysis fleet cannot accurately their vehicles in the fleet. With a view toward developing an accurate reproduce the analysis presented in this reflect manufacturers actual plans to characterization of the 2016 fleet to proposal. Insofar as future product apply fuel-saving technologies (e.g., serve as an analytical starting point, offerings are likely to be more similar to manufacturers may apply turbocharging corrections and updates to mid-year vehicles produced in 2016 than to to improve not just fuel economy, but data (e.g., to production estimates) were vehicles produced in earlier model also to improve vehicle performance) or sought, in addition to corroboration or years, using available data regarding the manufacturers’ plans to change product correction of technical information 2016 model year provides the most offerings by introducing some vehicles obtained from commercial and other realistic, publicly releasable foundation and brands and discontinuing other sources (to the extent that information for constructing a forecast of the future vehicles and brands, precisely because was not included in compliance data), vehicle market for this proposal. that information is typically although future product planning A number of comments to the Draft confidential business information (CBI). information from manufacturers (e.g., TAR, EPA’s Proposed Determination, A fully-publicly-releasable analysis fleet future product offerings, products to be and EPA’s 2017 Request for Comment 69 holds vehicle characteristics unchanged discontinued) was not requested, as stated that the most up-to-date analysis over time and arguably lacks some level most manufacturers view such fleet possible should be used, because a of accuracy when projected into the information as CBI. Manufacturers more up-to-date analysis fleet will better future. For example, over time, offered a range of corrections to indicate capture how manufacturers apply manufacturers introduce new products engineering characteristics (e.g., technology and will account better for vehicle model/configuration and even entire brands. On the other footprint, curb weight, transmission introductions and deletions.70 On the hand, plans announced in press releases type) of specific vehicle model/ other hand, some commenters suggested do not always ultimately bear out, nor configurations, as well as updates to fuel economy and production volume that because manufacturers continue do commercially-available third-party improving vehicle performance and forecasts. Assumptions could be made estimates in mid-year reporting. After following up on a case-by-case basis to utility over time, an older analysis fleet about these issues to improve the should be used to estimate how the fleet accuracy of a publicly-releasable investigate significant differences, the analysis fleet was updated. could have evolved had manufacturers analysis fleet, but concerns include that Sales, footprint, and fuel economy applied all technological potential to this information would either be largely values with final compliance data were incorrect, or information would be also updated if that data was available. 69 82 FR 39551 (Aug. 21, 2017). 70 released that manufacturers would In a few cases, final production and fuel For example, in 2016 comments to dockets consider CBI. We seek comment on how EPA–HQ–OAR–2015–0827 and NHTSA–2016– economy values may be slightly 0068, the Alliance of Automobile Manufacturers to address this issue going forward, different for specific model year 2016 commented that ‘‘the Alliance supports the use of recognizing the competing interests vehicle models and configurations than the most recent data available in establishing the involved and also recognizing typical baseline fleet, and therefore believes that NHTSA’s are indicated in today’s analysis; selection [of, at the time, model year 2015] was timeframes for CAFE and CO2 standards however, other vehicle characteristics more appropriate for the Draft TAR.’’ (Alliance at rulemakings. (e.g., footprint, curb weight, technology 82, Docket ID. EPA–HQ–OAR–2015–0827–4089) content) important to the analysis Global Automakers commented that ‘‘a one-year (b) Use of MY 2016 CAFE Compliance difference constitutes a technology change-over for should be accurate. While some Data Versus Other Starting Points up to 20% of a manufacturer’s fleet. It was also commenters have, in the past, raised generally understood by industry and the agencies concerns that non-final CAFE that several new, and potentially significant, Based on the assumption that a technologies would be implemented in MY 2015. publicly-available analysis fleet compliance data is subject to change, The use of an older, outdated baseline can have continues to be desirable, for this significant impacts on the modeling of subsequent 68 NPRM, an analysis fleet was constructed CO2 emissions rates are directly related to fuel Reference Case and Control Case technologies.’’ economy levels, and the CAFE model uses the latter (Global Automakers at A–10, Docket ID. EPA–HQ– starting with CAFE compliance to calculate the former. OAR–2015–0827–4009).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00021 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43006 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

73 fuel economy rather than continuing to confidence into an analysis fleet for economy and reduce CO2 emissions. improve vehicle performance and modeling supporting deliberations With a portfolio of descriptive utility.71 Because manufacturers change regarding today’s proposal. technologies arranged by manufacturer and improve product offerings over In short, the 2016 fleet was, in fact, and model, the analysis fleet can be time, conducting analysis with an older the most up-to-date fleet that could be summarized and project how vehicles in analysis fleet (or with a fleet using fuel produced for this NPRM. Moreover, that fleet may improve over time via the economy levels and CO emissions rates 2 during late 2016 and early 2017, nearly application of additional technology. that have been adjusted to reflect an all manufacturers provided comments In many cases, vehicle technology is assumed return to levels of performance clearly observable from the 2016 on the characterization of their vehicles and utility typical of some past model compliance data (e.g., compliance data in the analysis fleet, and many provided year) would miss this real-world trend. indicates clearly which vehicles have specific feedback about their vehicles, While such an analysis could turbochargers and which have including aerodynamic drag demonstrate what industry could do if, continuously variable transmissions), coefficients, tire rolling resistance for example, manufacturers devoted all but in some cases technology levels are technological improvements toward values, transmission efficiencies, and less observable. For the latter, like levels other information used in the analysis. raising fuel economy and reducing CO2 of mass reduction, the analysis emissions (and if consumers decided to NHTSA worked with manufacturers to categorized levels of technology already purchase these vehicles), we do not clarify and correct some information used in a given vehicle. Similarly, believe it would be consistent with a and integrated the information into a engineering judgment was used to transparent examination of what effects single input file for use in the CAFE determine if higher mass reduction different levels of standards would have model. Accordingly, the current levels may be used practicably and on individual manufacturers and the analysis fleet is reasonable to use for safely in a given vehicle. fleet as a whole. purposes of the NPRM analysis. Either in mid-year compliance data Generally, all else being equal, using As always, however, ways to improve for MY 2016 or, separately and at the a newer analysis fleet will produce more the analysis fleet used for subsequent agencies’ invitation (as discussed realistic estimates of impacts of modeling to evaluate potential new above), most manufacturers identified potential new standards than using an CAFE and CO standards will undergo most of the technology already present outdated analysis fleet. However, among 2 continuous consideration. As described in each of their MY 2016 vehicle model/ relatively current options, a balance above, the compliance data is only the configurations. This information was must be struck between, on one hand, starting point for developing the not as complete for all manufacturers’ inputs’ freshness, and on the other, products as needed for today’s analysis, 72 analysis fleet; much additional inputs’ completeness and accuracy. so in some cases, information was During assembly of the inputs for information comes directly from manufacturers (such as details about supplemented with publicly available today’s analysis, final compliance data data, typically from manufacturer media was available for the MY 2015 model technologies, platforms, engines, transmissions, and other vehicle sites. In limited cases, manufacturers year but not in a few cases for MY 2016. did not supply information, and However, between mid-year compliance information, that may not be present in compliance data), and other information information from commercial and information and manufacturers’ specific publicly available sources was used. updates discussed above, a robust and must come from commercial and public detailed characterization of the MY sources (for example, fleet-wide (d) Mapping Technology Content of 2016 fleet was developed. However, information like market share, because 2016 Fleet to Argonne Technology while information continued to develop individual manufacturers do not Effectiveness Simulation Work provide this kind of information). If regarding the MY 2017 and, to a lesser While each vehicle model/ newer compliance data (i.e., MY 2017) extent MY 2018 and even MY 2019 configuration in the analysis fleet has its becomes available and can be analyzed fleets, this information was—even in list of observed technologies and mid-2017—too incomplete and during the pendency of this rulemaking, equipment, the ways in which inconsistent to be assembled with and if all of the other necessary steps manufacturers apply technologies and can be performed, the analysis fleet will 71 equipment do not always coincide For example, in 2016 comments to dockets be updated, as feasible, and made perfectly with how the analysis EPA–HQ–OAR–2015–0827 and NHTSA–2016– publicly available. The agencies seek 0068, UCS stated ‘‘in modeling technology characterizes the various technologies effectiveness and use, the agencies should use 2010 comment on the option used today and that improve fuel economy and reduce levels of performance as the baseline.’’ (UCS at 4, any other options, as well as on the CO2 emissions. To improve how the Docket ID. EPA–HQ–OAR–2015–0827–4016). tradeoffs between, on one hand, fidelity 72 Comments provided through a recent peer observed vehicle fleet ‘‘fits into’’ the review of the CAFE model recognize the need for with manufacturers’ actual plans and, analysis, each vehicle model/ this balance. For example, referring to NHTSA’s on the other, the ability to make detailed configuration is ‘‘mapped’’ to the full- 2016 analysis documented in the draft TAR, one of analysis inputs and outputs publicly the peer reviewers commented as follows: ‘‘The available. NHTSA decision to use MY 2015 data is wise. In 73 These technologies are generally grouped into the TAR they point out that a MY 2016 foundation (c) Observed Technology Content of the following categories: Vehicle technologies would require the use of confidential data, which include mass reduction, aerodynamic drag is less desirable. Clearly they would also have a 2016 Fleet reduction, low rolling resistance tires, and others. qualitative vision of the MY 2016 landscape while Engine technologies include engine attributes employing MY 2015 as a foundation. Although MY As explained above, the analysis fleet describing fuel type, engine aspiration, valvetrain 2015 data may still be subject to minor revision, is defined not only by the vehicle configuration, compression ratio, number of this is unlikely to impact the predictive ability of models/configurations it contains but cylinders, size of displacement, and others. the model . . . A more complex alternative Transmission technologies include different approach might be to employ some 2016 changes also by the technologies on those transmission arrangements like manual, 6-speed in technology, and attempt a blend of MY 2015 and vehicles. Each vehicle model/ automatic, 8-speed automatic, continuously MY 2016, while relying of estimation gained from configuration in the analysis fleet has an variable transmission, and dual-clutch only MY 2015 for sales. This approach may add associated list of observed technologies transmissions. Hybrid and electric powertrains may some relevancy in terms of technology, but might complement traditional engine and transmission introduce substantial error in terms of sales.’’ and equipment that can improve fuel designs or replace them entirely.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00022 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43007

vehicle simulation modeling 74 by Instead, Argonne simulated 10 different observed technologies and equipment Argonne National Laboratory that is vehicle types, corresponding to the on that vehicle.75 This mapping to a used to estimate the effectiveness of the ‘‘technology classes’’ generally used in specific simulation result most closely fuel economy-improving/CO2 CAFE analysis over the past several representing a given vehicle model/ emissions-reducing technologies rulemakings (e.g., small car, small configuration’s initial technology considered. Argonne produces full- performance car, pickup truck, etc.). ‘‘state’’ enables the CAFE model to vehicle simulation modeling for many Each of those 10 different vehicle types estimate the same vehicle model/ combinations of technologies, on many was assigned a set of ‘‘baseline configuration’s fuel economy after types of vehicles, but it did not simulate characteristics,’’ to which Argonne application of some other combination literally every single vehicle model/ added combinations of fuel-saving of technologies, leading to an alternative configuration in the analysis fleet technologies and then ran simulations technology state. because it would be impractical to to determine the fuel economy achieved BILLING CODE 4910–59–P assemble the requisite detailed when applying each combination of information—much of which would technologies to that vehicle type given 75 Mapping vehicle model/configurations in the likely only be provided on a its baseline characteristics. These analysis fleet to Argonne simulations was generally confidential basis—specific to each inputs, discussed at greater length in straightforward, but occasionally the mapping was vehicle model/configuration and Sections II.D and II.G, provide the basis complicated by factors like a vehicle model/ because the scale of the simulation for the CAFE model’s estimation of fuel configuration being a great match for simulations within more than one technology class (in which effort would correspondingly increase economy levels and CO2 emission rates. case, the model/configuration was assigned to the by at least two orders of magnitude. In the analysis fleet, inputs assign technology class that it best matched), or when each specific vehicle model/ technologies on the model/configuration were 74 Full-vehicle simulation modeling uses software configuration to a technology class, and difficult to observe directly (like friction reduction and physics models to compute and estimate energy once there, map to the simulation or parasitic loss characteristics of a transmission, in use of a vehicle during explicit driving conditions. which case the agencies relied on manufacturer- Section II.D below contains more information on within that technology class most reported data or CBI to help map the vehicle to a the Argonne work for this analysis. closely matching the combination of simulation).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00023 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43008 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00024 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.009 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43009

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00025 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.010 43010 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00026 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.011 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43011

BILLING CODE 4910–59–C may appear on many vehicles on a families to retain reasonable powertrain (e) Shared Vehicle Platforms, Engines, platform, and changes to that engine complexity.82 and Transmissions may or may not carry through to all the Like with engines, manufacturers vehicles. Some engines are shared often use transmissions that are the Another aspect of characterizing same or similar on multiple vehicles.83 vehicle model/configurations in the across a range of different vehicle To reflect this reality, shared analysis fleet is based on whether they platforms. Vehicle model/configurations transmissions were considered for share a ‘‘platform’’ with other vehicle in the analysis fleet that share engines manufacturers as appropriate. To define model/configurations. A ‘‘platform’’ belonging to the same platform are also refers to engineered underpinnings identified as such. common transmissions, the agencies shared on several differentiated It is important to note that considered transmission type (manual, products. Manufacturers share and manufacturers define common engines automatic, dual-clutch, continuously standardize components, systems, differently. Some manufacturers variable), number of gears, and vehicle tooling, and assembly processes within consider engines as ‘‘common’’ if the architecture (front-wheel-drive, rear- their products (and occasionally with engines shared an architecture, wheel-drive, all-wheel-drive based on a the products of another manufacturer) to components, or manufacturing front-wheel-drive platform, or all-wheel- cost-effectively maintain vibrant processes. Other manufacturers take a drive based on a rear-wheel-drive portfolios.76 narrower definition, and only assume platform). If vehicles shared these Vehicle model/configurations derived ‘‘common’’ engines if the parts in the attributes, these transmissions were from the same platform are so identified engine assembly are the same. In some grouped for the analysis. Vehicles in the in the analysis fleet. Many analysis fleet with the same cases, manufacturers designate each 84 manufacturers’ use of vehicle platforms engine in each application as a unique transmission configuration will adopt is well documented in the public record 79 transmission technology together, as powertrain. Engine families for each 85 and widely recognized among the manufacturer were tabulated and described above. vehicle engineering community. assigned 80 based on data-driven Having all vehicles that share a Engineering knowledge, information criteria. If engines shared a common platform (or engines that are part of a from trade publications, and feedback cylinder count and configuration, family) adopt fuel economy-improving/ from manufacturers and suppliers was displacement, valvetrain, and fuel type, CO2 emissions-reducing technologies also used to assign vehicle platforms in those engines may have been considered together, subject to refresh/redesign the analysis fleet. together. Additionally, if the constraints, reflects the real-world When the CAFE model is deciding compression ratio, horsepower, and considerations described above but also where and how to add technology to displacement of engines were only overlooks some decisions manufacturers vehicles, if one vehicle on the platform slightly different, those engines were might make in the real world in receives new technology, other vehicles considered to be the same for the response to market pull, meaning that on the platform also receive the purposes of redesign and sharing. even though the analysis fleet is technology as part of their next major Vehicles in the analysis fleet with the incredibly complex, it is also over- 77 redesign or refresh. Similar to vehicle same engine family will therefore adopt simplified in some respects compared to platforms, manufacturers create engines engine technology in a coordinated the real world. For example, the CAFE 78 that share parts. One engine family fashion.81 By grouping engines together, model does not currently attempt to the CAFE model controls future engine simulate the potential for a 76 The concept of platform sharing has evolved manufacturer to shift the application of with time. Years ago, manufacturers rebadged technologies to improve performance vehicles and offered luxury options only on out common engine block castings with different premium nameplates (and manufacturers shared diameters to create engines with an array of rather than fuel economy. Therefore, the some vehicle platforms in limited cases). Today, displacements. Head assemblies for different model’s representation of the manufacturers share parts across highly displacement engines may share many components ‘‘inheritance’’ of technology can lead to differentiated vehicles with different body styles, and manufacturing processes across the engine estimates a manufacturer might sizes, and capabilities that may share the same family. Manufacturers may finish crankshafts with platform. For instance, the Honda Civic and Honda the same tools, to similar tolerances. Engines on the eventually exceed fuel economy CR–V share many parts and are built on the same same architecture may share pistons, connecting platform. Engineers design chassis platforms with rods, and the same engine architecture may include 82 This does mean, however, that for the ability to vary wheelbase, ride height, and even both six and eight cylinder engines. manufacturers that submitted highly atomized driveline configuration. Assembly lines can 79 For instance, a manufacturer may have listed engine and transmission portfolios, there is a produce hatchbacks and sedans to cost-effectively two engines for a pair that share designs for the practical cap on powertrain complexity and the utilize manufacturing capacity and respond to shifts engine block, the crank shaft, and the head because ability of the manufacturer to optimize the in market demand. Engines made on the same line the accessory drive components, oil pans, and displacement of (a.k.a. ‘‘right size’’) engines may power small cars or mid-size sport utility engine calibrations differ between the two. In perfectly for each vehicle configuration. vehicles. Additionally, although the agencies’ practice, many engines share parts, tooling, and 83 Manufacturers may produce transmissions that analysis, like past CAFE analyses, considers assembly resources, and manufacturers often have nominally different machining to castings, or vehicles produced for sale in the U.S., the agency coordinate design updates between two similar manufacturers may produce transmissions that are notes these platforms are not constrained to vehicle engines. internally identical, except for final output gear models built for sale in the United States; many 80 Engine family is referred to in the analysis as ratio. In some cases, manufacturers sub-contract manufacturers have developed, and use, global an ‘‘engine code.’’ with suppliers that deliver whole transmissions. In platforms, and the total number of platforms is 81 Specifically, if such vehicles have different other cases, manufacturers form joint-ventures to decreasing across the industry. Several automakers design schedules (i.e., refresh and redesign develop shared transmissions, and these (for example, General Motors and Ford) either plan schedules), and a subset of vehicles using a given transmission platforms may be offered in many to, or already have, reduced their number of engine add engine technologies in the course of a vehicles across manufacturers. Manufacturers use platforms to less than 10 and account for the redesign or refresh that occurs in an early model supplier and joint-venture transmissions to a greater overwhelming majority of their production volumes year (e.g., 2018), other vehicles using the same extent than engines. on that small number of platforms. engine ‘‘inherit’’ these technologies at the soonest 84 Transmission configurations are referred to in 77 The CAFE model assigns mass reduction ensuing refresh or redesign. This is consistent with the analysis as ‘‘transmission codes.’’ technology at a platform level, but many other a view that, over time, most manufacturers are 85 Similar to the inheritance approach outlined technologies may be assigned and shared at a likely to find it more practicable to shift production for engines, if one vehicle application of a shared vehicle nameplate or vehicle model level. to a new version of an engine than to indefinitely transmission family upgraded the transmission, 78 For instance, manufacturers may use different continue production of both the new engine and a other vehicle applications also upgraded the piston strokes on a common engine block or bore ‘‘legacy’’ engine. transmission at the next refresh or redesign year.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00027 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43012 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

standards as technology continues to In addition to refresh and redesign cycles (e.g., four to five years), and this propagate across shared platforms and schedules associated with vehicle approach was considered for the engines. In the past, there were some model/configurations, vehicles that analysis.87 Some manufacturers tend to examples of extended periods during share a platform subsequently have operate with faster redesign cycles and which some manufacturers exceeded platform-wide refresh and redesign may continue to do so, and one or both of the CAFE and/or GHG schedules for mass reduction manufacturers tend to redesign some standards, but in plenty of other technologies. products more frequently than others. examples, manufacturers chose to To develop the refresh/redesign However, especially considering that introduce (or even reintroduce) cycles used for the MY 2016 vehicles in information presented by manufacturers technological complexity into their the analysis fleet, information from largely supports estimates discussed vehicle lineups in response to buyer commercially available sources was above, applying a ‘‘one size fits all’’ preferences. Going forward, and used to project redesign cycles through acceleration of redesign cycles would recognizing the recent trend for MY 2022, as for NHTSA’s analysis for likely not improve the analysis; instead, consolidating platforms, it seems likely the Draft TAR published in 2016.86 doing so would likely reduce manufacturers will be more likely to Commercially available sources’ consistency with the real world, choose efficiency over complexity in estimates through MY 2022 are especially for light trucks. Moreover, if this regard; therefore, the potential generally supported by detailed some manufacturers accelerate should be lower that today’s analysis consideration of public announcements redesigns in response to new standards, turns out to be over-simplified plus related intelligence from suppliers doing so would likely involve costs compared to the real world. and other sources, and recognize that (greater levels of stranded capital, Options will be considered to further uncertainty increases considerably as reduced opportunity to benefit from refine the representation of sharing and the forecasting horizon is extended. For ‘‘learning’’-related cost reductions) inheritance of technology, possibly MYs 2023–2035, in recognition of that greater than reflected in other inputs to including model revisions to account for uncertainty, redesign schedules were the analysis. However, a wider range of tradeoffs between fuel economy and extended considering past pacing for technologies can practicably be applied performance when applying technology. each product, estimated schedules during mid-cycle ‘‘freshenings’’ than Please provide comments on the sharing through MY 2022, and schedules for has been represented by past analyses, and inheritance-related aspects of the other products in the same technology and this part of the analysis has been analysis fleet and the CAFE model along classes. As mentioned above, potentially expanded, as discussed below in with information that would support confidential forward-looking Section II.D.88 Also, in the sensitivity refinement of the current approach or information was not requested from analysis supporting today’s proposal development and implementation of manufacturers; nevertheless, all and presented in Chapter 13 of the alternative approaches. manufacturers had an opportunity to PRIA, one case involving faster redesign review the estimates of product-specific schedules and one involving slower (f) Estimated Product Design Cycles redesign schedules, a few manufacturers redesign schedules has been analyzed. provided related forecasts and, for the Another aspect of characterizing Manufacturers use diverse strategies most part, that information corroborated vehicle model/configurations in the with respect to when, and how often analysis fleet is based on when they can the estimates. Some commenters suggested they update vehicle designs. While most next be refreshed or redesigned. vehicles have been redesigned sometime Redesign schedules play an important supplanting these estimated redesign schedules with estimates applying faster in the last five years, many vehicles role in determining when new have not. In particular, vehicles with technologies may be applied. Many 86 In some cases, data from commercially lower annual sales volumes tend to be technologies that improve fuel economy available sources was found to be incomplete or redesigned less frequently, perhaps and reduce CO2 emissions may be inconsistent with other available information. For giving manufacturers more time to difficult to incorporate without a major instance, commercially available sources identified amortize the investment needed to bring product redesign. Therefore, each some newly imported vehicles as new platforms, but the international platform was midway through the product to market. In some cases, vehicle model in the analysis fleet has the product lifecycle. While new to the U.S. market, manufacturers continue to produce and an associated redesign schedule, and the treating these vehicles as new entrants would have sell vehicles designed more than a CAFE model uses that schedule to resulted in artificially short redesign cycles if decade ago. restrict significant advances in some carried forward, in some cases. Similarly, commercially available sources labeled some technologies (like major mass reduction) product refreshes as redesigns, and vice versa. In 87 In response to the EPA’s August 21, 2017, to redesign years, while allowing these limited cases, the data was revised to be Request for Comments (docket numbers EPA–HQ– manufacturers to include minor consistent with other available information or OAR–2015–0827 and NHTSA–2016–0068), see, e.g., advances (such as improved tire rolling typical redesign and refresh schedules, for the CARB at 28 (Docket ID. EPA–HQ–OAR–2015–0827– purpose of the CAFE modeling. In these limited 9197), EDF at 12 (Docket ID. EPA–HQ–OAR–2015– resistance) during a vehicle ‘‘refresh,’’ or cases, the forecast time between redesigns and 0827–9203), and NRDC, et. al. at 29–33 (Docket ID. a smaller update made to a vehicle, refreshes was updated to match the observed past EPA–HQ–OAR–2015–0827–9826). which can happen between redesigns. product timing. 88 NRDC, et al., at 32.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00028 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43013

BILLING CODE 4910–59–P refresh and redesign cycles. Table II–3 vehicle technology class in the MY 2016 Each manufacturer may use different below summarizes the average time analysis fleet. Across the industry, strategies throughout their product between redesigns, by manufacturer, by manufacturers average 6.5 years portfolio, and a component of each vehicle technology class.89 Dashes mean between product redesigns. strategy may include the timing of the manufacturer has no volume in that

89 Technology class, or tech class, refers to a vehicles. As explained, each vehicle is assigned to a representative simulation to estimate technology group of fuel-economy simulations of similar effectiveness for purposes of the analysis.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00029 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.012 43014 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

There are a few notable observations industry, light-duty vehicle designs last technology class. A value of ‘‘0.0’’ from this table. Pick-up trucks have for about 6.5 years. means that every vehicle for a much longer redesign schedules (7.8 Even if two manufacturers have manufacturer in that vehicle technology years on average) than small cars (5.5 similar redesign cadence, the model class, represented in the MY 2016 years on average). Some manufacturers years in which the redesigns occur may analysis fleet was new in MY 2016. redesign vehicles often (every 5.2 years still be different and dependent on Across the industry, manufacturers in the case of Hyundai), while other where each of the manufacturer’s redesigned MY 2016 vehicles an average manufacturers redesign vehicles less products are in their life cycle. of 3.2 years earlier. often (FCA waits on average 8.6 years Table II–4 summarizes the average age between vehicle redesigns). Across the of manufacturers’ offering by vehicle

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00030 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.013 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43015

BILLING CODE 4910–59–C vehicle emissions. EPCA, as amended methodologies used to develop current Based on historical observations and by EISA, expressly requires that CAFE attribute-based standards, and the refresh/redesign schedule forecasts, standards for passenger cars and light agencies’ current proposal to continue careful consideration to redesign cycles trucks be based on one or more vehicle to do so for MYs 2022–2026. Section for each manufacturer and each vehicle attributes related to fuel economy and II.C.5 discusses the methodologies used is important. Simply assuming every be expressed in the form of a to reconsider the mathematical function vehicle is redesigned by 2021 and by mathematical function.90 While the for the proposed standards. 2025 is not appropriate, as this would CAA includes no specific requirements misrepresent both the likely timing of regarding GHG regulation, EPA has 1. Why attribute-based standards, and redesigns and the likely time between chosen to adopt standards consistent what are the benefits? redesigns in most cases. with the EPCA/EISA requirements in the interest of simplifying compliance Under attribute-based standards, C. Development of Footprint-Based 91 every vehicle model has fuel economy Curve Shapes for the industry since 2010. Section II.C.1 describes the advantages of and CO2 targets, the levels of which As in the past four CAFE rulemakings, attribute standards, generally. Section depend on the level of that vehicle’s the most recent two of which included II.C.2 explains the agencies’ specific determining attribute (for this proposed related standards for CO2 emissions, decision to use vehicle footprint as the rule, footprint is the determining NHTSA and EPA are proposing to set attribute over which to vary stringency attribute, as discussed below). The attribute-based CAFE standards that are for past and current rules. Section II.C.3 manufacturer’s fleet average defined by a mathematical function of discusses the policy considerations in performance is calculated by the vehicle footprint, which has observable selecting the specific mathematical harmonic production-weighted average correlation with fuel economy and function. Section II.C.4 discusses the of those targets, as defined below:

90 49 U.S.C. 32902(a)(3)(A). 91 Such an approach is permissible under section based approach in issuing standards under 202(a) of the CAA, and EPA has used the attribute- analogous provisions of the CAA.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00031 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.014 43016 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Here, i represents a given model 92 in vehicle model has its own target 2. Why footprint as the attribute? a manufacturer’s fleet, Productioni (determined by a size-related attribute), It is important that the CAFE and CO2 represents the U.S. production of that properly fitted attribute-based standards standards be set in a way that does not model, and Targeti represents the target provide little, if any, incentive to build encourage manufacturers to respond by as defined by the attribute-based smaller vehicles simply to meet a fleet- selling vehicles that are less safe. standards. This means no vehicle is wide average, because smaller vehicles Vehicle size is highly correlated with required to meet its target; instead, are subject to more stringent compliance vehicle safety—for this reason, it is manufacturers are free to balance targets. important to choose an attribute improvements however they deem best correlated with vehicle size (mass or within (and, given credit transfers, at Second, attribute-based standards, if properly fitted, better respect some dimensional measure). Given this least partially across) their fleets. consideration, there are several policy The idea is to select the shape of the heterogeneous consumer preferences and technical reasons why footprint is mathematical function relating the than do single-valued standards. As considered to be the most appropriate standard to the fuel economy-related discussed above, a single-valued attribute upon which to base the attribute to reflect the trade-offs standard encourages a fleet mix with a standards, even though other vehicle manufacturers face in producing more larger share of smaller vehicles by of that attribute over fuel efficiency (due creating incentives for manufacturers to size attributes (notably, curb weight) are to technological limits of production use downsizing the average vehicle in more strongly correlated with fuel economy and emissions. and relative demand of each attribute). their fleet (possibly through fleet First, mass is strongly correlated with If the shape captures these trade-offs, mixing) as a compliance strategy, which fuel economy; it takes a certain amount every manufacturer is more likely to may result in manufacturers building of energy to move a certain amount of continue adding fuel efficient vehicles for compliance reasons that mass. Footprint has some positive technology across the distribution of the consumers do not want. Under a size- attribute within their fleet, instead of correlation with frontal surface area, related, attribute-based standard, likely a negative correlation with potentially changing the attribute—and reducing the size of the vehicle is a less other correlated attributes, including aerodynamics, and therefore fuel viable compliance strategy because economy, but the relationship is less fuel economy—as a part of their smaller vehicles have more stringent compliance strategy. Attribute-based deterministic. Mass and crush space regulatory targets. As a result, the fleet (correlated with footprint) are both standards that achieve this have several mix under such standards is more likely advantages. important safety considerations. As First, assuming the attribute is a to reflect aggregate consumer demand discussed below and in the measurement of vehicle size, attribute- for the size-related attribute used to accompanying PRIA, NHTSA’s research based standards reduce the incentive for determine vehicle targets. of historical crash data indicates that manufacturers to respond to CAFE Third, attribute-based standards holding footprint constant, and standards by reducing vehicle size in provide a more equitable regulatory decreasing the mass of the largest ways harmful to safety.93 Larger framework across heterogeneous vehicles, will have a net positive safety vehicles, in terms of mass and/or crush manufacturers who may each produce impact to drivers overall, while holding space, generally consume more fuel, but different shares of vehicles along footprint constant and decreasing the are also generally better able to protect attributes correlated with fuel mass of the smallest vehicles will have occupants in a crash.94 Because each economy.95 A single, industry-wide a net decrease in fleetwide safety. Properly fitted footprint-based standards CAFE standard imposes 92 provide little, if any, incentive to build If a model has more than one footprint variant, disproportionate cost burden and here each of those variants is treated as a unique smaller vehicles to meet CAFE and CO compliance challenges on 2 model, i, since each footprint variant will have a standards, and therefore help minimize unique target. manufacturers who produce more the impact of standards on overall fleet 93 The 2002 NAS Report described at length and vehicles with attributes inherently safety. quantified the potential safety problem with average correlated with lower fuel economy— fuel economy standards that specify a single Second, it is important that the numerical requirement for the entire industry. See i.e. manufacturers who produce, on attribute not be easily manipulated in a Transportation Research Board and National average, larger vehicles. As discussed manner that does not achieve the goals Research Council. 2002. Effectiveness and Impact of above, retaining the ability for Corporate Average Fuel Economy (CAFE) of EPCA or other goals, such as safety. Standards, Washington, DC: The National manufacturers to produce vehicles Although weight is more strongly Academies Press (‘‘2002 NAS Report’’) at 5, finding which respect heterogeneous market correlated with fuel economy than 12, available at https://www.nap.edu/catalog/ preferences is an important footprint, there is less risk of 10172/effectiveness-and-impact-of-corporate- consideration. Since manufacturers may average-fuel-economy-cafe-standards (last accessed manipulation (changing the attribute(s) June 15, 2018). Ensuing analyses, including by target different markets as a part of their to achieve a more favorable target) by NHTSA, support the fundamental conclusion that business strategy, ensuring that these increasing footprint under footprint- standards structured to minimize incentives to manufacturers do not incur a downsize all but the largest vehicles will tend to based standards than there would be by produce better safety outcomes than flat standards. disproportionate share of the regulatory increasing vehicle mass under weight- 94 Bento, A., Gillingham, K., & Roth, K. (2017). cost burden is an important part of based standards. It is relatively easy for The Effect of Fuel Economy Standards on Vehicle conserving consumer choices within the a manufacturer to add enough weight to Weight Dispersion and Accident Fatalities. NBER market. Working Paper No. 23340. Available at http:// a vehicle to decrease its applicable fuel www.nber.org/papers/w23340 (last accessed June economy target a significant amount, as 15, 2018). 95 2002 NAS Report at 4–5, finding 10. compared to increasing vehicle

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00032 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.015 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43017

footprint, which is a much more nature of the function. The CAA • Given the same industry-wide complicated change that typically takes provides no specific direction regarding average required fuel economy or CO2 place only with a vehicle redesign. CO2 regulation, and EPA has continued standard, dramatically steeper standards Further, some commenters on the MY to harmonize this aspect of its CO2 tend to place greater compliance 2011 CAFE rulemaking were concerned regulations with NHTSA’s CAFE burdens on limited-line manufacturers that there would be greater potential for regulations. The relationship between (depending of course, on which vehicles gaming under multi-attribute standards, fuel economy (and GHG emissions) and are being produced). such as those that also depend on footprint, though directionally clear • If cutpoints are adopted, given the weight, torque, power, towing (i.e., fuel economy tends to decrease and same industry-wide average required capability, and/or off-road capability. As CO2 emissions tend to increase with fuel economy, moving small-vehicle discussed in NHTSA’s MY 2011 CAFE increasing footprint), is theoretically cutpoints to the left (i.e., up in terms of 96 final rule, it is anticipated that the vague, and quantitatively uncertain; in fuel economy, down in terms of CO2 possibility of gaming is lowest with other words, not so precise as to a priori emissions) discourages the introduction footprint-based standards, as opposed to yield only a single possible curve. of small vehicles, and reduces the weight-based or multi-attribute-based The decision of how to specify this incentive to downsize small vehicles in standards. Specifically, standards that mathematical function therefore reflects ways that could compromise overall incorporate weight, torque, power, some amount of judgment. The function highway safety. towing capability, and/or off-road can be specified with a view toward • If cutpoints are adopted, given the capability in addition to footprint would achieving different environmental and same industry-wide average required not only be more complex, but by petroleum reduction goals, encouraging fuel economy, moving large-vehicle providing degrees of freedom with different levels of application of fuel- cutpoints to the right (i.e., down in respect to more easily-adjusted saving technologies, avoiding any terms of fuel economy, up in terms of attributes, they could make it less adverse effects on overall highway CO emissions) better accommodates the certain that the future fleet would 2 safety, reducing disparities of design requirements of larger vehicles actually achieve the projected average manufacturers’ compliance burdens, — especially large pickups — and fuel economy and CO levels. This is 2 and preserving consumer choice, among extends the size range over which not to say that a footprint-based system other aims. The following are among the downsizing is discouraged. will eliminate gaming, or that a specific technical concerns and footprint-based system will eliminate resultant policy tradeoffs the agencies 4. What mathematical functions have the possibility that manufacturers will have considered in selecting the details been used previously, and why? change vehicles in ways that of specific past and future curve shapes: compromise occupant protection, but Notwithstanding the aforementioned • Flatter standards (i.e., curves) footprint-based standards achieve the discretion under EPCA/EISA, data increase the risk that both the size of best balance among affected should inform consideration of potential vehicles will be reduced, potentially considerations. Please provide mathematical functions, but how comments on whether vehicular compromising highway safety, and relevant data is defined and interpreted, footprint is the most suitable attribute reducing any utility consumers would and the choice of methodology for upon which to base standards. have gained from a larger vehicle. fitting a curve to that data, can and • Steeper footprint-based standards should include some consideration of 3. What mathematical function should may create incentives to upsize specific policy goals. This section be used to specify footprint-based vehicles, potentially oversupplying summarizes the methodologies and standards? vehicles of certain footprints beyond policy concerns that were considered in In requiring NHTSA to ‘‘prescribe by what consumers would naturally developing previous target curves (for a regulation separate average fuel demand, and thus increasing the complete discussion see the 2012 FRIA). economy standards for passenger and possibility that fuel savings and CO2 As discussed below, the MY 2011 non-passenger automobiles based on 1 reduction benefits will be forfeited final curves followed a constrained or more vehicle attributes related to fuel artificially. logistic function defined specifically in economy and express each standard in • Given the same industry-wide the final rule.97 The MYs 2012–2021 the form of a mathematical function’’, average required fuel economy or CO2 final standards and the MYs 2022–2025 EPCA/EISA provides ample discretion standard, flatter standards tend to place augural standards are defined by regarding not only the selection of the greater compliance burdens on full-line constrained linear target functions of attribute(s), but also regarding the manufacturers. footprint, as shown below: 98

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

96 See 74 FR at 14359 (Mar. 30, 2009). 98 The right cutpoint for the light truck curve was for light trucks for MYs 2017–2021 is the maximum 97 See 74 FR 14196, 14363–14370 (Mar. 30, 2009) moved further to the right for MYs 2017–2021, so of the MY 2016 target curves and the target curves for NHTSA discussion of curve fitting in the MY that more possible footprints would fall on the for the give MY standard. This is defined further in sloped part of the curve. In order to ensure that, for 2011 CAFE final rule. the 2012 final rule. See 77 FR 62624, at 62699–700 all possible footprints, future standards would be at (Oct. 15, 2012). least as high as MY 2016 levels, the final standards

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00033 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.016 43018 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

slope, c, and the intercept, d, of the functions defining the standards. as percentage improvements over a linear portion of the curve are specified NHTSA then identified footprints at previous MY target curve. as gpm per change in square feet, and which to apply minimum and (c) MYs 2017 and Beyond Standards gpm, respectively. maximum values (rather than letting the (Constrained Linear) The min and max functions will take standards extend without limit) and the minimum and maximum values transposed these functions vertically The mathematical functions finalized within their associated parentheses. (i.e., on a gallons-per-mile basis, in 2012 for MYs 2017 and beyond Thus, the max function will first find uniformly downward) to produce the changed somewhat from the functions the maximum of the fitted line at a promulgated standards. In the preceding for the MYs 2012–2016 standards. These given footprint value and the lower rule, for MYs 2008–2011 light truck changes were made to both address asymptote from the perspective of gpm. standards, NHTSA examined a range of comments from stakeholders, and to If the fitted line is below the lower potential functional forms, and further consider some of the technical asymptote it is replaced with the floor, concluded that, compared to other concerns and policy goals judged more which is also the minimum of the floor considered forms, the constrained preeminent under the increased and the ceiling by definition, so that the logistic form provided the expected and uncertainty of the impacts of finalizing target in mpg space will be the appropriate trend (decreasing fuel and proposing standards for model 103 reciprocal of the floor in mpg space, or economy as footprint increases), but years further into the future. simply, a. If, however, the fitted line is avoided creating ‘‘kinks’’ the agency Recognizing the concerns raised by full- not below the lower asymptote, the was concerned would provide line OEMs, it was concluded that fitted value is returned from the max distortionary incentives for vehicles continuing increases in the stringency of function and the min function takes the with neighboring footprints.100 the light truck standards would be more minimum value of the upper asymptote feasible if the light truck curve for MYs (in gpm space) and the fitted line. If the (b) MYs 2012–2016 Standards 2017 and beyond was made steeper than fitted value is below the upper (Constrained Linear) the MY 2016 truck curve and the right asymptote, it is between the two For the MYs 2012–2016 rule, (large footprint) cut-point was extended asymptotes and the fitted value is potential methods for specifying only gradually to larger footprints. To appropriately returned from the min mathematical functions to define fuel accommodate these considerations, the 2012 final rule finalized the slope fit to function, making the overall target in economy and CO2 standards were mpg the reciprocal of the fitted line in reevaluated. These methods were fit to the MY 2008 fleet using a sales- gpm. If the fitted value is above the the same MY 2008 data as the MY 2011 weighted, ordinary least-squares upper asymptote, the upper asymptote standard. Considering these further regression, using a fleet that had is returned is returned from the min specifications, the constrained logistic technology applied to make the function, and the overall target in mpg form, if applied to post-MY 2011 technology application across the fleet is the reciprocal of the upper asymptote standards, would likely contain a steep more uniform, and after adjusting the in gpm space, or b. mid-section that would provide undue data for the effects of weight-to- In this way curves specified as incentive to increase the footprint of footprint. Information from an updated constrained linear functions are midsize passenger cars.101 A range of MY 2010 fleet was also considered to specified by the following parameters: methods to fit the curves would have support this decision. As the curve was vertically shifted (with fuel economy a = upper limit (mpg) been reasonable, and a minimum specified as mpg instead of gpm or CO b = lower limit (mpg) absolute deviation (MAD) regression 2 emissions) upwards, the right cutpoint c = slope (gpm per sq. ft.) without sales weighting on a was progressively moved for the light d = intercept (gpm) technology-adjusted car and light truck truck curves with successive model The slope and intercept are specified fleet was used to fit a linear equation. years, reaching the final endpoint for as gpm per sq. ft. and gpm instead of This equation was used as a starting MY 2021; this is further discussed and mpg per sq. ft. and mpg because fuel point to develop mathematical functions shown in Chapter 4.3 of the PRIA. consumption and emissions appear defining the standards. Footprints were roughly linearly related to gallons per then identified at which to apply 5. Reconsidering the Mathematical mile (the reciprocal of the miles per minimum and maximum values (rather Functions for This Proposal gallon). than letting the standards extend (a) Why is it important to reconsider the (a) NHTSA in MY 2008 and MY 2011 without limit). Finally, these mathematical functions? CAFE (Constrained Logistic) constrained/piecewise linear functions were transposed vertically (i.e., on a By shifting the developed curves by a single factor, it is assumed that the For the MY 2011 CAFE rule, NHTSA gpm or CO2 basis, uniformly downward) estimated fuel economy levels by by multiplying the initial curve by a underlying relationship of fuel footprint from the MY 2008 fleet after single factor for each MY standard to consumption (in gallons per mile) to normalization for differences in produce the final attribute-based targets vehicle footprint does not change technology,99 but did not make for passenger cars and light trucks significantly from the model year data adjustments to reflect other vehicle described in the final rule.102 These used to fit the curves to the range of attributes (e.g., power-to-weight ratios). transformations are typically presented model years for which the shifted curve Starting with the technology-adjusted shape is applied to develop the passenger car and light truck fleets, 100 See 71 FR 17556, 17609–17613 (Apr. 6, 2006) standards. However, it must be NHTSA used minimum absolute for NHTSA discussion of ‘‘kinks’’ in the MYs 2008– recognized that the relationship deviation (MAD) regression without 2011 light truck CAFE final rule (there described as sales weighting to fit a logistic form as ‘‘edge effects’’). A ‘‘kink,’’ as used here, is a portion 103 The MYs 2012–2016 final standards were of the curve where a small change in footprint signed April 1st, 2010—putting 6.5 years between a starting point to develop mathematical results in a disproportionally large change in its signing and the last affected model year, while stringency. the MYs 2017–2021 final standards were signed 99 See 74 FR 14196, 14363–14370 (Mar. 30, 2009) 101 75 FR at 25362. August 28th, 2012—giving just more than nine for NHTSA discussion of curve fitting in the MY 102 See generally 74 FR at 49491–96; 75 FR at years between signing and the last affected final 2011 CAFE final rule. 25357–62. standards.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00034 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43019

between vehicle footprint and fuel conducted that do not require adjustments to the data to make the economy is not necessarily constant underlying engineering models of how fleets less technologically heterogeneous over time; newly developed fuel economy and footprint might be were performed. There were several technologies, changes in consumer expected to be related, and a separate adjustments to the data that were demand, and even the curves analysis that uses vehicle simulation common to all of the statistical analyses themselves could, if unduly susceptible results as the basis to estimate the considered. to gaming, influence the observed relationship from a perspective more With a view toward isolating the relationships between the two vehicle explicitly informed by engineering relationship between fuel economy and characteristics. For example, if certain theory was conducted as well. Despite footprint, the few diesels in the fleet technologies are more effective or more changes in the new vehicle fleet both in were excluded, as well as the limited marketable for certain types of vehicles, terms of technologies applied and in number of vehicles with partial or full their application may not be uniform market demand, the underlying electric propulsion; when the fleet is over the range of vehicle footprints. statistical relationship between footprint normalized so that technology is more Further, if market demand has shifted and fuel economy has not changed homogenous, application of these between vehicle types, so that certain significantly since the MY 2008 fleet technologies is not allowed. This is vehicles make up a larger share of the used for the 2012 final rule; therefore, consistent with the methodology used fleet, any underlying technological or it is proposed to continue to use the in the 2012 final rule. curve shapes fit in 2012. The analysis market restrictions which inform the The above adjustments were applied average shape of the curves could and reasoning supporting this decision follows. to all statistical analyses considered, change. That is, changes in the regardless of the specifics of each of the technology or market restrictions (b) What statistical analyses did NHTSA methods, weights, and technology level themselves, or a mere re-weighting of consider? of the data, used to view the different vehicles types, could reshape In considering how to address the relationship of vehicle footprint and the fit curves. various policy concerns discussed fuel economy. Table II–5, below, For the above reasons, the curve above, data from the MY 2016 fleet was summarizes the different assumptions shapes were reconsidered using the considered, and a number of descriptive considered and the key attributes of newest available data, from MY 2016. statistical analyses (i.e., involving each. The analysis was performed With a view toward corroboration observed fuel economy levels and considering all possible combinations of through different techniques, a range of footprints) using various statistical these assumptions, producing a total of descriptive statistical analyses were methods, weighting schemes, and eight footprint curves.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00035 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.017 43020 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(1) Current Technology Level Curves and fuel consumption. The results of all footprints; augmenting a largely updated analysis for maximum statistical analysis with an analysis that The ‘‘current technology’’ level curves technology level curves are also shown exclude diesels and vehicles with more explicitly incorporates engineering in Chapter 4.4.2.2 of the PRIA. electric propulsion, as discussed above, theory helps to corroborate that the Especially if vehicles progress over time but make no other changes to each relationship between fuel economy and toward more similar size-specific model year fleet. Comparing the MY footprint is in fact being characterized. efficiency, further removing variation in 2016 curves to ones built under the Tractive energy is the amount of technology application both better energy it will take to move a vehicle.104 same methodology from previous model isolates the relationship between fuel Here, tractive energy effectiveness is year fleets shows whether the observed consumption and footprint and further defined as the share of the energy curve shape has changed significantly supports the curve slopes finalized in content of fuel consumed which is over time as standards have become the 2012 final rule. more stringent. Importantly, these converted into mechanical energy and curves will include any market forces (c) What other methodologies were used to move a vehicle—for internal which make technology application considered? combustion engine (ICE) vehicles, this variable over the distribution of The methods discussed above are will vary with the relative efficiency of footprint. These market forces will not descriptive in nature, using statistical specific engines. Data from ANL be present in the ‘‘maximum analysis to relate observed fuel economy simulations suggest that the limits of technology’’ level curves: By making levels to observed footprints for known tractive energy effectiveness are technology levels homogenous, this vehicles. As such, these methods are approximately 25% for vehicles with variation is removed. The current clearly based on actual data, answering internal combustion engines which do technology level curves built using both the question ‘‘how does fuel economy not possess ISG, other hybrid, plug-in, regression types and both regression appear to be related to footprint?’’ pure electric, or fuel cell technology. weight methodologies from the MY However, being independent of explicit A tractive energy prediction model 2008, MY 2010, and MY 2016 fleets, engineering theory, they do not answer was also developed to support today’s shown in more detail in Chapter 4.4.2.1 the question ‘‘how might one expect proposal. Given a vehicle’s mass, frontal of the PRIA, support the curve slopes fuel economy to be related to footprint?’’ area, aerodynamic drag coefficient, and finalized in the 2012 final rule. The Therefore, as an alternative to the above rolling resistance as inputs, the model curves built from most methodologies methods, an alternative methodology will predict the amount of tractive using each fleet generally shift, but was also developed and applied that, energy required for the vehicle to remain very similar in slope. This using full-vehicle simulation, comes complete the Federal test cycle. This suggests that the relationship of closer to answer the second question, model was used to predict the tractive footprint to fuel economy, including providing a basis to either corroborate energy required for the average vehicle both technology and market limits, has answers to the first, or suggest that of a given footprint 105 and ‘‘body not significantly changed. further investigation could be technology package’’ to complete the (2) Maximum Technology Level Curves important. cycle. The body technology packages As discussed in the 2012 final rule, considered are defined in Table II–6, As in prior rulemakings, technology several manufacturers have below. Using the absolute tractive differences between vehicle models confidentially shared with the agencies were considered to be a significant what they described as ‘‘physics-based’’ energy predicted and tractive energy factor producing uncertainty regarding curves, with each OEM showing effectiveness values spanning possible the relationship between fuel significantly different shapes for the ICE engines, fuel economy values were consumption and footprint. Noting that footprint-fuel economy relationships. then estimated for different body attribute-based standards are intended This variation suggests that technology packages and engine tractive to encourage the application of manufacturers face different curves energy effectiveness values. additional technology to improve fuel given the other attributes of the vehicles 104 Thomas, J. ‘‘Drive Cycle Powertrain efficiency and reduce CO2 emissions in their fleets (i.e., performance Efficiencies and Trends Derived from EPA Vehicle across the distribution of footprint in characteristics) and/or that their curves Dynamometer Results,’’ SAE Int. J. Passeng. Cars— the fleet, approaches were considered in reflected different levels of technology Mech. Syst. 7(4):2014, doi:10.4271/2014–01–2562. which technology application is application. In reconsidering the shapes Available at https://www.sae.org/publications/ simulated for purposes of the curve of the proposed MYs 2021–2026 technical-papers/content/2014-01-2562/ (last accessed June 15, 2018). fitting analysis in order to produce fleets standards, a similar estimation of 105 The mass reduction curves used elsewhere in that are less varied in technology physics-based curves leveraging third- this analysis were used to predict the mass of a content. This approach helps reduce party simulation work form Argonne vehicle with a given footprint, body style box, and ‘‘noise’’ (i.e., dispersion) in the plot of National Laboratories (ANL) was mass reduction level. The ‘Body style Box’ is 1 for hatchbacks and minivans, 2 for pickups, and 3 for vehicle footprints and fuel consumption developed. Estimating physics-based sedans, and is an important predictor of levels and identify a more technology- curves better ensures that technology aerodynamic drag. Mass is an essential input in the neutral relationship between footprint and performance are held constant for tractive energy calculation.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00036 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43021

Chapter 6 of the PRIA shows the improve the fuel economy of new manufacturers occasionally use a resultant CAFE levels estimated for the vehicles, in both the near future and the technology in today’s new vehicle fleet vehicle classes ANL simulated for this timeframe of this proposed rulemaking, (like turbocharged engines). For a few analysis, at different footprint values to meet the fuel economy and CO2 technologies considered in the 2012 and by vehicle ‘‘box.’’ Pickups are standards proposed in this rulemaking. rulemaking, manufacturers began considered 1-box, hatchbacks and The analysis evaluated costs for these implementing the technologies but have minivans are 2-box, and sedans are 3- technologies, and looked at how these since largely pivoted to other box. These estimates are compared with costs may change over time. The technologies due to consumer the MY 2021 standards finalized in analysis also considered how fuel- acceptance issues (for instance, in some 2012. The general trend of the simulated saving technologies may be used on cases drivability and performance feel data points follows the pattern of the many types of vehicles (ranging from issues associated with dual clutch previous MY 2021 standards for all small cars to trucks) and how the transmissions without a torque technology packages and tractive energy technologies may perform in improving converter) or limited commercial effectiveness values presented in the fuel economy and CO2 emissions in success. The analysis utilizes new PRIA. The tractive energy curves are combination with other technologies. information as manufacturers’ use of intended to validate the curve shapes With cost and effectiveness estimates for technologies evolves. against a physics-based alternative, and technologies, the analysis can forecast Some of the emerging technologies the analysis suggests that the curve how manufacturers may respond to described in the Draft TAR were not shapes track the physical relationship potential standards and can estimate the included in this analysis, but this between fuel economy and tractive associated costs and benefits related to includes some additional technologies energy for different footprint values. technology and equipment changes. not previously considered. As industry Physical limitations are not the only This assists the assessment of invents and develops new fuel-savings forces manufacturers face; they must technological feasibility and is a technologies, and as suppliers and also produce vehicles that consumers building block for the consideration of manufacturers produce and apply the will purchase. For this reason, in setting economic practicability of potential technologies, and as consumers react to future standards, the analysis will standards. the new technologies, efforts are continue to consider information from NHTSA, EPA, and CARB issued the continued to learn more about the statistical analyses that do not Draft Technical Assessment Report capabilities and limitations of new homogenize technology applications in (Draft TAR) 106 as the first step in the technologies. While a technology is in addition to statistical analyses which EPA MTE process. The Draft TAR early development, theoretical do, as well as a tractive energy analysis provided an opportunity for the constructs, limited access to test data, similar to the one presented above. agencies to share with the public and CBI is relied on to assess the The relationship between fuel updated technical analysis relevant to technology. After manufacturers economy and footprint remains development of future standards. For commercialize the technology and bring directionally discernable but this NPRM, the analysis relies on products to market, the technology may quantitatively uncertain. Nevertheless, portions of the analysis presented in the be studied in more detail, which each standard must commit to only one Draft TAR, along with new information generally leads to the most reliable function. Approaching the question that has been gathered and developed information about the technology. In ‘‘how is fuel economy related to since conducting that analysis, and the addition, once in production, the footprint’’ from different directions and significant, substantive input that was technology is represented in the fuel applying different approaches will received during the public comment economy and CO2 status of the baseline provide the greatest confidence that the period. fleet. The technology analysis is kept as single function defining any given The Draft TAR considered many current as possible in light of the technologies previously assessed in the ongoing technology development and standard appropriately and reasonably 107 reflects the relationship between fuel 2012 final rule. In some cases, implementation in the automotive economy and footprint. Please provide manufacturers have nearly universally industry. adopted a technology in today’s new comments on this tentative conclusion Some technology assumptions have vehicle fleet (for example, electric and the above discussion. been updated since the MYs 2017–2025 power steering). In other cases, final rule and, in many cases, since the D. Characterization of Current and 2016 Draft TAR. In some cases, EPA and Anticipated Fuel-Saving Technologies 106 Available at https://www.nhtsa.gov/staticfiles/ rulemaking/pdf/cafe/Draft-TAR-Final.pdf (last NHTSA presented different analytical The analysis evaluates a wide array of accessed June 15, 2018). approaches in the Draft TAR; the technologies manufacturers could use to 107 77 FR 62624 (Oct. 15, 2012). analysis is now presented using the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00037 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.018 43022 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

same direct manufacturing costs, retail fuel-saving technologies to a retail price. Model or ALPHA), and applied cost costs, and learning rates. In addition, Today’s analysis utilizes the direct effective technologies to manufacturers’ the effectiveness of fuel-economy manufacturing costs (DMC) and the fleets in response to potential standards. technologies is now assessed based on retail price equivalent (RPE) Many stakeholders commented that the the same assumptions, and with the methodology published in the Draft OMEGA model oversimplified fleet- same tools. Finally, manufacturers’ TAR. wide analysis, resulting in significant response to stringency alternatives is Two tools to estimate effectiveness of shortcomings.114 For instance, OMEGA forecast with the same simulation fuel-saving technologies were used in assumed manufacturers would redesign the Draft TAR, and for today’s analysis, model. all vehicles in the fleet by 2021, and Since the 2017 and later final rule, only one tool was used (Autonomie).111 then again by 2025; stakeholders many cost assessments, including tear Previously, EPA developed ‘‘ALPHA’’, down studies, were funded and an in-house model that estimated fuel- purported that these assumptions did completed, and presented as part of the savings for technologies, which not reflect practical constraints in many 115 Draft TAR analysis. These studies provided a foundation for EPA’s manufacturers’ business models. evaluated transmissions, engines, analysis. EPA’s ‘‘ALPHA’’ results were Additionally, stakeholders commented hybrid technologies, and mass used to calibrate a much simpler that OMEGA did not adequately take reduction.108 As a result, the analysis ‘‘Lumped Parameter Model’’ that was into consideration common parts like uses updated cost estimates for many developed by EPA to estimate shared engines, shared transmissions, technologies, some of which have been technology effectiveness for many and engineering platforms. The CAFE updated since the Draft TAR. In technologies. The Lumped Parameter model does consider refresh and addition to those studies, the analysis Model (LPM) approximated simulation redesign cycles and parts sharing. The also leveraged research reports from modeling results instead of directly CAFE model can evaluate responses to other organizations to assess costs.109 using the results and lead to less any policy alternative on a year-by-year Today’s analysis also updates the costs accurate estimates of technology basis, as required by EPCA/EISA 116 and to 2016 dollars, as in many cases effectiveness. Many stakeholders as allowed by the CAA, and can also technology costs were estimated several questioned the efficacy of the Lumped account for manufacturers’ year-by-year years ago. Parameter Model and ALPHA plans that involve ‘‘carrying forward’’ The analysis uses an updated, peer- assumptions and outputs in technology from prior model years, and reviewed model developed by ANL for 112 combination, especially as the tool some other vehicles possibly applying the Department of Energy to provide a was used to evaluate increasingly ‘‘extra’’ technology in anticipation of more rigorous estimate for battery costs. heterogeneous combinations of standards in ensuing model years. For The new battery model provides an technologies in the baseline fleet.113 For estimate future for battery costs for today’s analysis, EPA and NHTSA used today’s analysis, an updated version of hybrids, plug-in hybrids, and electric an updated version of the Autonomie the CAFE model is used—an improved vehicles, taking into account the model—an improved version of what version of what NHTSA presented in different battery design characteristics NHTSA presented in the 2016 Draft the 2016 Draft TAR—to assess and taking into account the size of the TAR—to assess technology effectiveness manufacturers’ response to policy battery for different applications.110 of technologies and combinations of alternatives. See Section II.A.1 above for In the Draft TAR, two possible technologies. The Department of further discussion of the decision to use methodologies to estimate indirect costs Energy’s ANL developed Autonomie the CAFE model for the NPRM analysis. from direct manufacturing costs, and the underpinning model Each aforementioned change is described as ‘‘indirect cost multipliers’’ assumptions leveraged research from discussed briefly in the remainder of and ‘‘retail price equivalent’’ were the DOE’s Vehicle Technologies Office this section and in much greater detail presented. Both of these methodologies and feedback from the public. in Chapter 6 of the PRIA. A brief attempted to relate the price of parts for Autonomie is commercially available summary of the technologies considered and widely used; third parties such as in this proposal is discussed below. 108 FEV prepared several cost analysis studies for suppliers, automakers, and academic Please provide comments on all aspects EPA on subjects ranging from advanced 8-speed researchers (who publish findings in transmissions to belt alternator starter, or Start/Stop peer reviewed academic journals) of the analysis as discussed here and as systems. NHTSA also contracted with Electricore, detailed in the PRIA. EDAG, and Southwest Research on teardown commonly use the Autonomie studies evaluating mass reduction and simulation software. transmissions. The 2015 NAS report on fuel Similarly for today’s analysis, only economy technologies for light-duty vehicles also 114 The Alliance of Automobile Manufacturers evaluated the agencies’ technology costs developed one tool is used. Previously, EPA developed ‘‘OMEGA,’’ a tool that looked commented that ‘‘the OMEGA model is over- based on these teardown studies, and the optimized and unrealistic . . . many of these issues technology costs used in this proposal were at costs of technologies and either are not present or are accounted for in DOT’s updated accordingly. These studies are discussed in effectiveness of technologies (as detail in Chapter 6 of the PRIA accompanying this Volpe model. The Alliance therefore recommends proposal. estimated by EPA’s Lumped Parameter that EPA focus on ensuring needs specific to its 109 For example, the agencies relied on reports regulatory analysis are appropriately addressed in from the Department of Energy’s Office of Energy 111 ANL’s Full-Vehicle Simulation Autonomie the Volpe model.’’ Alliance at 48 (Docket ID. EPA– Efficiency & Renewable Energy’s Vehicle Model is discussed in Chapter 6 of the PRIA and HQ–OAR–2015–0827–9194). Technologies Office. More information on that in the ANL Model Documentation available at 115 For example, FCA provided the following office is available at https://www.energy.gov/eere/ Docket No. NHTSA–2018–0067. comments: ‘‘EPA’s expectation of 10–20% mass 112 vehicles/vehicle-technologies-office. Other agency At NHTSA–2016–0068–0082, p. 49, FCA reduction rates across 70% of FCA’s fleet, which reports that were relied on for technology or other provided the following comments, ‘‘FCA believes includes a 70% truck mix, is simply unreasonable information are referenced throughout this proposal EPA is overestimating the benefits of technology. As as the magnitude of change would require complete and accompanying PRIA. the LPM is calibrated to those projections, so too product redesigns in less than eight years 110 For instance, battery electric vehicles with is the LPM too optimistic.’’ FCA also shared the high levels of mass reduction may use a smaller chart, ‘‘LPM vs. Actual for 8 Speed Transmissions.’’ shortening existing production needed to amortize battery than a comparable vehicle with less mass 113 See e.g., Automotive News ‘‘CAFE math gets the large capital cost involved.’’ FCA at 19 (Docket reduction technology and still deliver the same trickier as industry innovates’’ (Kulisch), March 26, ID. EPA–HQ–OAR–2015–0827–6160). range on a charge. 2018. 116 49 U.S.C. 32902(b)(2)(B).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00038 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43023

1. Data Sources and Processes for costs (the retail price equivalent, or assess costs because the product is not Developing Individual Technology RPE) was taken into account. Finally, yet in the marketplace for evaluation. In Assumptions costs for technologies may change over these cases, third-party estimates and Technology assumptions were time as industry streamlines design and confidential information from suppliers developed that provide a foundation for manufacturing processes. Potential cost and manufacturers are relied upon; conducting a fleet-wide compliance improvements with learning effects (LE) however, there are some common analysis. As part of this effort, the were also considered. The retail cost of pitfalls with relying on confidential analysis estimated technology costs, equipment in any future year is business information to estimate costs. projected technology effectiveness estimated to be equal to the product of The agencies and the source may have had incongruent or incompatible values, and identified possible the DMC, RPE, and LE. Considering the definitions of ‘‘baseline.’’ 119 The source limitations for some fuel-saving retail cost of equipment, instead of may have provided direct manufacturer technologies. There is a preference to merely direct manufacturing costs, is costs at a date many years in the future, use values developed from careful important to account for the real-world and assumed very high production review of commercialized technologies; price effects of a technology, as well as volumes, important caveats to consider however, in some cases for technologies market realities. Absent government for agency analysis. In addition, a that are new, and are not yet for sale in mandate, a manufacturer will not source, under no contractual obligation any vehicle, the analysis relied on undertake expensive development and to the agencies, may provide incomplete information from other sources, support costs to implement technologies and/or misleading information. In other including CBI and third-party research without realistic prospects of consumer cases, intellectual property reports and publications. Many willingness to pay enough for such considerations and strategic business emerging technologies are still being technology to allow for the partnerships may have contributed to a evaluated for the analysis supporting manufacturer to recover its investment. manufacturer’s cost information and the final rule, including those that are (1) Direct Manufacturing Costs could be difficult to account for in the currently emerging. In many instances, the analysis used model as not all manufacturer’s may For today’s analysis, one set of cost agency-sponsored tear-down studies of have access to proprietary technologies assumptions, one set of effectiveness vehicles and parts to estimate the direct at stated costs. New information is values (developed with one tool), and manufacturing costs of individual carefully evaluated in light of these one set of assumptions about the technologies. In the simplest cases, the common pitfalls, especially regarding limitations of some technologies are studies produced results that confirmed emerging technologies. The analysis presented. Many sources of data were third-party industry estimates, and used third-party, forward looking evaluated, in addition to many aligned with confidential information information for advanced cylinder stakeholder comments received on the provided by manufacturers and deactivation and variable compression Draft TAR. Throughout the process of suppliers. In cases with a large ratio engines, and while these cost developing the assumptions for today’s difference between the tear-down study estimates may be cursory (as is the case analysis, the preferred approach was to results and credible independent with many emerging technologies prior harmonize on sources and sources, study assumptions were to commercialization), the agencies took methodologies that were data-driven scrutinized, and sometimes the analysis care to use early information provided and reproducible in independent was revised or updated accordingly.117 fairly and reasonably. While costs for verification, produced using tools Studies were conducted on vehicles and fuel-saving technologies reflect the best utilized by OEMs, suppliers, and technologies that would cover a breadth estimates available today, technology academic institutions, and using tools of fuel-savings technologies, but because cost estimates will likely change in the that could support both CAFE and CO 2 tear-down studies can be time-intensive future as technologies are deployed and analysis. A single set of assumptions and expensive, the agencies did not as production is expanded. For also facilitates and focuses public sponsor teardown studies for every emerging technologies, the best comment by reducing burden on technology. For some technologies, information available at the time of the stakeholders who seek to review all of independent tear-down studies were analysis was utilized, and cost the supporting documentation for this also utilized, in addition to other assumptions will continue to be proposal. publications and confidential business updated. (a) Technology Costs information.118 Due to the variety of (2) Indirect Costs The analysis estimated present and technologies and their applications, a detailed tear-down study could not be As explained above, in addition to future costs for fuel-saving technologies, direct manufacturing costs, the analysis taking into consideration the type of conducted for every technology, including pre-production technologies. estimates and considers indirect vehicle, or type of engine if technology manufacturing costs. To estimate costs vary by application. Cost estimates Many fuel-saving technologies were considered that are pre-production, or indirect costs, direct manufacturing were developed based on three main costs are multiplied by a factor to inputs. First, direct manufacturing costs sold in very small pilot volumes. For emerging technologies that could be represent the average price for fuel- (DMC), or the component costs of the saving technologies at retail. This factor, physical parts and systems, were applied in the rulemaking timeframe, a tear-down study cannot be conducted to referred to as the retail price considered, with estimated costs equivalence (RPE), accounts for indirect assuming high volume production. 117 For instance, in previous analysis, EPA costs like engineering, sales, and DMCs generally do not include the referenced an old study that purported the first 7– administrative support, as well as other indirect costs of tools, capital 10% of mass reduction to be ‘‘free’’ or at a overhead costs, business expenses, significant ‘‘cost savings’’ to for many vehicles and equipment, and financing costs, nor do warranty costs, and return on capital they cover indirect costs like many manufacturers. 118 The analysis referenced studies from private engineering, sales, and administrative businesses and business analysts for emerging 119 ‘‘Baseline’’ here refers to a reference part, support. Second, indirect costs via a technologies and for off-the-shelf technologies that piece of equipment, or engineering system that scalar markup of direct manufacturing were commercially mature. efficiency improvements and costs are relative to.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00039 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43024 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

considerations. This approach to the industry and their uncertain effect on responds dynamically to volume, and RPE remains unchanged from the RPE capital amortization, and given the volume responds dynamically to approach NHTSA presented in the Draft difficulty of handling this uncertainty in learning, an internally consistent model TAR. the CAFE model, this analysis does not solution would likely require iteration The RPE was chosen for this analysis account for stranded capital. However, of the CAFE model to seek a stable instead of indirect cost multipliers these trends will be monitored to assess solution within the model’s (ICM) because it provides the best the role of stranded capital moving representation multiyear planning. Thus estimate of indirect costs. For a more forward. far, these challenges suggest it would be detailed discussion of the approach to The analysis continues to rely on infeasible to calculate degrees of indirect costs, see PRIA Chapter 9. projected refresh and redesign cycles in volume-based learning in a manner that the CAFE model to moderate the (3) Stranded Capital Costs responds dynamically to modeled cadence for technology adoption and technology application. Nevertheless, Past analyses accounted for costs limit the occurrence of stranded capital the agencies invite comment on the associated with stranded capital when and the need to account for it. Stranded issue, and seek data and methods that fuel economy standards caused a capital is an important consideration to would provide the basis for a technology to be replaced before its appropriately account for costs if there practicable approach to doing so. costs were fully amortized. The idea is too rapid of a turnover for certain Today’s analysis also updates the way behind stranded capital is that technologies. learning effects apply to costs. In the manufacturers amortize research, Draft TAR analysis, NHTSA applied (4) Cost Learning development, and tooling expenses over learning curves only to the incremental many years, especially for engines and Manufacturers make improvements to direct manufacturing costs or costs over transmissions. The traditional production processes over time, often the previous technology on the tech production life-cycles for transmissions resulting in lower costs. Today’s tree. In practice, two things were and engines have been a decade or analysis estimates cost learning by observed: (1) If the incremental direct longer. If a manufacturer launches or considering Wright’s learning theory, manufacturing costs were positive, updates a product with fuel-saving which states that as every time technologies could not become less technology, and then later replaces that cumulative volume for a product expensive than their predecessors on technology with an unrelated or doubles, the cost lowers by a scalar the tech tree, and (2) absolute costs over different fuel-saving technology before factor. The analysis accounts for baseline technology depended on the the equipment and research and learning effects with model year-based learning curves of root technologies on development investments have been cost learning forecasts for each the tech tree. Today’s analysis applies fully paid off, there will be unrecouped, technology that reduce direct learning effects to the incremental cost or stranded, capital costs. Quantifying manufacturing costs over time. over the null technology state on the stranded capital costs attempted to Historical use of technologies were tech tree. After this step, the analysis account for such lost investments. In the evaluated, and industry forecasts were calculates year-by-year incremental Draft TAR analysis, there were only a reviewed to estimate future volumes for costs over preceding technologies on the few technologies for a few the purpose of developing the model tech tree to create the CAFE model manufacturers that were projected to year-based technology cost learning inputs. have stranded capital costs. curves. The CAFE model does not Direct manufacturing costs and As more technologies are included in dynamically update learning curves, learning effects for many technologies this analysis, and as the CAFE model based on compliance pathways chosen were reviewed by evaluating historical has been expanded to account for in simulation. use of technologies and industry platform and engine sharing and As discussed above, cost inputs to the forecasts to estimate future volumes. updated with redesign and refresh CAFE model incorporate estimates of This approach produced reasonable cycles, accounting for stranded capital volume-based learning. As an estimates for technologies already in has become increasingly complex. alternative approach, Volpe Center staff production. For technologies not yet in Separately, the fact that manufacturers have considered modifications such that production in MY 2016, the cumulative may be shifting their investment the CAFE model would calculate volume in MY 2016 is zero, because strategies in ways that may affect degrees of volume-based learning manufacturers have not yet produced stranded capital calculations was dynamically, responding to the model’s the technologies. For pre-production considered. For instance, Ford and application of affected technologies. cost estimates, the analysis often relies General Motors agreed to jointly While it is intuitive that the degree of on confidential business information develop next generation transmission cost reduction achieved through sources to predict future costs. Many technologies,120 and some suppliers sell experience producing a given sources for pre-production cost similar transmissions to multiple technology should depend on the actual estimates include significant learning manufacturers. These arrangements accumulated experience (i.e., volume) effects, often providing cost estimates allow manufacturers to share in capital producing that technology, staff have assuming high volume production, and expenditures, or amortize expenses thus far found such dynamic often for a timeframe late in the first more quickly. Manufacturers implementation in the CAFE model production generation or early in the increasingly share parts on vehicles infeasible. Insufficient data has been second generation of the technology. around the globe, achieving greater scale available regarding manufacturers’ Rapid doubling and re-doubling of a low and greatly affecting tooling strategies historical application of specific cumulative volume base with Wright’s and costs. Given these trends in the technology. Also, insofar as underlying learning curves can provide unrealistic direct manufacturing costs already make cost estimates. In addition, direct 120 See, e.g., Nick Bunkley, Ford to invest $1.4 some assumptions about volume and manufacturing cost projections can vary billion to build 10-speed transmissions for 2017 F– scale, insufficient information is depending on the initial production 150, Automotive News (Apr. 26, 2016), http:// www.autonews.com/article/20160426/OEM01/ currently available to determine how to volume assumed. Direct costs with 160429878/ford-to-invest-$1.4-billion-to-build-10- dynamically adjust these underlying learning were carefully examined, and speed-transmissions-for-2017. costs. It should be noted that if learning adjustments were made to the starting

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00040 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43025

point for those technologies on the improvements of technologies. First, stated as model inputs and are available learning curve to better align with the technology effectiveness often differs for review and comment. For today’s assumptions used for the initial direct significantly depending on the type of analysis, every effort was made to make cost estimate. See PRIA Chapter 9 for vehicle and the other technologies that the input specifications and modeling more detailed information on cost are on the vehicle, and this is shown in assumptions available for review and learning. full-vehicle simulations. Different comment. PRIA Chapter 6 and technologies may provide different fuel referenced documents provide more (b) Technology Effectiveness economy improvements depending on detailed information. (1) Technology Effectiveness Simulation whether they are implemented alone or Technology development and Modeling in tandem with other technologies. application will be monitored to acquire Full-vehicle simulation modeling was Single point estimates often more information for the final rule. The used to estimate the fuel economy oversimplify these important, complex agencies may update the analysis for the improvements manufacturers could relationships and lead to less accurate final rule based on comments and/or make to their fleet by adding new effectiveness estimates. Also, because new information that becomes available. technologies, taking into account MY manufacturers often implement a Today’s analysis utilizes effectiveness 2016 vehicle specifications, as well as number of fuel-saving technologies estimates for technologies developed 121 how combinations of technologies simultaneously at vehicle redesigns, it is using Autonomie software, a physics- interact. Full-vehicle simulation generally difficult to isolate the effect of based full-vehicle simulation tool modeling uses computer software and individual technologies using laboratory developed and maintained by the physics-based models to predict how measurement of production vehicles Department of Energy’s ANL. combinations of technologies perform alone. Simulation modeling offers the Autonomie has a long history of together. opportunity to isolate the effects of development and widespread The simulation and modeling requires individual technologies by using a application by users in industry, detailed specifications for each single or small number of baseline academia, research institutions and 122 technology that describes its efficiency configurations and incrementally government. Real-world use has and performance-related characteristics. adding technologies to those baseline contributed significantly to aspects of Those specifications generally come configurations. This provides a Autonomie important to producing from design specifications, laboratory consistent reference point for the realistic estimates of fuel economy and measurements, simulation or modeling, incremental effectiveness estimates for CO2 emission rates, such as estimation and may involve additional analysis. each technology and for combinations of and consideration of performance, For example, the analysis used engine technologies for each vehicle type and utility, and driveability metrics (e.g., maps showing fuel use vs. engine torque reduces potential double counting or towing capability, shift business, vs. engine speed, and transmission undercounting technology effectiveness. frequency of engine on/off transitions). maps taking into account gear efficiency Note: It is most important that the This steadily increasing realism has, in for a range of loads and speeds. With incremental effectiveness of each turn, steadily increased confidence in physics-based technology specifications, technology and combinations be the appropriateness of using Autonomie full-vehicle simulation modeling can be accurate and relative to a consistent to make significant investment used to estimate technology baseline, because it is the incremental decisions. Notably, DOE uses effectiveness for various combinations effectiveness that is applied to each Autonomie for analysis supporting and permutations of technologies for vehicle model/configuration in the MY budget priorities and plans for programs many vehicle classes. To develop the 2016 baseline fleet (and to each vehicle managed by its Vehicle Technologies specifications used for the simulation model/configuration’s absolute fuel Office (VTO) and to decide among and modeling, laboratory test data was economy value) to determine the competing vehicle technology R&D evaluated for production and pre- absolute fuel economy of the model/ projects. production technologies, technical configuration with the additional In the 2015 National Academies of publications, manufacturer and supplier technology. The absolute fuel economy Science (NAS) study of fuel economy CBI, and simulation modeling of values of the simulation modeling runs improving technologies, the Committee specific technologies. Evaluating by themselves are used only to recommended that the agencies use full- recently introduced production determine the incremental effectiveness vehicle simulation to improve the products to inform the technology and are never used directly to assign an analysis method of estimating 123 effectiveness models of emerging absolute fuel economy value to any technology effectiveness. The technologies is preferred because doing vehicle model/configuration for the committee acknowledged that so allows for a more reliable analysis of rulemaking analysis. Therefore, developing and executing tens or incremental improvements over commenters on technology effectiveness hundreds of thousands of constantly previous technologies; however, some should be specific about the incremental changing vehicle packages models in technologies were considered that are effectiveness of technologies relative to 121 not yet in production. As technologies other specifically defined technologies. More information about Autonomie is The fuel economy of a specific vehicle available at https://www.anl.gov/technology/ evolve and new applications emerge, project/autonomie-automotive-system-design (last this work will be continued and may or simulation modeling run in isolation accessed June 21, 2018). include additional technologies and/or may be less useful. 122 ANL Model Documentation, ‘‘A Detailed updated modeling for the final rule. The Second, full-vehicle simulation Vehicle Simulation Process To Support CAFE modeling requires explicit Standards’’ ANL/ESD–18/6. details of new and emerging 123 National Research Council. 2015. Cost, technologies are discussed in PRIA specifications and assumptions for each Effectiveness, and Deployment of Fuel Economy Chapter 6. technology; therefore, these Technologies for Light-Duty Vehicles. Washington, Using full-vehicle simulation assumptions can be presented for public DC: The National Academies Press [hereinafter review and comment. For instance, ‘‘2015 NAS Report’’] at pg. 263, available at https:// modeling has two primary advantages www.nap.edu/catalog/21744/cost-effectiveness- over using single or limited point transmission gear efficiencies, shift and-deployment-of-fuel-economy-technologies-for- estimates for fuel efficiency logic, and gear ratios are explicitly light-duty-vehicles (last accessed June 21, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00041 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43026 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

real-time is extremely challenging. simulations were run. In addition, Additional transmission technologies While initially this approach was not simulation modeling was conducted to and more levels of aerodynamic considered practical to implement, a determine the appropriate amount of technologies than NHTSA presented in process developed by Argonne in engine downsizing needed to maintain the Draft TAR analysis were also added collaboration with NHTSA and the DOT baseline performance across all modeled for today’s analysis. Having additional Volpe Center has succeeded in enabling vehicle performance metrics when technologies allowed the agencies to large scale simulation modeling. For advanced mass reduction technology or assign baselines and estimate fuel- more details about the Autonomie advanced engine technology was savings opportunities with more simulation model and its submodels applied, so these simulations take into precision. and inputs, see PRIA Chapter 6.2. account performance neutrality, given The 10 vehicle types (referred to as Today’s analysis modeled more than logical engine down-sizing ‘‘technology classes’’ in the modeling 50 fuel economy-improving opportunities associated with specific documentation) are shown in Table II– technologies, and combinations thereof, technologies. 7. Each vehicle type (technology class) on 10 vehicle types (an increase from Some baseline vehicle assumptions represented a large segment of vehicles, five vehicle types in NHTSA’s Draft used in the simulation modeling were such as medium cars, small SUVs, and TAR analysis). While 10 vehicle types updated based on public comment and medium performance SUVs.125 Baseline may seem like a small number, a large the assessment of the MY 2016 parameters were defined with ANL for portion of the production volume in the production fleet. The analysis included each technology class, including MY 2016 fleet have specifications that updated assumptions about curb weight, baseline curb weight, time required to are very similar, especially in highly component inertia, as well as accelerate from stop to 60 miles per competitive segments (for instance, technology properties like baseline hour, time required to accelerate from many mid-sized sedans, many small rolling resistance, aerodynamic drag 50 miles per hour to 80 miles per hour, SUVs, and many large SUVs coalesce coefficients, and frontal areas. Many of ability of the vehicle to maintain around similar specifications, the assumptions are aligned with constant 65 miles per hour speed on a respectively), and baseline simulations published research from the Department six percent upgrade, and (for some have been aligned around these modal of Energy’s Vehicle Technologies Office classes) assumptions about towing specifications. The sequential addition and other independent sources.124 capability. of these technologies generated more than 100,000 unique technology 124 Pannone, G. ‘‘Technical Analysis of Vehicle Load Reduction Potential for Advanced Clear Cars,’’ 125 Separate technology classes were created for combinations per vehicle class. The April 29, 2015. Available at https://www.arb.ca.gov/ high performance and low performance vehicles to analysis included 10 technology classes, research/apr/past/13-313.pdf (last accessed June 21, better account for performance diversity across the so more than one million full-vehicle 2018). fleet.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00042 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43027

From these baseline specifications, engines to the extent that performance engine. Manufacturers have repeatedly incremental combinations of fuel saving was maintained for the least capable told the agencies that the high costs for technologies were applied. As the performance criteria to maintain vehicle redesign and the increased combinations of technologies change, so utility for that criteria; therefore, manufacturing complexity that would too may predicted performance. sometimes other performance attributes result from resizing engines for such The analysis attempts to maintain may improve. For instance, the amount small changes in the vehicle preclude performance by resizing engines at a few of engine resizing may be determined doing so. The analysis should not, in specific incremental technology steps. based on its high speed acceleration fact, include engine resizing with the Steps from one technology to another time if it is the least capable criteria, but application of every technology or for typically associated with a major that resizing may also improve the low combinations of technologies that drive vehicle redesign, or engine redesign, speed acceleration time.128 The analysis small performance changes so that the were identified, and engine resizing was did not re-size the engine in response to analysis better reflects what is feasible restricted only to these steps. The adding technologies that have small for manufacturers to do.129 analysis allowed engine resizing when effects on vehicle performance. For 2. CAFE model mass reduction of 10% or greater was instance, if a vehicle’s weight is reduced applied to the vehicle glider mass,126 by a small amount causing the 0–60 The CAFE model is designed to and when one powertrain architecture mile per hour time to improve slightly, simulate compliance with a given set of was replaced with another the analysis would not resize the CAFE or CO2 standards for each architecture.127 The analysis resized manufacturer that sells vehicles in the belt-integrated starter generators, and other basic United States. The model begins with a 126 The vehicle glider is defined here as the technologies. However, switching from a naturally vehicle without the engine, transmission, and aspirated engine to a turbo-downsized engine is an 129 For instance, a vehicle would not get a driveline. See PRIA Chapter 6.3 for further engine architecture change typically associated modestly bigger engine if the vehicle comes with information. with a major redesign and radical change in engine floor mats, nor would the vehicle get a modestly 127 Some engine and accessory technologies may displacement. smaller engine without floor mats. This example be added to an engine without an engine 128 The simulation database, or summary of demonstrates small levels of mass reduction. If architecture change. For instance, manufacturers simulation outputs, includes all of the estimated manufacturers resized engines for small changes, may adapt, but not replace engine architectures to performance metrics for each combination of manufacturers would have dramatically more part include cylinder deactivation, variable valve lift, technology as modeled. complexity, potentially losing economies of scale.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00043 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.019 43028 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

representation of the MY 2016 vehicle • Technology costs too low or too matched to historical values for each model offerings for each manufacturer high. Tear down cost studies, CBI, manufacturer. that includes the specific engines and literature, and the 2015 NAS study • Customer acceptance under transmissions on each model variant, information were referenced to estimate estimated or over estimated. Resale observed sales volumes, and all fuel technology costs. In cases that one prices for hybrid vehicles, electric economy improving technology that is technology appeared exemplary on cost vehicles, and internal combustion already present on those vehicles. From and effectiveness relative to all other engine vehicles were evaluated to assess there the model adds technology, in technologies, information was acquired consumer willingness to pay for those response to the standards being from additional sources to confirm or technologies. The analysis accounts for considered, in a way that minimizes the reject assumptions. Cost assumptions the differential in the cost for those cost of compliance and reflects many for emerging technologies are technologies and the amount consumers real-world constraints faced by continuously being evaluated. have actually paid for those automobile manufacturers. The model • Technology effectiveness too high technologies. Separately, new dual- addresses fleet year-by-year compliance, or too low in combination with other clutch transmissions and manual taking into consideration vehicle refresh vehicle technologies. Technology transmissions were applied to vehicles and redesign schedules and shared effectiveness was evaluated using the already equipped with these platforms, engines, and transmissions Autonomie full-vehicle simulation transmission architectures. among vehicles. modeling, taking into account the Please provide comments on all As a result of simulating compliance, impact of other technologies on the assumptions for fuel economy and CO2 the CAFE model provides the vehicle and the vehicle type. Inputs and technology costs, effectiveness, technology pathways that manufacturers modeling for the analysis took into availability, and applicability to could use to comply with regulations, account laboratory test data for vehicles in the fleet. including how technologies could be production and some pre-production The technology effectiveness applied to each of their vehicle model/ technologies, technical publications, modeling results show effectiveness of a configurations in response to a given set manufacturer and supplier CBI, and technology often varies with the type of of standards. The model calculates the simulation modeling of specific vehicle and the other technologies that impacts of the simulated standard: technologies. Evaluating recently are on the vehicle. Figure II–1 and Technology costs, fuel savings (both in introduced production products to Figure II–2 show the range of gallons and dollars), CO2 reductions, inform the technology effectiveness effectiveness for each technology for the social costs and benefits, and safety models of emerging technologies was range of vehicle types and technology impacts. preferred; however, some technologies combinations included in this NPRM The current analysis reflects several that are not yet in production were analysis. The data reflect the change in changes made to the CAFE model since considered, via CBI. Simulation effectiveness for applying each 2012, when NHTSA used the model to modeling used carefully chosen baseline technology by itself while all other estimate the effects, costs, and benefits configurations to provide a consistent, technologies are held unchanged. The of final CAFE standards for light-duty reasonable reference point for the data show the improvement in fuel vehicles produced during MYs 2017– incremental effectiveness estimates. consumption (in gallons per mile) and 2021 and augural standards for MYs • Vehicle performance not considered tailpipe CO2 over the combined 2-cycle 2022–2025. The changes are discussed or applied in an infeasible manner. test procedures. For many technologies, in Section II.A.1, above, and PRIA Performance criteria, including low effectiveness values ranged widely; only Chapter 6. speed acceleration (0–60 mph time), a few technologies for which high speed acceleration (50–80 mph effectiveness may be reasonably 3. Assumptions About Individual time), towing, and gradeability (six represented as a fixed offset were Technology Cost and Effectiveness percent grade at 65 mph) were also identified. Values considered. In the simulation modeling, For engine technologies, the Cost and effectiveness values were resizing was applied to achieve the effectiveness improvement range is estimated for each technology included same performance level as the baseline relative to a comparably equipped in the analysis, with a summary list of for the least capable performance vehicle with only variable valve timing all technologies provided in Table II–1 criteria but only with significant design (VVT) on the engine. For automatic (List of Technologies with Data Sources changes. The analysis struck a balance transmission technologies, the for Technology Assignments) of by employing a frequency of engine effectiveness improvement range is over Preamble Chapter II.B, above. In all, downsizing that took product a 5-speed automatic transmission. For more than 50 technologies were complexity and economies of scale into manual transmission technologies, the considered in today’s analysis, and the account. effectiveness improvement range is over analysis evaluated many combinations • Availability of technologies for a 5-speed manual transmission. For road of these technologies on many production application too soon or too load technologies like aerodynamics, applications. Potential issues in late. A number of technologies were rolling resistance, and mass reduction, assessing technology effectiveness and evaluated that are not yet in production. the effectiveness improvement ranges cost were identified, including: CBI was gathered on the maturity and are relative to the least advanced • Baseline (MY 2016) vehicle timing of these technologies and the technology state, respectively. For technology level assessed as too low, or likely cadence at which manufacturers hybrid and electric drive systems that too high. Compliance information was might adopt these technologies. wholly replace an engine and extensively reviewed and supplemented • Product complexity and design transmission, the effectiveness with available literature on many MY cadence constraints too low or too high. improvement ranges are relative to a 2016 vehicle models. Manufacturers Product platforms, refresh and redesign comparably equipped vehicle with a could also review the baseline cycles, shared engines, and shared basic engine with VVT only and a 5- technology assignments for their transmissions were also considered in speed automatic transmission. For vehicles, and the analysis incorporates the analysis. Product complexity and hybrid or electrification technologies feedback received from manufacturers. the cadence of product launches were that complement other advanced engine

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00044 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43029

and transmission technologies, the instance, parallel strong hybrids and range of estimated effectiveness values. effectiveness improvement ranges are belt integrated starter generators retain Figure II–3 below shows a hierarchy of relative to a comparably equipped engine technologies, such as a turbo technologies discussed. vehicle without the hybrid or charged engine or an Atkinson cycle BILLING CODE 4910–59–P electrification technologies (for engine). Many technologies have a wide

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00045 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.020 43030 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00046 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.021 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43031

4. Engine Technologies and ‘‘advanced engine technologies’’ are an entirely new engine architecture. In used only to define how the CAFE the CAFE model, basic engine There are a number of engine model applies a specific engine technologies may be applied in technologies that manufacturers can use technology and handles incremental combination with other basic engine to improve fuel economy and CO2. Some engine technologies can be costs and effectiveness improvements. technologies; advanced engine incorporated into existing engines with ‘‘Basic engine technologies’’ refer to technologies (defined by an engine map) minor or moderate changes to the technologies that, in many cases, can be stand alone as an exclusive engine engines, but many engine technologies adapted to an existing engine with technology. The words ‘‘basic’’ and require an entirely new engine minor or moderate changes to the ‘‘advanced’’ are not meant to confer any architecture. engine. ‘‘Advanced engine information about the level of In this section and for this analysis, technologies’’ refer to technologies that sophistication of the technology. Also, the terms ‘‘basic engine technologies’’ generally require significant changes or many advanced engine technology

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00047 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.022 43032 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

definitions include some basic engine the incremental engine costs and engine, and the agencies evaluated technologies, but these basic effectiveness values often depend on many production engines to better technologies are already accounted for engine architecture. The agencies understand how costs and capabilities in the costs and effectiveness values of modeled single overhead cam (SOHC), may vary with engine configuration. the advance engine. The ‘‘basic engine dual overhead cam (DOHC), and Table II–8, Table II–9, Table II–10 below technologies’’ need not be (and are not) overhead valve (OHV) engines shows the summary of absolute costs 130 applied in addition to the ‘‘advanced separately to account for differences in for different technologies. engine technologies’’ in the CAFE engine architecture. The agencies model. adjusted costs for some engine 130 ‘‘Absolute’’ being in reference to cost above the lowest level of technology considered in Engines come in a wide variety of technologies based on the number of simulations. For instance, an engine of the same shapes, sizes, and configurations, and cylinders and number of banks in the architecture with no VVT, VVL, SGDI, or DEAC.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00048 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43033

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00049 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.023 43034 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00050 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.024 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43035

BILLING CODE 4910–59–C Many types of production powertrains analysis, and engine maps were were reviewed and tested for this developed for each combination of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00051 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.025 43036 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

engine technologies. For a given engine VVT is nearly universally used in the turbo technologies were modeled and configuration, some production engines MY 2016 fleet. considered separately in today’s may be less efficient than the engine analysis. (2) Variable Valve Lift (VVL) maps presented in the analysis, and (b) ‘‘Advanced’’ Engine Technologies some may be more efficient. Developing Variable Valve Lift (VVL) dynamically engine maps that reasonably adjusts the travel of the valves to The analysis included ‘‘advanced’’ represented most vehicles equipped optimize airflow over a broad range of engine technologies that can deliver with the engine technology, and that are engine operating conditions. The high levels of effectiveness but often in production today, was the preferred technology increases effectiveness by require a significant engine design approach for this analysis. Additionally, reducing pumping losses and may change or a new engine architecture. In some advanced engines were included improve efficiency by affecting in- the CAFE model, ‘‘basic’’ engine in the simulation that are not yet in cylinder charge (fuel and air mixture), technologies may be considered in production. The engine maps for these motion, and combustion. combination and applied before advanced engine technologies. engines were either based on CBI or (3) Stoichiometric Gasoline Direct ‘‘Advanced’’ engine technologies were theoretical. The most recently Injection (SGDI) released production engines are still generally include one or more basic being reviewed, and the analysis may Stoichiometric Gasoline Direct engine technologies in the simulation, include updated engine maps in the Injection (SGDI) sprays fuel at high without the need to layer on ‘‘basic’’ future or add entirely new engine maps pressure directly into the combustion engine technologies on top of to the analysis if either action could chamber, which provides cooling of the ‘‘advanced’’ engines. Once an advanced improve the quality of the fleet-wide in-cylinder charge via in-cylinder fuel engine technology is applied, the model analysis. vaporization to improve spark knock does not reconsider the basic engine Stakeholders provided many tolerance and enable an increase in technologies. The characterization of comments on the engine maps that were compression ratio and/or more optimal each advanced engine technology takes presented in the Draft TAR. These spark timing for improved efficiency. into account the prerequisite comments were considered, and today’s SGDI appears in about half of basic technologies. analysis utilizes several engine maps engines produced in MY 2016, and the Many of the newest advanced engine that were updated since the Draft TAR. technology is used in many advanced technologies improve effectiveness over Most notably, for turbocharged and engines as well. their predecessors, but the engines may also include sophisticated materials or downsized engines, the engine maps (4) Basic Cylinder Deactivation (DEAC) were adjusted in high torque, low speed manufacturing processes that contribute operating conditions to address engine Basic Cylinder Deactivation (DEAC) to efficiency improvements. For knock with regular octane fuel to align disables intake and exhaust valves and instance, one recently introduced turbo with the fuel octane that manufacturers prevents fuel injection into some charged engine uses sodium filled valve recommend be used for the majority of cylinders during light-load operation. stems.131 Another recently introduced vehicles. In the Draft TAR, NHTSA The engine runs temporarily as though high compression ratio engine uses a assumed high octane fuel to develop it were a smaller engine, which reduces sophisticated laser cladding process to engine maps. See the discussion below pumping losses and improves manufacture valve seats and improve and in PRIA Chapter 6.3 for more efficiency. Manufacturers typically airflow.132 To fully consider these details. Please provide comment on the disable one-cylinder bank with basic advancements (and their potential appropriateness of assuming the use of cylinder deactivation. In the MY 2016 benefits), the incremental costs of these lower octane fuels. fleet, manufacturers used DEAC on V6, technologies, as well as the effectiveness V8, V10, and V12 engines on OHV, improvements, must be accounted for. (a) ‘‘Basic’’ Engine Technologies SOHC, and DOHC engine (1) Turbocharged Engines The four ‘‘basic’’ engine technologies configurations. With some engine in today’s model are Variable Valve configurations in some operating Turbo engines recover energy from Timing (VVT), Variable Valve Lift conditions, DEAC creates noise- hot exhaust gas and compress intake air, (VVL), Stoichiometric Gasoline Direct vibration-and-harshness (NVH) thereby increasing available airflow and Injection (SGDI), and basic Cylinder challenges. NVH challenges are increasing specific power level. Due to Deactivation (DEAC). Over the last significant for V6 and I4 DEAC specific power improvements on turbo decade, manufacturers upgraded many configurations. For I4 engine engines, engine displacement can be engines with these engine technologies. configurations, manufacturers can downsized. The downsizing reduces Implementing these technologies operate the DEAC function of an engine pumping losses and improves fuel involves changes to the cylinder head of in very few operating conditions, with economy at lower loads. For the NPRM the engine, but the engine block, limited potential to save fuel. No analysis, a level of downsizing is crankshaft, pistons, and connecting rods manufacturers sold I4 DEAC engines in assumed to be applied that achieves require few, if any, changes. In today’s the MY 2016 fleet. Typically, the performance similar to the baseline analysis, manufacturers may apply the smaller the engine displacement, the naturally-aspirated engine. This four basic engine technologies in less opportunity DEAC provides to assumes manufacturers would apply the various combinations, just as improve fuel consumption. benefits toward improved fuel economy manufacturers have done recently. Manufacturers and suppliers continue to evaluate more improved versions of 131 See Honda, ‘‘2018 Honda Accord Press Kit— (1) Variable Valve Timing (VVT) Powertrain,’’ Oct. 2, 2017. Available at http:// cylinder deactivation, including news.honda.com/newsandviews/article.aspx?g= Variable Valve Timing (VVT) is a advanced cylinder deactivation and honda-automobiles&id=9932-en. (last accessed June family of valve-train designs that pairing basic cylinder deactivation with 21, 2018). dynamically adjusts the timing of the turbo charged engines. No 132 Hakariya et al., ‘‘The New Toyota Inline 4- Cylinder 2.5L Gasoline Engine,’’ SAE Technical intake valves, exhaust valves, or both, in manufacturers produced such Paper 2017–01–1021 (Mar. 28, 2017), available at relation to piston position. This family technologies in the MY 2016 fleet. https://www.sae.org/publications/technical-papers/ of technologies reduces pumping losses. Advanced cylinder deactivation and content/2017-01-1021/.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00052 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43037

and not trade off fuel economy with regular octane fuel. Using the performance and greater engine improvements to increase overall updated criteria assures the NPRM downsizing for lower pumping losses. vehicle performance. In practice, analysis reflects real-world constraints CEGR1 technology is used in only a few manufacturers have often also improved faced by manufacturers to assure engine vehicles in the MY 2016 fleet, and many some vehicle performance attributes at durability, and acceptable drivability, of these vehicles include high- the expense of not maximizing potential noise and harshness, and addresses the performance utility either for towing or fuel economy improvements. over-estimation of potential fuel acceleration. Manufacturers may develop engines economy improvements related to the 133 (a) Turbocharged Engine Technologies to operate on varying levels of boost, fuel octane assumptions, which did not Not Considered with higher levels of boost achieving fully account for these constraints, in higher engine specific power and the Draft TAR. Compared with the Previous analyses considered turbo enabling greater levels of engine NHTSA analysis in the Draft TAR, these charged engines with even higher BMEP downsizing and corresponding engine maps adjust the fuel use at high than today’s Turbo2 and CEGR1 reductions in pumping losses for torque and low speed operation and at technologies, but today’s analysis does improved efficiency. However, engines high speed operation to fully account not present 27 bar BMEP turbo engines. operating at higher boost levels are for knock limitations with regular Turbo engines with very high BMEP generally more susceptible to engine octane fuel. have demonstrated limited potential to knock,134 especially at higher torques The analysis assumes engine improve fuel economy due to practical and low engine speeds. Additionally, downsizing with the addition of turbo limitations on engine downsizing and engines with higher boost levels technology. For instance, in the tradeoffs with launch performance and typically require larger induction and simulations, manufacturers may have drivability. Based on the analysis, and exhaust system components, dissipate replaced a naturally-aspirated V8 engine based on CBI, CEGR2 turbo engine greater amounts of heat, and with with a turbo V6 engine, and technology was not included in this greater levels of engine downsizing have manufacturers may have replaced a NPRM analysis. increased challenges with turbo lag.135 naturally-aspirated V6 engine with a (2) High Compression Ratio Engines For these reasons, three levels of turbo turbo I4 engine. When manufacturers (Atkinson Cycle Engines) downsizing technologies are separately reduced the number of banks or Atkinson cycle gasoline engines use modeled in this analysis. cylinders of an engine, some cost changes in valve timing (e.g., late- The analysis also modeled savings is projected due to fewer intake-valve-closing or LIVC) to reduce turbocharged engines with parallel cylinders and fewer valves. Such cost the effective compression ratio while hybrid technology. In simulations with savings is projected to help offset the maintaining the expansion ratio. This high stringencies, many manufacturers additional costs of turbo charger specific approach allows a reduction in top- produced turbo-hybrid electric vehicles. hardware, making turbo downsizing a dead-center (TDC) clearance ratio (e.g., In the MY 2016 fleet, of the vehicles that very attractive technology progression 137 increase in ‘‘mechanical’’ or ‘‘physical’’ use parallel hybrid technology, many for some engines. compression ratio) to increase the use turbocharged engines. (a) TURBO1 effective expansion ratio without Since the Draft TAR, the turbo family increasing the effective compression Level 1 Turbo Charging (TURBO1) engine maps were updated to reflect ratio to a point that knock-limited adds a turbo charger to a DOHC engine operation on 87 AKI regular octane operation is encountered. Increasing the 136 with SGDI, VVT, and continuously VVL. fuel. In the Draft TAR, turbo engine expansion ratio in this manner improves The engine operates at up to 18 bar maps were developed assuming thermal efficiency but also lowers peak brake mean effective pressure (BMEP). premium fuel. For this rulemaking BMEP, particularly at lower engine Manufacturers used Turbo1 analyses, pathways to improving fuel speeds. economy and CO2 are analyzed, while technology in a little less than a quarter Often knock concerns for these also maintaining vehicle performance, of the MY 2016 fleet with particularly engines limit applications in high load, capability, and other attributes. This high concentrations in premium low RPM conditions. Some includes assuming there is no change in vehicles. manufacturers have mitigated knock the fuel octane required to operate the (b) TURBO2 concerns by lowering back pressure vehicle. Using 87 AKI regular octane with long, intricate exhaust systems, but Level 2 Turbo Charging (TURBO2) fuel is consistent with the fuel octane these systems must balance knock operates at up to 24 bar BMEP. The step that manufacturers specify for the performance with emissions tradeoffs, from Turbo1 to Turbo2 is accompanied majority of vehicles, and enables the and the increased size of the exhaust with additional displacement modeling to account for important manifold can pose packaging concerns, downsizing for reduced pumping losses. design and calibration issues associated particularly on V-engine Very few manufacturers have Turbo2 configurations.138 133 Boost refers to the degree to which the technology in the MY 2016 fleet. Only a few manufacturers produced turbocharger compresses the intake air for the engine, which may affect the specific power of the (c) CEGR1 internal combustion engine vehicles engine. with Atkinson cycle engines in MY Turbo Charging with Cooled Exhaust 134 Knock refers to rapid uncontrolled combustion Gas Recirculation (CEGR1) improves the in the cylinder part way through the combustion 138 Some HCR1 4-cylinder (I–4) engines use an process, which can create an audible sound and can knock resistance of Turbo2 engines by intricate 4–2–1 exhaust manifold to lower damage the engine. mixing cooled inert exhaust gases into backpressure and to improve engine efficiency. 135 Turbo lag refers to the delay time between the engine’s air intake. That allows Manufacturers sometimes fitted such an exhaust power demanded and power delivered; it is greater boost levels, more optimal spark system into a front-wheel-drive vehicle with an I– typically associated with rapid accelerations from a 4 engine by using a high underbody tunnel or stopped vehicle at idle. timing for improved fuel economy, and rearward dashpanel (trading off some interior 136 Specifically, 87 Anti-Knock Index (AKI) Tier space), but packaging such systems on rear-wheel- 3 certification fuel. 87 AKI is also known as 87 137 In particular, the step from a naturally- drive vehicles may pose challenges, especially if the (R+M)/2 or 87 (Research Octane + Motor Octane)/ aspirated V6 to a turbo I4 was particularly cost engine has two banks and would therefore require 2. effective in agency simulations. room for two such exhaust manifolds.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00053 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43038 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

2016; however, these engines are engine as outlined in the EPA SAE attributable to friction and thermal commonly paired with hybrid electric paper has ever been commercially efficiency before drawing conclusions. vehicle technologies due to the synergy produced or even produced as a Using vehicle level data may of peak efficiency of Atkinson cycle prototype in a lab setting. Furthermore, misrepresent or conflate complex engines and immediate torque from the engine map has not been validated interactions between a high thermal electric motors in strong hybrids. with hardware and bench data, even on efficiency engine, engine friction Atkinson cycle engines are very a prototype level (as no such engine reduction, accessory load common on power split hybrids and are exists to test to validate the engine improvements, transmission sometimes observed as part of a parallel map). technologies, mass reduction, hybrid system or plug-in hybrid system. Previously, EPA relied heavily on the aerodynamics, rolling resistance, and Atkinson cycle engines played a HCR2 (or sometimes referred to as ATK2 other vehicle technologies. For instance, prominent role in EPA’s January 2017 in previous EPA analysis) engine as a some of the newest high compression final determination, which has since cost effective pathway to compliance for ratio engines show improved thermal been withdrawn. Today’s analysis stringent alternatives, but many engine efficiency, in large part due to improved recognizes that the technology is not experts questioned its technical accessory loads or reduced parasitic suitable for many vehicles due to feasibility and near term commercial losses from accessory systems.144 The performance, emissions and packaging practicability. Stakeholders asked for CAFE model allows for incremental issues, and/or the extensive capital and the engine to be removed from improvement over existing HCR1 resources that would be required for compliance simulations until the technologies with the addition of manufacturers to shift from other performance could be validated with improved accessory devices (IACC), a powertrain technology pathways (such engine hardware.141 142 While for the technology that is available to be as turbocharging and downsizing) to Draft TAR, the agencies ran full-vehicle applied on many baseline MY 2016 standalone Atkinson cycle engine simulations with the theoretical engine vehicles with HCR1 engines and may be technology. map and made these available in the applied as part of a pathway of CAFE model, HCR2 technology as compliance to further improve the (a) HCR1 described in EPA’s SAE paper was not effectiveness of existing HCR1 engines. A number of Asian manufacturers included in today’s analysis because have launched Atkinson cycle engines there has been no observable physical (c) Emerging Gasoline Engine in smaller vehicles that do not use demonstration of the speculative Technologies technology, and many questions remain hybrid technologies. These production Manufacturers and suppliers continue engines have been benchmarked to about its practicability as specified, especially in high load, low engine to invest in many emerging engine characterize HCR1 technology for technologies, and some of these today’s analysis. speed operating conditions. Simulations with EPA’s HCR2 engine map produce technologies are on the cusp of Today’s analysis restricted the commercialization. Often, application of stand-alone Atkinson results that approach (and sometimes exceed) diesel powertrain efficiency.143 manufacturers submit information about cycle engines in the CAFE model in new engine technologies that they may some cases. The engines benchmarked Given the prominence of this unproven technology in previous rule-makings, soon bring into production. When this for today’s analysis were not suitable for happens, a collaborative effort is MY 2016 baseline vehicle models that the CAFE model may be configured to consider the application of HCR2 undertaken with suppliers and have 8-cylinder engines and in many manufacturers to learn as much as cases 6-cylinder engines. technology for reference only. As new engines emerge that achieve possible and sometimes begin (b) HCR2 high thermal efficiency, questions may simulation modeling efforts. Bench data, or performance data for preproduction EPA conceptualized a ‘‘future’’ be raised as to whether the HCR2 engine is a simulation proxy for the new engine vehicles and engines, is usually closely Atkinson cycle engine and published held confidential business information. the theoretical engine map in an SAE technology. It is important to conduct a 139 140 thorough evaluation of the actual new To properly characterize the paper. For this engine, EPA staff technologies, it is often necessary to began with a best-in-class 2.0L Atkinson production engines to measure the brake specific fuel consumption and to wait until the engine technologies are in cycle engine and then increased the production to study them. efficiency of the engine map further, characterize the improvements through the theoretical application of (1) Advanced Cylinder Deactivation 141 At NHTSA–2016–0068–0082, FCA additional technologies in combination, recommended, ‘‘Remove ATK2 from OMEGA (ADEAC) like cylinder deactivation, engine model until the performance is validated.’’, p. viii. friction reduction, and cooled exhaust And FCA stated, ‘‘ATK2—High Compression Advanced cylinder deactivation gas recirculation. This engine remains engines coupled with Cylinder Deactivation and systems (or rolling or dynamic cylinder Cooled EGR are unlikely to deliver modeled results, deactivation systems) allows a further entirely speculative, as no production meet customer needs, or be ready for commercial application.’’, p. 6–9. degree of cylinder deactivation than 139 Ellies, B., Schenk, C., and Dekraker, P., 142 At Docket ID No EPA–HQ–OAR–2015–0827– DEAC. The technology allows the ‘‘Benchmarking and Hardware-in-the-Loop 6156, The Alliance of Automobile Manufacturers engine to vary the percentage of Operation of a 2014 MAZDA SkyActiv 2.0L 13:1 commented, ‘‘[There] is no current example of cylinders deactivated and the sequence Compression Ratio Engine,’’ SAE Technical Paper combined Atkinson, plus cooled EGR, plus cylinder 2016–01–1007, 2016. Available at https:// deactivation technology in the present fleet to verify in which cylinders are deactivated, www.sae.org/publications/technical-papers/ EPA’s modeled benefits and . . . EPA could not essentially providing ‘‘displacement on content/2016-01-1007/. provide physical test results replicating its modeled demand’’ for low load operations, so 140 Lee, S., Schenk, C., and McDonald, J., ‘‘Air benefits of these combined technologies,’’ p. 40. long as the calibration avoids certain Flow Optimization and Calibration in High- 143 Thomas, J. ‘‘Drive Cycle Powertrain frequencies. Compression-Ratio Naturally Aspirated SI Engines Efficiencies and Trends Derived from EPA Vehicle with Cooled-EGR,’’ SAE Technical Paper 2016–01– Dynamometer Results,’’ SAE Int. J. Passeng. Cars— 0565, 2016. Available at https://www.sae.org/ Mech. Syst. 7(4):2014. Available at https:// 144 For instance, the MY 2018 2.5L Camry engine publications/technical-papers/content/2016-01- www.sae.org/publications/technical-papers/ that uses HCR technology also reduces parasitic 0565/. content/2014-01-2562/. losses with a variable capacity oil pump.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00054 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43039

ADEAC systems may be integrated Should VCR technology be employed in modeling for past rulemakings, and is into the valvetrains with moderate the timeframe of this proposed not included in this NPRM analysis, modifications on OHV engines. rulemaking? Why or why not? Do primarily because effectiveness, cost, However, while the ADEAC operating commenters believe VCR technology and mass market implementation concept remains the same on DOHC will see widespread adoption in the US readiness data are not available. engines, the valvetrain hardware vehicle fleet? Why or why not? What Please comment on the potential use configuration is very different, and vehicle segments may it best be suited of HCCI technology in the timeframe application on DOHC engines is for, and which segments would it not be covered by this rule. More specifically, projected to be more costly per cylinder best suited for? Why or why not? What should HCCI be included in the final due to the valvetrain differences. cost and effectiveness values should be rulemaking analysis for this proposed Some preproduction 8-cylinder OHV used if VCR is modeled for analysis? rulemaking? Why or why not? Please prototype vehicles were briefly Please provide supporting data. provide supporting data, including evaluated for this analysis, but no Additionally, please provide any effectiveness values, costs in relation production versions of the technology comments on the sponsored work varying engine types and applications, have been studied. related to VCR, described further in and production timing that supports the Today’s analysis relied on CBI to PRIA Chapter 6.3. timeframe of this rulemaking. estimate costs and effectiveness values of ADEAC. Since no engine map was (3) Compression Ignition Gasoline (d) Diesel Engines Engines (SpCCI, HCCI) available at the time of the NPRM Diesel engines have several analysis, ADEAC was estimated to For many years, engine developers, characteristics that give superior fuel improve a basic engine with VVL, VVT, researchers, manufacturers have efficiency, including reduced pumping SGDI, and DEAC by three percent (for 4 explored ways to achieve the inherent losses due to lack of (or greatly reduced) cylinder engines) six percent (for efficiency of a diesel engine while throttling, high pressure direct injection engines with more than 4 cylinders). maintaining the operating of fuel, a combustion cycle that operates ADEAC systems will continue to be characteristics of a gasoline engine. A at a higher compression ratio, and a very studied as production begins. potential pathway for striking this lean air/fuel mixture relative to an balance is utilizing compression equivalent-performance gasoline engine. (2) Variable Compression Ratio Engines ignition for gasoline fueled engines, (VCR) This technology requires additional more commonly referred to as enablers, such as a NO adsorption Engines using variable compression X Homogeneous Charge Compression catalyst system or a urea/ammonia ratio (VCR) technology appear to be at Ignition (HCCI). selective catalytic reduction system for a production-intent stage of Ongoing, periodic discussions with control of NO emissions during lean development but also appear to be manufacturers on future fuel saving X (excess air) operation. targeted primarily towards limited technologies and powertrain plans have, production, high performance and very generally, included HCCI as a long-term (e) Alternative Fuel Engines strategy. The technology appears to high BMEP (27–30 bar) applications. (1) Compressed Natural Gas (CNG) Variable compression ratio engines always be a strong consideration as, in work by changing the length of the theory, it provides the ‘‘best of both Compressed Natural Gas (CNG) piston stroke of the engine to operate at worlds,’’ meaning a way to provide engines use compressed natural gas as a a more optimal compression ratio and diesel engine efficiency with gasoline fuel source. The fuel storage and supply improve thermal efficiency over the full engine performance and emissions systems for these engines differ range of engine operating conditions. levels. tremendously from gasoline, diesel, and A number of manufacturers and Developments in both the research flex fuel vehicles. suppliers provided information about and the potential production (2) Flex Fuel Engines VCR technologies, and several design implementation of HCCI for the US concepts were reviewed that could market is continually assessed. In 2017, Flex fuel engines can run on regular achieve a similar functional outcome. In a significant, potentially production gasoline and fuel blended with ethanol. addition to design concept differences, breakthrough was announced by Mazda These vehicles may require additional intellectual property ownership regarding a gasoline-fueled engine equipment in the fuel system to complicates the ability of the agencies to employing Spark Controlled effectively supply different blends of define a VCR hardware system that Compression Ignition (SpCCI), where fuel to the engine. HCCI is employed for a portion of its could be widely adopted across the (f) Lubrication and Friction Reduction industry. normal operation and spark ignition is For today’s analysis, VCR engines used at other times.145 Soon after, Low-friction lubricants including low have a spot on the technology Mazda publicly stated they plan to viscosity and advanced low friction simulation tree, but VCR is not actively introduce this engine as part of the lubricant oils are now available (and used in the NPRM simulation. Skyactiv family of engines in 2019.146 widely used). If manufacturers choose to Reasonable representations of costs and However, HCCI was not included in make use of these lubricants, they may technology characterizations remain the simulation and vehicle fleet need to make engine changes and open questions for VCR engine conduct durability testing to technology and the analysis. 145 Mazda Next-Generation Technology—Press accommodate the lubricants. The level NHTSA is sponsoring work to Information, Mazda USA (Oct. 24, 2017), https:// of low friction lubricants exceeded 85% insidemazda.mazdausa.com/press-release/mazda- develop engine maps for additional next-generation-technology-press-information/ (last penetration in the MY 2016 fleet. combinations of technologies. Some of visited Apr. 13, 2018). Reduction of engine friction can be these technologies being researched 146 Mazda introduces updated 2019 CX–3 at 2018 achieved through low-tension piston presently, including VCR, may be used New York International Auto Show, Mazda USA rings, roller cam followers, improved (Mar. 28, 2018), https:// in the analysis supporting the final rule. insidemazda.mazdausa.com/press-release/mazda- material coatings, more optimal thermal Please provide comment on variable introduces-2019-cx-3-2018-new-york-auto-show/ management, piston surface treatments, compression ratio engine technology. (last visited Apr. 13, 2018). and other improvements in the design of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00055 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43040 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

engine components and subsystems that Currently, throughout the United could also limit consumer choice, or, in improve efficient engine operation. States, pump fuel is a blend of 90% the case of higher elevation regions, Manufacturers have already widely gasoline and 10% ethanol. It is standard present an opportunity for consumers to adopted both lubrication and friction practice for refiners to manufacture use a fuel grade that is below the reduction technologies. This analysis gasoline and ship it, usually via minimum recommended. As such, includes advanced engine maps that pipelines, to bulk fuel terminals across vehicle manufacturers employ strategies already assume application of low- the country. In many cases, refiners for scenarios where a lower than friction lubricants and engine friction supply lower octane fuels than the recommended, or required, fuel grade is reduction technologies. Therefore, minimum 87-octane required by law to used, mitigating engine damage over the additional friction reduction is not these terminals. The terminals then life of a vehicle. considered in today’s analysis. perform blending operations to bring the When knock (also referred to as The use and commercial development fuel octane level up to the minimum detonation) is encountered during of improved lubricants and friction required by law, and higher. In some engine operation, at the most basic reduction components will continue to cases, typically to lowest fuel grade, the level, non-turbo charged engines can be monitored, including conical boring ‘‘base fuel’’ is blended with ethanol, reduce or eliminate knock by adjusting and oblong cylinders, and future which has a typical octane rating of the timing of the spark that ignites the analyses may be updated if new approximately 113. For example, in fuel, as well as the amounts of fuel information becomes available. 2013, the State of Nebraska Ethanol injected at each intake stroke 5. Fuel Octane Board defined requirements for refiners (‘‘fueling’’). In turbo-charged to 84-octane gas for blending to achieve applications, boost levels are typically (a) What is fuel octane level? 87-octane prior to final dispensing to reduced along with spark timing and Gasoline octane levels are an integral consumers.150 fueling adjustments. Past rulemakings part of potential engine performance. have also discussed other techniques (b) Fuel Octane Level and Engine that may be employed to allow higher According the United States Energy Performance Information Administration (EIA), compression ratios, more optimal spark octane ratings are measures of fuel A typical, overarching goal of optimal timing to be used without knock, such stability. These ratings are based on the spark-ignited engine design and as the addition of cooled exhaust gas pressure at which a fuel will operation is to maximize the greatest recirculation (EGR). Regardless of the spontaneously combust (auto-ignite) in amount of energy from the fuel type of spark-ignition engine or a testing engine.147 Spontaneous available, without manifesting technology employed, reducing or combustion is an undesired condition detrimental impacts to the engine over preventing knock results in the loss of that will lead to serious engine damage its expected operating conditions. potential power output, creating a and costly repairs for consumers if not Design factors, such as compression ‘‘knock-limited’’ constraint on properly managed. The higher an octane ratio, intake and exhaust value control performance and efficiency. number, the more stable the fuel, specifications, combustion chamber and Despite limits imposed by available mitigating the potential for spontaneous piston characteristics, among others, are fuel grades, manufacturers continue to combustion, also commonly known as all impacted by octane (stability) of the make progress in extracting more power ‘‘knock.’’ Modern engine control fuel consumers are anticipated to use.151 and efficiency from spark-ignited systems are sophisticated and allow Vehicle manufacturers typically engines. Production engines are safely manufacturers to detect when ‘‘knock’’ develop their engines and engine operating with regular 87 AKI fuel with occurs during engine operation. These control system calibrations based on the compression ratios and boost levels control systems are designed to adjust fuel available to consumers. In many once viewed as only possible with operating parameters to reduce or cases, manufacturers may recommend a premium fuel. According to the eliminate ‘‘knock’’ once detected. fuel grade for best performance and to Department of Energy, the average In the United States, consumers are prevent potential damage. In some gasoline octane level has remained typically able to select from three cases, manufacturers may require a fundamentally flat starting in the early distinct grades of fuel, each of which specific fuel grade for both best 1980’s and decreased slightly starting in provides a different octane rating. The performance and/or to prevent potential the early 2000s. During this time, octane levels can vary from region to engine damage. however, the average compression ratio region, but on the majority, the octane Consumers, though, may or may not for the U.S. fleet has increased from 8.4 levels offered are regular (the lowest choose to follow the recommendation or to 10.52, a more than 20% increase, octane fuel–generally 87 Anti-Knock requirement for a specific fuel grade. yielding the statement that, ‘‘There is Index (AKI) also expressed as (the Additionally, regional fuel availability some concern that in the future, auto average of Research Octane + Motor manufacturers will reach the limit of Octane), midgrade (the middle range visited Mar. 19, 2018). 85 octane fuel is available technological increases in compression in high-elevation regions where the barometric octane fuel–generally 89–90 AKI), and ratios without further increases in the pressure is lower causing naturally-aspirated 152 premium (the highest octane fuel– engines to operate with less air and, therefore, at octane of the fuel.’’ generally 91–94 AKI).148 At higher lower torque and power. This creates less benefit As such, manufacturers are still elevations, the lowest octane rating and need for higher octane fuels as compared to at limited by the available fuel grades to lower elevations where engine airflow, torque, and available can drop to 85 AKI.149 consumers and the need to safeguard power levels are higher. the durability of their products for all of 150 Nebraska Ethanol Board, Oil Refiners Change 147 U.S. Energy Information Administration, What Nebraska Fuel Components, Nebraska.gov, http:// the available fuels; thus, the potential is Octane?, https://www.eia.gov/energyexplained/ ethanol.nebraska.gov/wordpress/oil-refiners- _ index.cfm?page=gasoline home#tab2 (last visited change-nebraska-fuel-components/ (last visited 152 Fact of the Week, Fact #940: August 29, 2016 Mar. 19, 2018). Mar. 19, 2018). Diverging Trends of Engine Compression Ratio and 148 Id. 151 Additionally, PRIA Chapter 6 contains a brief Gasoline Octane Rating, U.S. Department of Energy, 149 See e.g., U.S. Department of Energy and U.S. discussion of fuel properties, octane levels used for https://www.energy.gov/eere/vehicles/fact-940- Environmental Protection Agency, What is 85 engine simulation and in real-world testing, and august-29-2016-diverging-trends-engine- octane, and is it safe to use in my vehicle?, https:// how octane levels can impact performance under compression-ratio-and-gasoline-octane (last visited www.fueleconomy.gov/feg/octane.shtml#85 (last these test conditions. Mar. 21, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00056 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43041

improvement in the design of spark- and/or the Draft TAR regarding the transmission may work with ignition engines continues to be potential for increasing octane levels for complementary equipment to improve overshadowed by the fuel grades the U.S. market. fuel economy. In the meetings with HOLC and the available to consumers. Today’s analysis significantly FFA, the groups advocated for the (c) Potential of Higher Octane Fuels potential benefits high octane fuels increased the number of transmissions Automakers and advocacy groups could provide via the blending of non- modeled in full-vehicle simulations, have expressed support for increases to petroleum feedstocks to increase octane attempting to more closely align the fuel octane levels for the U.S. market levels available at the pump. The Department of Energy full-vehicle and are actively participating in groups’ positions on benefits took both simulations with existing vehicles. Department of Energy research programs a technical approach by suggesting an Previously, EPA included just five on the potential of higher octane fuel octane level of 100 is desired for the transmissions 155 by vehicle class in usage.153 154 Some positions for potential marketplace, as well as, the benefits their analysis, and often EPA future octane levels include advocacy from potential increased national energy represented upgrades among manual, for today’s premium grade becoming the security by reduced dependencies on automatic, continuously variable, and base grade of fuel available, which foreign petroleum. dual clutch transmissions with the same could enable low cost design changes 156 157 (d) Fuel Octane—Request for Comments effectiveness and cost values that would improve fuel economy and within a vehicle class. Today’s analysis CO2. Challenges associated with this Please comment on the potential simulated nearly 20 transmissions, with approach include the increased fuel cost benefits, or dis-benefits, of considering explicit assumptions about gear ratios, the impacts of increased fuel octane to consumers who drive vehicles gear efficiencies, gear spans, shift logic, designed for current regular octane levels available to consumers for and transmission architecture.158 159 grade fuel that would not benefit from purposes of the model. More This analysis improves transparency by the use of the higher cost higher octane specifically, please comment on how fuel. The net costs for a shift to higher increasing fuel octane levels would play making clear the assumptions octane fuel would persist well into the a role in product offerings and engine underlying the transmissions in the full- future. Net benefits for the transition technologies. Are there potential vehicle simulations and by increasing would not be achieved until current improvements to fuel economy and CO2 the number of transmissions simulated regular octane fuel is not available in reductions from higher octane fuels? since the Draft TAR. Methods will be the North American market, causing Why or why not? What is an ideal continuously evaluated to improve manufacturers to redesign all engines to octane level for mass-market transmission models in full-vehicle operate the higher octane fuel, and then consumption balanced against cost and simulations. For the box plots of after those vehicles have been in potential benefits? What are the effectiveness values, as shown in the production a sufficient number of model negatives associated with increasing the PRIA Chapter 6, all automatic years to largely replace the current on- available octane levels and, potentially, transmissions are relative to a 5-speed road vehicle fleet. The transition to net eliminating today’s lower octane fuel automatic, and all manual transmissions positive benefits could take many years. blends? Please provide supporting data are relative to a 5-speed manual. Table In anticipation of this proposed for your position(s). II–11 below shows the absolute costs of rulemaking, organizations such as the 6. Transmission Technologies transmission used for this analysis High Octane Low Carbon Alliance including learning and retail price Transmissions transmit torque from (HOLC) and the Fuel Freedom equivalent. Foundation (FFA), have shared their the engine to the wheels. Transmissions positions on the potential for making may improve fuel efficiency primarily 155 Null, TRX11, TRX12, TRX21, TRX22. higher octane fuels available for the U.S. through two mechanisms: (1) 156 Draft TAR, p. 5–297 through 5–298 market. Other stakeholders also Transmissions with more gears allow summarizes effectiveness values previously commented to past NHTSA rulemakings the engine to operate more regularly at assumed for stepping between transmission the most efficient speed-load points, technologies (Null, TRX11, TRX12, TRX21, TRX22). 157 153 Mark Phelan, High octane gas coming—but and (2) transmissions may have Draft TAR, p. 5–299. ‘‘For future vehicles, it you’ll pay more for it, Detroit Free Press (Apr. 25, improvements in friction (gears, was assumed that the costs for transitioning from 2017), https://www.freep.com/story/money/cars/ one technology level (TRX11–TRX22) to another mark-phelan/2017/04/25/new-gasoline-promises- bearings, seals, and so on), or level is the same for each transmission type (AT, lower-emissions-higher-mpg-and-cost-octane- improvements in shift efficiency that AMT, DCT, and CVT).’’ society-of-automotive-engineers/100716174/. help the transmission transfer torque 158 See PRIA Chapter 6.3. 154 The octane game: Auto industry lobbies for 95 more efficiently, lowering parasitic 159 Ehsan, I.S., Moawad, A., Kim, N., & Rousseau, as new regular, Automotive News (April 17, 2018), A. ‘‘A Detailed Vehicle Simulation Process To http://www.autonews.com/article/20180417/ losses. These mechanisms are very Support CAFE Standards.’’ ANL/ESD–18/6. Energy BLOG06/180419780/the-octane-game-auto- different, so full-vehicle simulation is Systems Division, Argonne National Laboratory. industry-lobbies-for-95-as-new-regular. helpful to understand how a 2018.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00057 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43042 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(a) Automatic Transmissions higher speed transmissions were more (1) Continuously Variable prevalent in the MY 2016 fleet. Transmissions Five-, six-, seven-, eight-, nine- and ten-speed automatic transmissions are ‘‘L2’’ and ‘‘L3’’ transmissions Continuously variable transmission optimized by changing the gear ratios to designate improved gear efficiency and (CVT) commonly uses V-shaped pulleys enable the engine to operate in a more reduced parasitic losses. Few connected by a metal belt rather than efficient operating range over a broader transmissions in the MY 2016 fleet have gears to provide ratios for operation. range of vehicle operating conditions. achieved ‘‘L2’’ efficiency, and the Unlike manual and automatic While a six speed transmission highest level of transmission efficiencies transmissions with fixed transmission application was most prevalent for the modeled are assumed to be available in ratios, continuously variable MYs 2012–2016 final rule, eight and MY 2022. transmissions can provide fully variable and an infinite number of transmission

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00058 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.026 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43043

ratios that enable the engine to operate gear is pre-selected, which allows for analysis limits the application of in a more efficient operating range over faster and smoother shifting. The 2012– improved DCTs to vehicles that already a broader range of vehicle operating 2016 final rule limited DCT applications use DCTs. Many of these vehicles are conditions. In this NPRM, two levels of to a maximum of 6-speeds. Both 6-speed imported performance products. CVTs are considered for future vehicles. and 8-speed DCT transmissions are (b) Manual Transmissions The second level CVT would have a considered in today’s proposal. wider transmission ratio, increased Dual clutch transmissions are very Manual 6- and 7-speed transmissions torque capacity, improvements in oil effective transmission technologies, and offer an additional gear ratio, sometimes pump efficiency, lubrication previous rule-making projected rapid, with a higher overdrive gear ratio, over improvements, and friction reduction. and wide adoption into the fleet. a 5-speed manual transmission. Similar While CVTs work well with light loads, However, early DCT product launches to automatic transmissions, more gears the technology as modeled is not in the U.S. market experienced shift often means the engine may operate in suitable for larger vehicles such as harshness and poor launch the efficient zone more frequently. trucks and large SUVs. performance, resulting in customer satisfaction issues—some so extreme as 7. Vehicle Technologies (2) Dual Clutch Transmissions to prompt vehicle buyback As discussed earlier in Section 160 Dual clutch or automated shift campaigns. Most manufacturers are II.D.1.b)(1), several technologies were manual transmissions (DCT) are similar not using DCTs in the U.S. market due considered for this analysis, and Table to manual transmissions except for the to the customer satisfaction issues. II–12, Table II–13, and Table II–14 vehicle controls shifting and launch Manufacturers used DCTs in about three below shows the full list of vehicle functions. A dual-clutch automated shift percent of the MY 2016 fleet. Today’s technologies analyzed and the manual transmission uses separate associated absolute cost.161 160 Ford Powershift Transmission Settlement, clutches for even-numbered and odd- http://fordtransmissionsettlement.com/ (last visited numbered gears, so the next expected June 21, 2018). 161 Mass reduction costs are in $/lb.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00059 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43044 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00060 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.027 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43045

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00061 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.028 43046 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(a) Reduced Rolling Resistance rolling resistance reduction. In this (b) Reduced Aerodynamic Drag Lower-rolling-resistance tires have NPRM, baseline vehicle reference Coefficient characteristics that reduce frictional rolling resistance values were determined based on the MY 2016 Aerodynamic drag reduction can be losses associated with the energy achieved via two approaches, either by dissipated mainly in the deformation of vehicles rather than the MY 2008 reducing the drag coefficients or the tires under load, thereby improving vehicles used in the 2012 final rule. reducing vehicle frontal area. To reduce fuel economy and reducing CO Rolling resistance values were assigned 2 the drag coefficient, skirts, air dams, emissions. New for this proposal, and based on CBI shared by manufacturers. underbody covers, and more also marking an advance over low In some cases, low rolling resistance rolling resistance tires considered aerodynamic side view mirrors can be tires can affect traction, which may be applied. In the MY 2017–2025 final rule during the heavy duty greenhouse gas untenable for some high performance rulemaking,162 is a second level of lower and the 2016 Draft TAR, the analysis vehicles. For cars and SUVs with more rolling resistance tires that reduce included two levels of aerodynamic than 405 horsepower, the analysis frictional losses even further. The first technologies. The second level included level of low rolling resistance tires will restricted the application of the highest active grille shutters, rear visors, and have 10% rolling resistance reduction levels of rolling resistance. For cars and larger under body panels. This NPRM while the second level would have 20% SUVs with more than 500 horsepower, expanded the aerodynamic drag the analysis restricted the application of improvements from two levels to four to 162 See 76 FR 57106, at 57207, 57229 (Sep. 15, any additional rolling resistance provide more discrete levels. The NPRM 2011). technology. levels are 5%, 10%, 15%, and 20%

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00062 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.029 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43047

improvement relative to baseline modeling. That level of drag reduction (e) Secondary Axle Disconnect (SAX) reference vehicles. The agencies relied is likely not technologically feasible on the wind tunnel testing performed by with today’s technology, and the Front or secondary axle disconnect for National Research Council (NRC), analysis accordingly restricted the all-wheel drive systems provides a Canada, Transport Canada (TC), and application of advanced levels of torque distribution disconnect between Environment and Climate Change, aerodynamics in some instances, such front and rear axles when torque is not Canada (ECCC) to quantify the as in this case, due to bodystyle form required for the non-driving axle. This aerodynamic drag impacts of various drag limitations. Separate from form results in the reduction of associated OEM aerodynamic technologies and to drag limitations, some high performance parasitic energy losses. explore the improvement potential of vehicles already use advanced 8. Electrification Technologies these technologies by expanding the aerodynamics technologies to generate capability and/or improving the design down force, and sometimes these For this NPRM, the analysis of of MY 2016 state-of-the-art aerodynamic applications must trade-off between electrification technologies relies treatments. The agencies estimated the aerodynamic drag coefficient reduction primarily on research published by the level of aerodynamic drag in each and down force. Today’s analysis does Department of Energy, ANL.164 ANL’s vehicle model in the MY 2016 baseline not apply 15% or 20% aerodynamic assumptions regarding all hybrid fleet and gathered CBI on aerodynamic drag coefficient reduction to cars and systems, including belt-integrated drag coefficients, so each vehicle has an SUVs with more than 405 horsepower. starter generators, strong parallel and appropriate initial value for further series hybrids, plug-in hybrids,165 and (c) Mass Reduction improvements. battery electric vehicles, and most Notably, today’s analysis assumes Mass Reduction can be achieved in projected technology costs were adopted aerodynamic drag reduction can only many ways, such as material for this analysis. In addition, the most come from reduction in the substitution, design optimization, part recent ANL BatPaC model is used to aerodynamic drag coefficient and not consolidation, improving manufacturing estimate battery costs. Table II–15 and from reduction of frontal area.163 For process, etc. The analysis utilizes mass Table II–16 below show the absolute some bodystyles, the agencies have no reduction levels of 5, 10, 15, and 20% costs of all electrification technologies evidence that manufacturers may be relative to a reference glider vehicle for estimated for this NPRM analysis able to achieve 15% or 20% each vehicle subsegment. For HEV, relative to a basic internal combustion aerodynamic drag coefficient reduction PHEV, and BEV vehicles, net mass engine vehicle with a 5-speed automatic relative to baseline for some bodystyles reduction was considered, including the transmission.166 (for instance, with pickup trucks) due to mass reduction applied to the glider and form drag limitions. Previously, EPA the added mass of electrification 164 Moawad et al., Assessment of vehicle sizing, analysis assumed some vehicles from all components. An extensive discussion of energy consumption, and cost through large-scale simulation of advanced engine technologies, bodystyles could (and would) reduce mass reduction technologies as well as Argonne National Laboratory (March 2016), aerodynamic forces by 20%, which in the cost of mass reduction is located in available at https://www.autonomie.net/pdfs/ some cases led to future pickup trucks Chapter 6.3 of the PRIA. The analysis Report%20ANL%20ESD-1528%20-%20Assessment having aerodynamic drag coefficients included an estimated level of mass %20of%20Vehicle%20Sizing,%20Energy%20 Consumption%20and%20Cost%20through better than some of today’s typical cars, reduction technology in each vehicle %20Large%20Scale%20Simulation%20 if frontal area were held constant. While model in the MY 2016 baseline fleet so of%20Advanced%20Vehicle%20Technologies%20- ANL created full-vehicle simulations for that each vehicle model has an %201603.pdf. trucks with 20% drag reduction, those appropriate initial value for further 165 Notably all power split hybrids, and all plug- improvements. in hybrid vehicles were assumed to be paired with simulations were not used in the CAFE a high compression ratio internal combustion (d) Low Drag Brakes (LDB) engine for this analysis. 163 EPA previously assumed that manufacturers 166 Note: These costs do not include value loss for could reduce frontal area as well as aerodynamic Low-drag brakes reduce the sliding HEVs, PHEVs, and BEVs. Powertrain hardware drag coefficient to achieve 20% aerodynamic force friction of disc brake pads on rotors between cars and small SUV’s is often similar, reduction relative to ‘‘Null’’ or initial aerodynamic especially if technology is used vehicles on the technology level; however, reducing frontal area when the brakes are not engaged same platform; however, battery pack sizes may would likely degrade other utility features like because the brake pads are pulled away vary meaningfully to deliver similar range in interior volume, or ingress/egress. from the rotors. different applications.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00063 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43048 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00064 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.030 EP24AU18.031 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43049

(a) Hybrid Technologies electric operation and more efficient a control system that allows the battery (1) 12-Volt Stop-Start brake-energy recovery. Engaging the pack to be substantially depleted under clutch allows efficient coupling of the electric-only or blended mechanical/ 12-volt Stop-Start, sometimes referred engine and electric motor and, when electric operation and batteries that can to as idle-stop or 12-volt micro hybrid, combined with a DCT transmission, be cycled in charge sustaining operation is the most basic hybrid system that reduces gear-train losses relative to at a lower state of charge than is typical facilitates idle-stop capability. These power-split or 2-mode hybrid systems. of other hybrid electric vehicles. These systems typically incorporate an Battery costs are now considered vehicles are sometimes referred to as enhanced performance battery and other separately from other HEV hardware. Range Extended Electric Vehicles features such as electric transmission P2 Hybrid systems typically rely on (REEV). In this NPRM analysis, PHEVs pump and cooling pump to maintain the internal combustion engine to with two all-electric ranges—both a 30 vehicle systems during idle-stop. deliver high, sustained power levels. mile and a 50 mile all-electric range— (2) Higher Voltage Stop-Start/Belt While many vehicles may use HCR1 have been included as potential Integrated Starter Generator engines as part of a hybrid powertrain, technologies. Again, battery costs are HCR1 engines may not be suitable for all now considered separately from other Higher Voltage Stop-Start/Belt vehicles, especially high performance PHEV hardware. Integrated Starter Generator (BISG), vehicles, or vehicles designed to carry The ANL Autonomie simulations sometimes referred to as a mild hybrid or tow large loads. Many manufacturers assumed all PHEVs use a high system, provides idle-stop capability may prefer turbo engines (with high compression ratio engine. Therefore, all and uses a higher voltage battery with specific power output) for P2 Hybrid vehicles equipped with PHEV increased energy capacity over typical systems. technology in the CAFE model inputs automotive batteries. The higher system (5) Power-Split Hybrid and simulations are assumed to have voltage allows the use of a smaller, more high compression ratio engines. In powerful electric motor. This system Power-split Hybrid (SHEVPS) is a practice, many PHEVs recently replaces a standard alternator with an hybrid electric drive system that introduced in the marketplace use enhanced power, higher voltage, higher replaces the traditional transmission turbo-charged engines in the PHEV with a single planetary gearset and a efficiency starter-alternator, that is belt system, and this is particularly true for motor/generator. This motor/generator driven and that can recover braking PHEVs produced by European uses the engine to either charge the energy while the vehicle slows down manufacturers and for other PHEV battery or supply additional power to (regenerative braking). Today’s analysis performance vehicle applications. assumes 48V systems on cars and small the drive motor. A second, more Please provide comment on the SUVs and high voltage systems for large powerful motor/generator is modeling of PHEV systems. Should SUVs and trucks. Future analysis may permanently connected to the vehicle’s turbo PHEVs be considered instead, or reference the application and operation final drive and always turns with the in addition to high compression ratio of 48V systems on large SUVs and wheels. The planetary gear splits engine PHEVs? Why or why not? What vehicle trucks, if applicable. power between the first motor/generator segments may turbo PHEVs best be and the drive motor to either charge the suited for, and which segments would it (3) Integrated Motor Assist (IMA)/Crank battery or supply power to the wheels. Integrated Starter Generator not be best suited for? What vehicle The power-split hybrid technology is segments may high compression ratio Integrated Motor Assist (IMA)/Crank included in this analysis as an enabling PHEVs best be suited for, and which integrated starter generator (CISG) technology supporting this proposal, segments would it not be best suited provides idle-stop capability and uses a (the agencies evaluate the P2 hybrid for? Similarly, the analysis currently high voltage battery with increased technology discussed above where considers PHEVs with 30-mile and 50- energy capacity over typical automotive power-split hybrids might otherwise mile all-electric range, and should other batteries. The higher system voltage have been appropriate). As stated above, ranges be considered? For instance, a allows the use of a smaller, more battery costs are now considered 20-mile all-electic range may decrease powerful electric motor and reduces the separately from other HEV hardware. the battery pack size, and hence the weight of the wiring harness. This Power-split hybrid technology as battery pack cost relative to a 30-mile system replaces a standard alternator modeled in this analysis is not suitable all-electric range system, while still with an enhanced power, higher for large vehicles that must handle high providing electric-vehicle functionality voltage, higher efficiency starter loads. in many applications. Do commenters alternator that is crankshaft-mounted The ANL Autonomie simulations believe PHEV technology will see and can recover braking energy while assumed all power-split hybrids use a widespread adoption in the US vehicle the vehicle slows down (regenerative high compression ratio engine. fleet? Why or why not? Please provide braking). Therefore, all vehicles equipped with supporting data. SHEVPS technology in the CAFE model (4) P2 Hybrid inputs and simulations are assumed to (b) Full Electrification and Fuel Cell P2 Hybrid (SHEVP2) is a newly have high compression ratio engines. Vehicles emerging hybrid technology that uses a transmission-integrated electric motor (6) Plug-in Hybrid Electric (1) Battery Electric placed between the engine and a Plug-in hybrid electric vehicles Electric vehicles (EV) are equipped gearbox or CVT, much like the IMA (PHEV) are hybrid electric vehicles with with all-electric drive and with systems system described above except with a the means to charge their battery packs powered by energy-optimized batteries wet or dry separation clutch that is used from an outside source of electricity charged primarily from grid electricity. to decouple the motor/transmission (usually the electric grid). These EVs with range of 200 miles have been from the engine. In addition, a P2 vehicles have larger battery packs with included as a potential technology in hybrid would typically be equipped more energy storage and a greater this NPRM. Battery costs are now with a larger electric machine. capability to be discharged than other considered separately from other EV Disengaging the clutch allows all- hybrid electric vehicles. They also use hardware.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00065 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43050 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(2) Fuel Cell Electric (c) Electric Vehicle Infrastructure and charge convenience is not taken BEVs and PHEVs may be charged at into account in the CAFE model. Also, Fuel cell electric vehicles (FCEVs) today’s analysis assumes HEVs, PHEVs, utilize a full electric drive platform but home or elsewhere. Home chargers may access electricity from a regular wall and BEVs have the same survival rates consume electricity generated by an outlet, or they may require special and mileage accumulation schedules as onboard fuel cell and hydrogen fuel. equipment to be installed at the home. vehicles with conventional powertrains, Fuel cells are electrochemical devices Commercial chargers may sometimes be and that HEVs, PHEVs, and BEVs never that directly convert reactants (hydrogen found at businesses or other travel receive replacement batteries before and oxygen via air) into electricity, with locations. These chargers often may being scrapped. The agencies invite the potential of achieving more than supply power to the vehicle at a faster comment on these assumptions and on twice the efficiency of conventional rate of charge but often require data and practicable methods to internal combustion engines. High significant capital investment to install. implement any alternatives. pressure gaseous hydrogen storage tanks Time to charge, and availability and are used by most automakers for FCEVs. convenience of charging are significant 9. Accessory Technologies The high pressure tanks are similar to factors for plug-in vehicle operators. For Two accessory technologies, electric those used for compressed gas storage in many consumers, accessible charging power steering (EPS) and improved more than 10 million CNG vehicles stations present inconveniences that accessories (IACC) (accessory worldwide, except that they are may deter the adoption of battery technologies categorized for the 2012 designed to operate at a higher pressure electric and plug-in hybrid vehicles. rule) were considered in this analysis, (350 bar or 700 bar vs. 250 bar for CNG). More detail about charging and and are described below.167 Table II–17 FCEVs are currently produced in charging infrastructure, including a discussion of potential electric vehicle and Table II–18 below shows the limited numbers and are available in impacts on the electric grid, is available estimated absolute costs including limited geographic areas. in the PRIA, Chapter 6. For today’s learning effects and retail price analysis, costs for installing chargers equivalent utilized in today’s analysis.

(a) Electric Power Steering (EPS) because it replaces a continuously types of light-duty vehicles, including operated hydraulic pump, thereby large trucks. However, this analysis Electric power steering (EPS)/ reducing parasitic losses from the applies the EHPS technology to large Electrohydraulic power steering (EHPS) accessory drive. Manufacturers have trucks and the EPS technology to all is an electrically-assisted steering informed the agencies that full EPS other light-duty vehicles. system that has advantages over systems are being developed for all traditional hydraulic power steering

167 For further discussion of accessory technologies, see Chapter 6 of the PRIA accompanying this NPRM.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00066 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.032 EP24AU18.033 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43051

(b) Improved Accessories (IACC) supercharger for engine breathing. The (b) Electrified Vehicle Powertrain Improved accessories (IACC) may efficiency may be significantly higher • Advanced battery chemistries— include high efficiency alternators, than MY 2016 turbocharged gasoline many emerging battery technologies electrically driven (i.e., on-demand) engines with competitive costs. This promise to eventually improve the cost, water pumps, variable geometry oil engine architecture could run on many safety, charging time, and durability in pumps, cooling fans, a mild fuels, including gasoline and diesel. comparison to the MY 2016 automotive regeneration strategy, and high Packaging constraints, emissions lithium-ion batteries. For instance, efficiency alternators. It excludes other compliance, and performance across a many view solid state batteries as a electrical accessories such as electric oil wide range of operating conditions promising medium-term automotive pumps and electrically driven air remain as open considerations. No technology. Solid state batteries replace conditioner compressors. In the MY production vehicles have been publicly the battery’s liquid electrolyte with a 2017–2025 final rule, two levels of IACC announced, and multiple manufacturers solid electrolyte to potentially improve were offered as a technology path (a low continue to evaluate the safety, power and energy density, improvement level and a high technology.169 170 durability, and cost. Some variations • improvement level). Since much of the Dual-fuel—engine concepts such as use ceramic, polymer, or sulfide-based market has incorporated some of these reactivity controlled compression solid electrolytes. Multiple automakers technologies in the MY 2016 fleet, the ignition (RCCI) combine multiple fuels and suppliers are exploring the analysis assumes all vehicles have (e.g. gasoline and diesel) in cylinder to technology and possible incorporated what was previously the improve brake thermal efficiency while commercialization that may occur in the low level, so only the high level remains reducing NOX and particulate early 2020s.175 176 177 as an option for some vehicles. emissions. This technology is still in the • Supercapacitors/Ultracapacitors— research phase.171 An electrical energy storage device with 10. Other Technologies Considered but • Smart accessory technologies—can Not Included in This Aanalysis higher power density but lower energy improve fuel efficiency through smarter density than batteries. Advanced Manufacturers, suppliers, and controls of existing systems given capacitors may reduce battery researchers continue to create a diverse imminent or expected controls inputs in degradation associated with charge and set of fuel economy technologies. Many real world driving conditions. For discharge cycles, with some tradeoffs to high potential technologies that are still instance, a vehicle could adjust the use cost and engineering complexity. in the early stages of the development of accessory systems to conserve energy Supercapacitors/Ultracapacitors are and commercialization process are still and fuel as a vehicle approaches a red currently not used in parallel or as a being evaluated for any final analysis. light. Vehicle connectivity and sensors standalone traction motor energy storage Due to uncertainties in the cost and can further refine the operation for more device.178 capabilities of emerging technologies, benefit and smoother operation.172 • Motor/Drivetrain: some new and pre-production • High Compression Miller Cycle Æ Lower-cost magnets for Brushless technologies are not yet a part of the Engine with Variable Geometry Direct Current (BLDC) motors—BLDC CAFE model simulation. Evaluating and Turbocharger or Electric Supercharger— motor technology, common in hybrid benchmarking promising fuel economy Atkinson cycle gasoline engines with and battery electric vehicles, uses rare technologies continues to be a priority sophisticated forced induction system earth magnets. By substituting and as commercial development matures. that requires advanced controls. The eliminating rare earths from the benefits of these technologies provide (a) Engine Technologies magnets, motor cost can be significantly better control of EGR rates and boost • reduced. This technology is announced, Variable compression ratio (VCR)— which is achieved with electronically but not yet in production. The varies the compression ratio and swept controlled turbocharger or supercharger. capability and material configuration of volume by changing the piston stroke on The electric version of this technology these systems remains a closely guarded all cylinders. Manufacturers accomplish which incorporates 48V is called E- trade secret.179 this by changing the effective length of boost.173 174 the piston connecting rod, with some maps were not available in time for the NPRM prototypes having a range of 8:1 to 14:1 169 Murphy, T. Achates: Opposed-Piston Engine analysis. Please see Chapter 6.3 of the PRIA compression ratio. In turbocharged makers tooling up, Wards Auto (Jan. 23, 2017), accompanying this proposal for more information. form, early publications suggest VCR http://wardsauto.com/engines/achates-opposed- 175 Schmitt, B. Ultrafast-Charging Solid-State EV may be possible to deliver up to 35% piston-engine-makers-tooling. Batteries Around The Corner, Toyota Confirms, 170 Our Formula, Achates Power, http:// Forbes (Jul. 25, 2017), https://www.forbes.com/ improved efficiency over the existing achatespower.com/our-formula/opposed-piston/ sites/bertelschmitt/2017/07/25/ultrafast-charging- equivalent-output naturally-aspirated (last visited June 21, 2018). solid-state-ev-batteries-around-the-corner-toyota- engine.168 171 Robert Wagner, Enabling the Next Generation confirms/#5736630244bb. • Opposed-piston engine—sometimes of High Efficiency Engines, Oak Ridge National 176 Moving toward clean mobility, Robert Bosch known as opposed-piston opposed- Laboratory, U.S. Department of Energy (2012), GmbH, https://www.bosch.com/explore-and- available at https://www.energy.gov/sites/prod/ experience/moving-toward-clean-mobility/ (last cylinder (OPOC), operates with two files/2014/03/f8/deer12_wagner_0.pdf. visited June 21, 2018). pistons per cylinder working in 172 EfficientDynamics—The intelligent route to 177 Reuters Staff, Honda considers developing all opposite reciprocal motion and running lower emissions, BMW Group (2007), https:// solid-state EV batteries, Reuters (Dec. 21, 2017), on a two-stroke combustion cycle. It has www.bmwgroup.com/content/dam/bmw-group- https://www.reuters.com/article/us-honda-nissan/ websites/bmwgroup_com/responsibility/downloads/ honda-considers-developing-all-solid-state-ev- no cylinder head or valvetrain but en/2007/Alex_ED__englische_Version.pdf. batteries-idUSKBN1EF0FM. requires a turbocharger and 173 Volkswagen at the 37th Vienna Motor 178 Burke, A. & Zhao,H. Applications of Symposium, Volkswagen (Apr. 28, 2016), https:// Supercapacitors in Electric and Hybrid Vehicles, 168 See e.g., VC—Turbo—The world’s first www.volkswagen-media-services.com/en/ Institute of Transportation Studies University of production-ready variable compression ratio detailpage/-/detail/Volkswagen-at-the-37th-Vienna- California, Davis (Apr. 2015), available at https:// engine, Nissan Motor Corporation (Dec. 13, 2017), Motor-Symposium/view/3451577/ steps.ucdavis.edu/wp-content/uploads/2017/05/ https://newsroom.nissan-global.com/releases/ 5f5a4dcc90111ee56bcca439f2dcc518?p_p_ 2015-UCD-ITS-RR-15-09-1.pdf. release-917079cb4af478a2d26bf8e5ac00ae49-vc- auth=M2yfP3Ze. 179 Buckland, K. & Sano, N. Toyota Readies turbo-the-worlds-first-production-ready-variable- 174 These engines may be considered in the Cheaper Electric Motor by Halving Rare Earth Use, compression-ratio-engine. analysis supporting the final rule, but these engine Continued

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00067 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43052 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Æ Integrated multi-phase integrated and high cost.185 These systems are not about which off-cycle technologies are electric vehicle drivetrains. Research yet in production. present on which vehicles (see Section has been conducted on 6-phase and 9- • Suspension energy recovery— X for further discussion of reporting off- phase integrated systems to potentially Multiple electromechanical and cycle technology information). reduce cost and improve power density. electrohydraulic suspension The 2012 final rule for MYs 2017 and Manufacturers may improve system technologies exist that can convert later outlined two test procedures to power density through integration of the motion from uneven roads into determine credit or FCIV eligibility for motor, inverter, control, and gearing. electricity.186 187 These technologies are A/C efficiency menu credits, the idle These systems are in the research limited to luxury vehicles with limited test, and the AC17 test. The idle test, phase.180 181 production volumes. This technology is performed while the vehicle is at idle, not produced in 2016 but planned for determined the additional CO2 (c) Transmission Technologies production as early as 2018.188 generated at idle when the A/C system is operated.191 The AC17 test is a four- • Beltless CVT—Most MY 2016, 11. Air Conditioning Efficiency and Off- part performance test that combines the commercially available CVTs rely on Cycle Technologies belt technology. A new architecture of existing SC03 driving cycle, the fuel (a) Air Conditioning Efficiency CVT replaces belts or pulleys with a economy highway test cycle, and a pre- Technologies conditioning cycle, and solar soak continuously variable variator, which is 192 a special type of planetary set with balls Air conditioning (A/C) is a virtually period. Manufacturers could use the and rings instead of gears. The standard automotive accessory, with idle test or AC17 test to determine technology promises to improve more than 95% of new cars and light improvement values for MYs 2014– efficiency, handle higher torques, and trucks sold in the United States 2016, while for MYs 2017 and later, the change modes more quickly. This equipped with mobile air conditioning AC17 test is the exclusive test that manufacturers can use to demonstrate technology may be commercially (MAC) systems. Most of the additional eligibility for menu A/C improvement available as early as 2020.182 air conditioning related load on an engine is due to the compressor, which values. • Multi-speed electric motor pumps the refrigerant around the system In MYs 2020 and later, manufacturers transmission—MY 2016 battery electric loop. The less the compressor operates will use the AC17 test to demonstrate vehicle transmissions are single-speed. or the more efficiently it operates, the eligibility for A/C credits and to Multiple gears can allow for more less load the compressor places on the partially quantify the amount of the torque multiplication at lower speeds or engine, and the better fuel consumption credit earned. AC17 test results equal to a downsized electric machine, increased will be. This high penetration means A/ or greater than the menu value will efficiency, and higher top speed. Two- C systems can significantly impact allow manufacturers to claim the full speed transmission designs are available energy consumed by the light duty menu value for the credit. A test result 183 but not currently in production. vehicle fleet. less than the menu value will limit the (d) Energy-Harvesting Technology Vehicle manufacturers can generate amount of credit to that demonstrated credits for improved A/C systems under on the AC17 test. In addition, for MYs • Vehicle waste heat recovery EPA’s GHG program and receive a fuel 2017 and beyond, A/C fuel consumption systems—Internal combustion engines consumption improvement value (FCIV) improvement values will be available convert the majority of the fuel’s energy equal to the value of the benefit not for CAFE calculations, whereas to heat. Thermoelectric generators and captured on the 2-cycle test under efficiency credits were previously only heat pipes can convert this heat to NHTSA’s CAFE program.189 Table II–19 available for GHG compliance. The electricity.184 Thermoelectric provides a ‘‘menu’’ of qualifying A/C agencies proposed these changes in the generators, often made of technologies, with the magnitude of 2012 final rule for MYs 2017 and later semiconductors, have been tested by each improvement value or credit largely as a result of new data collected, automakers but have traditionally not estimated based on the expected as well as the extensive technical 193 been implemented due to low efficiency reduction in CO2 emissions from the comments submitted on the proposal. technology.190 NHTSA converts the The pre-defined technology menu and Bloomberg (Feb, 20, 2018), https:// improvement in grams per mile to a associated car and light truck credit www.bloomberg.com/news/articles/2018-02-20/ FCIV for each vehicle for purposes of value is shown in Table II–19 below. toyota-readies-cheaper-electric-motor-by-halving- measuring CAFE compliance. As part of The regulations include a definition of rare-earth-use. a manufacturer’s compliance data, each technology that must be met to be 180 Burkhardt, Y., Spagnolo, A., Lucas, P., eligible for the menu credit.194 Zavesky, M., & Brockerhoff, P. ‘‘Design and analysis manufacturers will provide information of a highly integrated 9-phase drivetrain for EV Manufacturers are not required to applications ’’ 20 November 2014. DOI. 10.1109/ 185 Patel, P. Powering Your Car with Waste Heat, submit any other emissions data or ICELMACH.2014.6960219. IEEE xplore. MIT Technology Review (May 25, 2011), https:// information beyond meeting the 181 Patel, V., Wang, J., Nugraha, D., Vuletic, R., & www.technologyreview.com/s/424092/powering- definition and useful life Tousen, J. ‘‘Enhanced Availability of Drivetrain your-car-with-waste-heat/. requirements 195 to use the pre-defined Through Novel Multi-Phase Permanent Magnet 186 Baeuml, B. et al., The Chassis of the Future, Machine Drive’’ 2016. IEEE Transactions on Schaeffler, https://www.schaeffler.com/ Industrial Electronics Pages. 469–480. remotemedien/media/_shared_media/08_media_ 191 75 FR 25324, 25431 (May 7, 2010). The A/C 182 Murphy, T. Planets Aligning for Dana’s library/01_publications/schaeffler_2/symposia_1/ CO2 Idle Test is run with and without the A/C VariGlide Beltless CVT, Wards Auto (Aug. 22, downloads_11/Schaeffler_Kolloquium_2014_27_ system cooling the interior cabin while the vehicle’s 2017), http://wardsauto.com/technology/planets- en.pdf (last visited June 21, 2018). engine is operating at idle and with the system aligning-dana-s-variglide-beltless-cvt. 187 Advanced Suspension, Tenneco, http:// under complete control of the engine and climate 183 Faid, S. A Highly Efficient Two Speed www.tenneco.com/overview/rc_advanced_ control system. Transmission for Electric Vehicles (May 2015), suspension/ (last visited June 21, 2018). 192 77 FR 62624, 62723 (Oct. 15, 2012). available at http://www.evs28.org/event_file/event_ 188 Audi A8 Active Chassis, Audi, https:// 193 Id. file/1/pfile/EVS28_Saphir_Faid.pdf. www.audi.com/en/innovation/design/more_ 194 Id. at 62725. 184 Orr et al., A review of car waste heat recovery personal_comfort_a8_active_chassis.html (last 195 Lifetime vehicle miles travelled (VMT) for MY systems utilising thermoelectric generators and heat visited June 21, 2018). 2017–2025 are 195,264 miles and 225,865 miles for pipes, 101 Applied Thermal Engineering 490–495 189 77 FR 62624, 62720 (Oct. 15, 2012). passenger cars and light trucks, respectively. The (May 25, 2016). 190 40 CFR 86.1868–12 (2016). manufacturer must also demonstrate that the off-

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00068 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43053

credit value. Manufacturers’ use of of the standards.197 Many manufacturers amounted to more than 12 million Mg menu-based credits for A/C efficiency is have since incorporated A/C technology of fuel consumption improvement subject to a regulatory cap: 5.7 g/mi for throughout their fleets, and the values of the total net fuel consumption cars and trucks through MY 2016 and utilization of advanced A/C improvement values reported. This is separate caps of 5.0 g/mi for cars and technologies has become a significant equivalent to approximately four grams 7.2g/mi for trucks for later MYs.196 contributor to industry compliance per mile across the 2016 fleet. plans. As summarized in the EPA Accordingly, a significant amount of In the 2012 final rule for MYs 2017 Manufacturer Performance Report for new information about A/C technology and later, the agencies estimated that the 2016 model year,198 15 auto and the efficacy of test procedures has manufacturers would employ significant manufacturers included A/C efficiency become available since the 2012 final advanced A/C technologies throughout credits as part of their compliance rule. their fleets to improve fuel economy, demonstration in the 2016 MY. These The sections below provide a brief and this was reflected in the stringency history of the AC17 test procedure for 197 See e.g., 77 FR 62623, 62803–62806 (Oct. 15, evaluating A/C efficiency improving cycle technology is effective for the full useful life 2012). 198 technology and discuss stakeholder of the vehicle. Unless the manufacturer See Greenhouse Gas Emission Standards for Light-Duty Vehicles: Manufacturer Performance comments on the AC17 test procedure demonstrates that the technology is not subject to Report for the 2016 Model Year (EPA Report 420– in-use deterioration, the manufacturer must account approach and discuss A/C efficiency R18–002), U.S. EPA (Jan. 2018), available at https:// technology valuation through the off- for the deterioration in their analysis. nepis.epa.gov/Exe/ZyPDF.cgi?Dockey= 196 40 CFR 86.1868–12(b)(2) (2016). P100TGIA.pdf. cycle program.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00069 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43054 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(1) Evaluation of the AC17 Test technology and the other does not.199 than during just idle (as measured in the Procedure Since the Draft TAR The agencies believe that the AC17 test old idle test procedure). As described procedure is more capable of detecting above and in the 2012 final rule,200 the In developing the AC17 test the effect of more efficient A/C AC17 test is a four-part performance test procedure, the agencies sought to components and controls strategies that combines the existing SC03 driving develop a test procedure that could during a transient drive cycle rather more reliably generate an appropriate 200 See 77 FR 62624, 62723 (Oct. 15, 2012); Joint fuel consumption improvement value Technical Support Document: Final Rulemaking for 199 based on an ‘‘A’’ to ‘‘B’’ comparison, For an explanation of how the agencies, in 2017–2025 Light-Duty Vehicle Greenhouse Gas collaboration with stakeholders, developed the that is, a comparison of substantially Emission and Corporate Average Fuel Economy AC17 test procedure, see the 2017 and later final Standards, U.S. EPA, National Highway Traffic similar vehicles in which one has the rule at 77 FR 62624, 62723 (Oct. 15, 2012). Safety Administration at 5–40 (August 2012) .

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00070 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.034 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43055

cycle, the fuel economy highway cycle, The agencies disagree this makes the Similarly, the Alliance commented that, as well as a pre-conditioning cycle, and test unworkable. The regulation ‘‘[t]he future success of the MAC credit a solar soak period. describes the baseline vehicle as a program in generating emissions The agencies received several ‘‘similar’’ vehicle, selected with good reductions will depend to a large extent comments on the Draft TAR evaluation engineering judgment (such that the test on the manner in which it is of the AC17 test procedure. FCA comparison is not unduly affected by administered by EPA, especially with commented generally that A/C other differences). Also, OEMs respect to making the AC17 A-to-B efficiency technologies ‘‘are not expressed confidence in using A-to-B provisions function smoothly, without showing their full effect on this AC17 testing to qualify for fuel consumption becoming a prohibitive obstacle to fully test as most technologies provide benefit improvement values for software-based achieving the MAC indirect credits.’’ 204 at different temperatures and humidity A/C efficiency technologies. While As described in the Draft TAR, in conditions in comparison to a standard hardware technologies may pose a 2016, USCAR members initiated a test conditions. All of these technologies greater challenge in locating a Cooperative Research Program (CRP) are effective at different levels at sufficiently similar ‘‘A’’ baseline through the Society of Automotive different conditions. So there is not one vehicle, the engineering analysis Engineers (SAE) to develop bench size fits all in this very complex testing provision under 40 CFR 86.1868– testing standards for the four hardware approach. Selecting one test that 12(g)(2) provides an alternative to technologies in the fuel consumption captures benefits of all of these locating and performing an AC17 test on 201 improvement value menu (blower motor conditions has not been possible.’’ such a vehicle. Further, as the USCAR The agencies acknowledge that any control, internal heat exchanger, program in general and the GM Denso single test procedure is unlikely to improved evaporators and condensers, SAS compressor application specifically equally capture the real-world effect of and oil separator). The intent of the have shown, the test is able to resolve every potential technology in every program is to streamline the process of small differences in CO effectiveness potential use case. Both the agencies 2 conducting bench testing and (1.3 grams in the latter case) when and stakeholders understood this engineering analysis in support of an carefully conducted. difficulty when developing the AC17 application for A/C credits under 40 test procedure. While no test is perfect, Commenters on the Draft TAR also CFR part 86.1868–12(g)(2), by creating the AC17 test procedure represents an expressed a desire for improvements in uniform standards for bench testing and industry best effort at identifying a test the process by which manufacturers for establishing the expected GHG effect that would greatly improve upon the without an ‘‘A’’ vehicle (for the A-to-B of the technology in a vehicle idle test by capturing a greater range of comparison) could apply under the application. operating conditions. General industry engineering analysis provision, such as An update to the list of SAE standards evaluation of the AC17 test procedure is development of standardized under development originally presented in agreement that the test achieves this engineering analysis and bench testing in the Draft TAR is listed in Table II– objective. procedures that could support such 20. Since completion of the Draft TAR, FCA also noted that ‘‘[i]t is a major applications. For example, Toyota work has continued on these standards, problem to find a baseline vehicle that requested that ‘‘EPA consider an which appear to be nearing completion. is identical to the new vehicle but optional method for validation via an The agencies seek comment with the without the new A/C technology. This engineering analysis, as is currently latest completion of these SAE alone makes the test unworkable.’’ 202 being developed by industry.’’ 203 standards.

(2) A/C Efficiency Technology Valuation by utilizing technologies on the menu; also requires ‘‘A-to-B’’ comparison Through the Off-Cycle Program however, the agencies recognize that testing under the AC17 test, that is, manufacturers will develop additional testing substantially similar vehicles in The A/C technology menu, discussed which one has the technology and the at length above, includes several A/C technologies that are not currently listed efficiency-improving technologies that on the menu. These additional A/C other does not. were well defined and had been efficiency-improving technologies are To date, the agencies have received quantified for effectiveness at the time eligible for fuel consumption one type off-cycle application for an A/ of the 2012 final rule for MYs 2017 and improvement values on a case-by-case C efficiency technology. In December beyond. Manufacturers claimed the vast basis under the off-cycle program. 2014, General Motors submitted an off- majority of A/C efficiency credits to date Approval under the off-cycle program cycle application for the Denso SAS A/

201 See Comment by FCA US LLC, Docket ID 203 See Comment by Toyota (revised), Docket ID 204 See Comment by Alliance of Automobile NHTSA 2016–0068–0082, at 123–124. NHTSA–2016–0068–0088, at 23. Manufacturers, Docket ID EPA–HQ–OAR–2015– 202 Id. at 124. 0827–4089 and NHTSA–2016–0068–0072, at 160.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00071 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.035 43056 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

C compressor with variable crankcase that new A/C technologies not currently values and are not part of the off-cycle suction valve technology, requesting an on the menu will emerge over the time regulation (40 CFR 86.1869–12). off-cycle GHG credit of 1.1 grams CO2 frame of the MY 2021–2026 standards. However, it should be noted that off- per mile. In December 2017, BMW of This proposal requests comment on cycle applications submitted via the North America, Ford Motor Company, adding one additional A/C technology public process pathway are decided Hyundai Motor Company, and Toyota to the menu—the A/C compressor with individually on merits through a petitioned and received approval to variable crankcase suction valve process involving public notice and receive the off-cycle improvement value technology, discussed below (and also opportunity for comment. In deciding for the same A/C efficiency one off-cycle technology, discussed whether to approve or deny a request, technology.205 206 EPA, in consultation below). The agencies also request the agencies may take into account any with NHTSA, evaluated the applications comment on whether to change any fuel factors deemed relevant, including such and found methodologies described economy improvement values currently issues as the realization of claimed fuel therein were sound and appropriate.207 assigned to technologies on the menu. consumption improvement value in Accordingly, the agencies approved the Next, as mentioned above, the menu- real-world use. Such considerations fuel economy improvement value based improvement values for A/C could include synergies or interactions applications. efficiency established in the 2012 final among applied technologies, which The agencies received additional rule for MYs 2017 and by end are could potentially be addressed by stakeholder comments on the off-cycle subject to a regulatory cap. The rule set application of some form of cap or other approval process as an alternate route to a cap of 5.7 g/mi for cars and trucks applicable limit, if warranted. receiving A/C technology credit values. through MY 2016 and separate caps of Therefore, applying for A/C efficiency The Alliance requested that EPA 5.0 g/mi for cars and 7.2g/mi for trucks fuel consumption improvement values ‘‘simplify and standardize the for later MYs.210 Several commenters through the off-cycle provisions in 40 procedures for claiming off-cycle credits asked EPA to reconsider the CFR 86.1869–12 should not be seen as for the new MAC technologies that have applicability of the cap to non-menu A/ a route to unlimited A/C fuel been developed since the creation of the C efficiency technologies claimed consumption improvement values. The MAC indirect credit menu.’’ 208 Other through the off-cycle process and agencies discuss air conditioning commenters noted the importance of questioned the applicability of this cap efficiency improvement values further continuing to incentivize further on several different grounds. These in Section X of this NPRM. innovation in A/C efficiency comments appear to be in response to a technologies as new technologies Draft TAR passage that stated: (b) Off-Cycle Technologies emerge that are not listed on the menu ‘‘Applications for A/C efficiency credits ‘‘Off-cycle’’ emission reductions and or when manufacturers begin to reach made under the off-cycle credit program fuel consumption improvements can be regulatory caps. The commenters rather than the A/C credit program will achieved by employing off-cycle suggested that EPA should consider continue to be subject to the A/C technologies resulting in real-world adding new A/C efficiency technologies efficiency credit cap’’ (Draft TAR, p. 5– benefits but where that benefit is not to the menu and/or update the fuel 210). The agencies considered these adequately captured on the test consumption improvement values for comments and present clarification procedures used to demonstrate technology already listed on the menu, below. As additional context, the 2012 compliance with fuel economy emission particularly in cases where TSD states: standards. EPA initially included off- manufacturers can show through an off- ‘‘. . . air conditioner efficiency is an off- cycle technology credits in the MY cycle application that the technology cycle technology. It is thus appropriate [. . .] 2012–2016 rule and revised the program actually deserves more credit than that to employ the standard off-cycle credit in the MY 2017–2025 rule.212 NHTSA listed on the menu. For example, Toyota approval process [to pursue a larger credit adopted equivalent off-cycle fuel commented that ‘‘the incentive values than the menu value]. Utilization of bench consumption improvement values for for A/C efficiency should be updated tests in combination with dynamometer tests MYs 2017 and later in the MY 2017– and simulations [. . .] would be an 213 along with including new technologies appropriate alternate method of 2025 rule. being deployed.’’ 209 demonstrating and quantifying technology Manufacturers can demonstrate the The agencies note that some of these credits (up to the maximum level of credits value of off-cycle technologies in three comments are directed towards the off- allowed for A/C efficiency) [emphasis added]. ways: First, they may select fuel cycle technology approval process A manufacturer can choose this method even economy improvement values and CO2 generally, which is described in more for technologies that are not currently credit values from a pre-defined 211 detail in Section X of this preamble. included in the menu.’’ ‘‘menu’’ for off-cycle technologies that This suggests the concept of placing a Regarding the A/C technology menu meet certain regulatory specifications. limit on total A/C fuel consumption specifically, the agencies do anticipate As part of a manufacturer’s compliance improvement values, even when some data, manufacturers will provide are granted under the off-cycle program, 205 EPA Decision Document: Off-Cycle Credits for information about which off-cycle is not entirely new and that EPA BMW Group, Ford Motor Company, and Hyundai technologies are present on which Motor Company, U.S. EPA (Dec. 2017), available at considered the menu cap as being vehicles. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey= appropriate at the time. The pre-defined list of technologies P100TF06.pdf. A/C regulatory caps specified under 206 Alternative Method for Calculating Off-cycle and associated off-cycle light-duty 40 CFR 86.1868–12(b)(2) apply to A/C Credits under the Light-Duty Vehicle Greenhouse vehicle fuel economy improvement Gas Emissions Program: Applications from General efficiency menu-based improvement Motors and Toyota Motor North America, 83 FR values and GHG credits is shown in 214 8262 (Feb. 26, 2018). 210 40 C.F.R § 86.1868–12(b)(2) (2016). Table II–21 and Table II–22 below. A 207 Id. 211 Joint Technical Support Document: Final 208 Comment by Alliance of Automobile Rulemaking for 2017–2025 Light-Duty Vehicle 212 77 FR 62624, 62832 (Oct. 15, 2012). Manufacturers, Docket ID EPA–HQ–OAR–2015– Greenhouse Gas Emission and Corporate Average 213 Id. at 62839. 0827–4089 and NHTSA–2016–0068–0072, at 152. Fuel Economy Standards, U.S. EPA, National 214 For a description of each technology and the 209 Comment by Toyota (revised), Docket ID Highway Traffic Safety Administration at 5–58 derivation of the pre-defined credit levels, see NHTSA–2016–0068–0088, at 23. (August 2012). Chapter 5 of the Joint Technical Support Document:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00072 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43057

definition of each technology equipment to submit any other emissions data or Credits based on the pre-defined list are must meet to be eligible for the menu information beyond meeting the subject to an annual manufacturer fleet- credit is included at 40 CFR 86.1869– definition and useful life requirements wide cap of 10 g/mile. 12(b)(4). Manufacturers are not required to use the pre-defined credit value.

Manufacturers can also perform their cold temperature, enabling use an alternative methodology for own 5-cycle testing and submit test improvements to be measured for determining the value of an off-cycle results to the agencies with a request technologies that do not impact technology. This option is only explaining the off-cycle technology. The operation on the 2-cycle tests. Credits available if the benefit of the technology additional three test cycles have determined according to this cannot be adequately demonstrated different operating conditions including methodology do not undergo public using the 5-cycle methodology. high speeds, rapid accelerations, high review. Manufacturers may also use this option temperature with A/C operation and The third pathway allows to demonstrate reductions that exceed manufacturers to seek EPA approval to

Final Rulemaking for 2017–2025 Light-Duty Vehicle Fuel Economy Standards, U.S. EPA, National Highway Traffic Safety Administration (August Greenhouse Gas Emission and Corporate Average 2012).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00073 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.036 43058 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

those available via use of the determine these off-cycle credits in credit values, manufacturers have yet to predetermined menu list. The September 2014.215 Subsequently, FCA, report credits to EPA based on those manufacturer must also demonstrate Ford, and GM requested off-cycle alternative methodologies. that the off-cycle technology is effective credits using this same methodology. As discussed below, all three methods for the full useful life of the vehicle. FCA and Ford submitted applications have been used by manufacturers to Unless the manufacturer demonstrates for off-cycle credits from high efficiency generate off-cycle improvement values that the technology is not subject to in- exterior lighting, solar reflective glass/ and credits. use deterioration, the manufacturer glazing, solar reflective paint, and active must account for the deterioration in seat ventilation. Ford’s application also (1) Use of Off-Cycle Technologies to their analysis. demonstrated off-cycle benefits from Date Manufacturers must develop a active aerodynamic improvements methodology for demonstrating the Manufacturers used a wide array of (grille shutters), active transmission off-cycle technologies in MY 2016 to benefit of the off-cycle technology, and warm-up, active engine warm-up EPA makes the methodology available generate off-cycle GHG credits using the technologies, and engine idle stop-start. pre-defined menu. Table II–23 below for public comment prior to an EPA GM’s application described real-world determination, in consultation with shows the percent of each benefits of an air conditioning manufacturer’s production volume NHTSA, on whether to allow the use of compressor with variable crankcase the methodology to measure using each menu technology reported to suction valve technology. EPA approved EPA for MY 2016 by manufacturer. improvements. The data needed for this the credits for FCA, Ford, and GM in demonstration may be extensive. Table II–24 shows the g/mile benefit September 2015.216 Note, however, that Several manufacturers have requested each manufacturer reported across its and been granted use of alternative test although EPA granted the use of fleet from each off-cycle technology. methodologies for measuring alternative methodologies to determine Like Table II–23, Table II–24 provides improvements. In 2013, Mercedes the mix of technologies used in MY requested off-cycle credits for the 215 EPA Decision Document: Mercedes-Benz Off- 2016 by manufacturer and the extent to cycle Credits for MYs 2012–2016, U.S. EPA (Sept. which each technology benefits each following off-cycle technologies in use 2014), available at https://nepis.epa.gov/Exe/ or planned for implementation in the ZyPDF.cgi/P100KB8U.PDF?Dockey= manufacturer’s fleet. Fuel consumption 2012–2016 model years: Stop-start P100KB8U.PDF. improvement values for off-cycle systems, high-efficiency lighting, 216 EPA Decision Document: Off-cycle Credits for technologies were not available in the Fiat Chrysler Automobiles, Ford Motor Company, CAFE program until MY 2017; therefore, infrared glass glazing, and active seat and General Motors Corporation, U.S. EPA (Sept. ventilation. EPA approved 2015), available at https://nepis.epa.gov/Exe/ only GHG off-cycle credits have been methodologies for Mercedes to ZyPDF.cgi/P100N19E.PDF?Dockey=P100N19E.PDF. generated by manufacturers thus far.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00074 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43059

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00075 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.037 43060 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

In 2016, manufacturers generated the was the first year that manufacturers Rover generated the most off-cycle vast majority of credits using the pre- could generate credits using pre-defined credits on a fleet-wide basis, reporting defined menu.217 Although MY 2014 menu values, manufacturers have acted credits equivalent to approximately 6 g/ quickly to generate substantial off-cycle mile and 5 g/mile, respectively. Several 217 Thus far, the agencies have only granted one improvements. FCA and Jaguar Land other manufacturers report fleet-wide manufacturer (GM) off-cycle credits for technology credits in the range of approximately 1 based on 5-cycle testing. These credits are for an off-cycle technology used on certain GM gasoline- weather while the vehicle is stopped and the engine to 4 g/mile. In MY 2016, the fleet total electric hybrid vehicles, an auxiliary electric pump, is off, thus allowing the engine stop-start system to across manufacturers equaled which keeps engine coolant circulating in cold be active more frequently in cold weather. approximately 2.5 g/mile. The agencies

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00076 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.038 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43061

expect that as manufacturers continue extent that it clarifies the benefits and Note: They include costs which are expanding their use of off-cycle costs of the proposed action’s impacts transfers between different economic technologies, the fleet-wide effects will on car and light truck producers, actors—these will appear as both a cost continue to grow with some illustrates how these are transmitted to and a benefit in equal amounts (to manufacturers potentially approaching buyers of new vehicles, shows the separate affected parties). Societal cost the 10 g/mile fleet-wide cap. action’s collateral economic effects on and benefit values shown elsewhere in E. Development of Economic owners of used cars and light trucks, this document do not show costs which Assumptions and Information Used as and identifies how these impacts create are transfers for the sake of simplicity Inputs to the Analysis costs and benefits for the remainder of but report the same net societal costs the U.S. economy. It will achieve the and benefits. As it indicates, the 1. Purpose of Developing Economic second objective by showing clearly proposed action first reduces costs to Assumptions for Use in Modeling how the economy-wide or ‘‘social’’ manufacturers for adding technology Analysis benefits and costs of the proposed necessary to enable new cars and light (a) Overall Framework of Costs and action are composed of its direct effects trucks to comply with fuel economy and Benefits on vehicle producers, buyers, and users, emission regulations (line 1). It may also It is important to report the benefits plus the indirect or ‘‘external’’ benefits reduce fine payments by manufacturers and costs of this proposed action in a and costs it creates for the general who would have failed to comply with format that conveys useful information public. the more demanding baseline standards. about how those impacts are generated Table II–25 through Table II–28 Manufacturers are assumed to transfer and that also distinguishes the impacts present the economic benefits and costs these cost savings on to buyers by of those economic consequences for of the proposed action to reduce CAFE charging lower prices (line 5); although private businesses and households from and CO2 emissions standards for model this reduces their revenues (line 3), on the effects on the remainder of the U.S. years 2021–26 at three percent and balance, the reduction in compliance economy. A reporting format will seven percent discount rates in a format costs and lower sales revenue leaves accomplish the first objective to the that is intended to meet these objectives. them financially unaffected (line 4).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00077 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43062 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00078 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.039 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43063

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00079 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.040 43064 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00080 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.041 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43065

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00081 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.042 43066 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

As the tables show, most impacts of consumers who purchase, drive, and during model years prior to those the proposed action will fall on the subsequently sell or trade-in new covered by this action. Compared to the businesses and individuals who design, models (and ultimately bear the cost of baseline standards, if the preferred manufacture, and sell (at retail and fuel economy technology), and owners alternative is finalized, buyers of new wholesale) cars and light trucks, the of used cars and light trucks produced cars and light trucks will benefit from

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00082 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.043 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43067

their lower purchase prices and remainder of the U.S. economy. These additional benefits will be experienced financing costs (line 5). They will also impacts are ‘‘external,’’ in the sense that throughout much of the U.S. economy avoid the increased risks of being they are by-products of decisions by (line 17). Finally, some of the higher injured in crashes that would have private firms and individuals that alter fuel costs to buyers of new cars and resulted from manufacturers’ efforts to vehicle use and fuel consumption but light trucks will consist of increased reduce the weight of new models to are experienced broadly throughout the fuel taxes; this increase in revenue will comply with the baseline standards, U.S. economy rather than by the firms enable Federal and State government which represents another benefit from and individuals who indirectly cause agencies to provide higher levels of road reducing stringency vis-a`-vis the them. Increased refining and capacity or maintenance, producing baseline (line 6). consumption of petroleum-based fuel benefits for all road and transit users At the same time, new cars and light will increase emissions of carbon (line 18). trucks will offer lower fuel economy dioxide and other greenhouse gases that On balance, Table II–27 through Table with more lenient standards in place, theoretically contribute to climate II–28 show that the U.S. economy as a and this imposes various costs on their change, and some of the resulting (albeit whole will experience large net buyers and users. Drivers will uncertain) increase in economic economic benefits from the proposed experience higher costs as a damages from future changes in the action (line 22). While the proposal to consequence of new vehicles’ increased global climate will be borne throughout establish less stringent CAFE and GHG fuel consumption (line 7), and from the the U.S. economy (line 13). Similarly, emission standards will produce net added inconvenience of more frequent added fuel production and use will external economic costs, as the increase refueling stops required by their increase emissions of more localized air in environmental and energy security reduced driving range (line 8). They will pollutants (or their chemical externalities outweighs external benefits also forego some mobility benefits as precursors), and the resulting increase from reduced driving and higher fuel they use newly-purchased cars and light in the U.S. population’s exposure to tax revenue (line 19), the table also trucks less in response to their higher harmful levels of these pollutants will shows that combined benefits to vehicle fueling costs, although this loss will be lead to somewhat higher costs from its manufacturers, buyers, and users of cars almost fully offset by the fuel and other adverse effects on health (line 14). On and light trucks, and the general public costs they save by driving less (line 9). the other hand, it is expected that the (line 20), including the value of the lives On balance, consumers of new cars and proposed standards, by reducing new saved and injuries avoided, will greatly light trucks produced during the model vehicle prices relative to the baseline, outweigh the combined economic costs years subject to this proposed action will accelerate fleet turnover to cleaner, they experience as a consequence of this will experience significant economic safer, more efficient vehicles (as proposed action (line 21). benefits (line 10). compared to used vehicles that might The finding that this action to reduce By lowering prices for new cars and otherwise continue to be driven or the stringency of previously-established light trucks, this proposed action will purchased). CAFE and GHG standards will create cause some owners of used vehicles to As discussed in PRIA Section 9.8, significant net economic benefits— retire them from service earlier than increased consumption and imports of when it was initially claimed that they would otherwise have done, and crude petroleum for refining higher establishing those standards would also replace them with new models. In volumes of gasoline and diesel will also generate large economic benefits to effect, it will transfer some driving that impose some external costs throughout vehicle buyers and others throughout would have been done in used cars and the U.S. economy, in the form of the economy—is notable. This contrast light trucks under the baseline scenario potential losses in production and costs with the earlier finding is explained by to newer and safer models, thus for businesses and households to adjust the availability of updated information reducing costs for injuries (both fatal rapidly to sudden changes in energy on the costs and effectiveness of and less severe) and property damages prices (line 15 of the table), although technologies that will remain available sustained in motor vehicle crashes. This these costs should be tempered by to improve fuel economy in model years improvement in safety results from the increasing U.S. oil production.218 2021 and beyond, the fleet-wide fact that cars and light trucks have consequences for vehicle use, fuel Reductions in driving by buyers of new become progressively more protective in consumption, and safety from requiring cars and light trucks in response to their crashes over time (and also slightly less higher fuel economy (that is, higher operating costs will also reduce prone to certain types of crashes, such considering these consequences for used the external costs associated with their as rollovers). Thus, shifting some travel cars and light trucks as well as new contributions to traffic delays and noise from older to newer models reduces ones), and new estimates of some levels in urban areas, and these injuries and damages sustained by external costs of fuel in petroleum use. drivers and passengers because they are 218 Note: This output was based upon the EIA traveling in inherently safer vehicles 2. Macroeconomic Assumptions That Annual Energy Outlook from 2017. The 2018 Affect the Benefit Cost Analysis and not because it changes the risk Annual Energy Outlook projects the U.S. will be a profiles of drivers themselves. This net exporter by around 2029, with net exports Unlike previous CAFE and GHG reduction in injury risks and other peaking at around 0.5 mbd circa 2040. See Annual rulemaking analyses, the economic Energy Outlook 2018, U.S. Energy Information damage costs produces benefits to Administration, at 53 (Feb, 6, 2018), https:// context in which the alternatives are owners and drivers of older cars and www.eia.gov/outlooks/aeo/pdf/AEO2018.pdf. simulated is more explicit. While both light trucks. This also results in benefits Furthermore, pursuant to Executive Order 13783 this analysis and previous analyses in terms of improved fuel economy and (Promoting Energy Independence and Economy contained fuel price projections from Growth), agencies are expected to review and revise significant reductions of emissions from or rescind policies that unduly burden the the Annual Energy Outlook, which has newer vehicles (line 11). development of domestic energy resources beyond embedded assumptions about future Table II–27 through Table II–28 also what is necessary to protect the public interest or macroeconomic conditions, this show that the changes in fuel otherwise comply with the law. Therefore, it is analysis requires explicit assumptions reasonable to anticipate further increases in consumption and vehicle use resulting domestic production of petroleum. The agencies about future GDP growth, labor force from this proposed action will in turn may update the analysis and table to account for participation, and interest rates in order generate both benefits and costs to the this revised information. to evaluate the alternatives.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00083 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43068 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The analysis simulates compliance order to estimate their lifetime mileage cover a similar span of years. Due to the through MY 2032 explicitly and must accumulation and fuel consumption. long time horizon, a source that consider the full useful lives of those This means that any macroeconomic regularly produces such lengthy vehicles, approximately 40 years, in forecast influencing those factors must forecasts of these factors was selected:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00084 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.044 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43069

the 2017 OASDI Trustees Report from fuel calendar year and fuel type are the U.S. Social Security Administration. rulemaking, and this is one of several important presented in Table–II–30, in real 2016 While Table–II–29 only displays changes in circumstances supporting revision of dollars. Fuel prices in this analysis assumptions through CY 2050, the previously-issued standards. affect not only the value of each gallon remaining years merely continue the After significant portions of today’s analysis had of fuel consumed but relative valuation already been completed, EIA released AEO 2018, trends present in the table. which reports reference case fuel prices about 10% of fuel-saving technologies demanded The analysis once again uses fuel higher than reported in AEO 2017, though still well by the market as a result of their price projections from the 2017 Annual below the above-mentioned prices from AEO 2011. associated fuel savings. Energy Outlook.219 The projections by The sensitivity analysis therefore includes a case that applies fuel prices from the AEO 2018 net benefits between the two are about five percent 219 The central analysis supporting today’s reference case. The AEO 2018 low oil price case different, especially considering that decisions proposal uses reference case estimates of fuel prices reports fuel prices somewhat higher than the AEO regarding future standards are not single-factor reported in the Energy Information 2017 low oil price case, and the AEO 2018 high oil decisions, but rather reflect a balancing of factors, Administration’s (EIA’s) Annual Energy Outlook price case reports fuel prices very similar to the applying AEO 2018-based fuel prices would not 2017 (AEO 2017). Today’s proposal also examines AEO 2017 high oil price case. Adding the AEO 2018 materially change the extent to which today’s the sensitivity of this analysis to changes in key low and high oil price cases to the sensitivity analysis supports the selection of the preferred inputs, including fuel prices, and includes cases analysis would thus have provided little, if any, alternative. that apply fuel prices from the AEO 2017 low oil additional insight into the sensitivity of the analysis Like other inputs to the analysis, fuel prices will price and high oil price cases. The reference case to fuel prices. As shown in the summary of the be updated for the analysis supporting the final rule prices are considerably lower than AEO 2011-based sensitivity analysis, results obtained applying AEO after consideration of related new information and reference cases prices applied in the 2012 2018-based fuel prices are similar to those obtained public comment. applying AEO 2017-based fuel prices. For example,

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00085 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43070 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

3. New Vehicle Sales and Employment combination of manufacturer alternatives, for each manufacturer Assumptions compliance data, public data sources, down to the make/model level where and proprietary forecasts were used. the exact same number of each model In all previous CAFE and GHG When simulating compliance with variant was simulated to be sold in a rulemaking analyses, static fleet regulatory alternatives, the analysis given model year under both the least forecasts that were based on a projected identical sales across the stringent alternative (typically the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00086 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.045 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43071

baseline) and the most stringent will fully reflect future fuel cost savings fuel savings from purchasing a model alternative considered. To the extent consumers would realize from owning— with higher fuel economy. Most studies that an alternative matched the and potentially re-selling—more fuel- have relied on car buyers’ purchasing assumptions made in the production of efficient models. In this case, more behavior to estimate their willingness- the proprietary forecast, using a static stringent fuel economy standards will to-pay for future fuel savings; a typical fleet based upon those assumptions may impose net costs on vehicle owners and approach has been to use ‘‘discrete have been warranted. However, it seems can only result in social benefits by choice’’ models that relate individual intuitive that any sufficiently large span correcting externalities, since buyers’ choices among competing of regulatory alternatives would contain consumers would already fully vehicles to their purchase prices, fuel alternatives for which that static forecast incorporate private savings into their economy, and other attributes (such as was unrepresentative. A number of purchase decisions. If instead performance, carrying capacity, and commenters have encouraged consumers systematically undervalue reliability), and to infer buyers’ consideration of the potential impact of the cost savings generated by valuation of higher fuel economy from CAFE/GHG standards on new vehicle improvements in fuel economy when the relative importance of purchase prices and sales, and the changes to choosing among competing models, prices and fuel economy.221 Empirical compliance strategies that those shifts more stringent fuel economy standards estimates using this approach span a could necessitate.220 In particular, the will also lead manufacturers to adopt wide range, extending from substantial continued growth of the utility vehicle improvements in fuel economy that undervaluation of fuel savings to segment creates compliance challenges buyers might not choose despite the cost significant overvaluation, thus making it within some manufacturers’ fleets as savings they offer. difficult to draw solid conclusions about sales volumes shift from one region of The potential for car buyers to forego the influence of fuel economy on the footprint curve to another. improvements in fuel economy that vehicle buyers’ choices (see Helfand & Any model of sales response must offer savings exceeding their initial Wolverton, 2011; Green (2010) for satisfy two requirements: It must be costs is one example of what is often detailed reviews of these cross-sectional appropriate for use in the CAFE model, termed the ‘‘energy-efficiency gap.’’ studies). Because a vehicle’s price is and it must be econometrically This appearance of such a gap, between often correlated with its other attributes reasonable. The first of these the level of energy efficiency that would (both measured and unobserved), requirements implies that any variable minimize consumers’ overall expenses analysts have often used instrumental used in the estimation of the and what they actually purchase, is variables or other approaches to address econometric model, must also be typically based on engineering endogeneity and other resulting available as a forecast throughout the calculations that compare the initial concerns (e.g., Barry, et al. 1995). duration of the years covered by the cost for providing higher energy Despite these efforts, more recent simulations (this analysis explicitly efficiency to the discounted present research has criticized these cross- simulates compliance through MY value of the resulting savings in future sectional studies; some have questioned 2032). Some values the model calculates energy costs. the effectiveness of the instruments they endogenously, making them available in There has long been an active debate use (Allcott & Greenstone, 2012), while future years for sales estimation, but about why such a gap might arise and others have observed that coefficients others must be known in advance of the whether it actually exists. Economic estimated using non-linear statistical simulation. As the CAFE model theory predicts that individuals will methods can be sensitive to the simulates compliance, it accumulates purchase more energy-efficient products optimization algorithm and starting technology costs across the industry and only if the savings in future energy costs values (Knittel & Metaxoglou, 2014). over time. By starting with the last they offer promise to offset their higher Collinearity (i.e., high correlations) known transaction price and adding the initial costs. However, the additional among vehicle attributes—most notably accumulated technology cost to that cost of a more energy-efficient product among fuel economy, performance or value, the model is able to represent the includes more than just the cost of the power, and vehicle size—and between average selling price in each future technology necessary to improve its vehicles’ measured and unobserved efficiency; it also includes the model year assuming that manufacturers features also raises questions about the opportunity cost of any other desirable are able to pass all of their compliance reliability and interpretation of features that consumers give up when costs on to buyers of new vehicles. coefficients that may conflate the value they choose the more efficient Other variables used in the estimation of fuel economy with other attributes alternative. In the context of vehicles, must enter the model as inputs prior to (Sallee, et al., 2016; Busse, et al., 2013; whether the expected fuel savings the start of the compliance simulation. Allcott & Wozny, 2014; Allcott & outweigh the opportunity cost of Greenstone, 2012; Helfand & Wolverton, (a) How do car and light truck buyers purchasing a model offering higher fuel 2011). value improved fuel economy? economy will depend on how much its In an effort to overcome shortcomings How potential buyers value buyer expects to drive, his or her of past analyses, three recently improvements in the fuel economy of expectations about future fuel prices, published studies rely on panel data new cars and light trucks is an the discount rate he or she uses to value from sales of individual vehicle models important issue in assessing the benefits future expenses, the expected effect on to improve their reliability in and costs of government regulation. If resale value, and whether more efficient identifying the association between buyers fully value the savings in fuel models offer equivalent attributes such vehicles’ prices and their fuel economy costs that result from higher fuel as performance, carrying capacity, (Sallee, et al. 2016; Allcott & Wozny, economy, manufacturers will reliability, quality, or other 2014; Busse, et al., 2013). Although they presumably supply any improvements characteristics. differ in certain details, each of these Published literature has offered little that buyers demand, and vehicle prices consensus about consumers’ 221 In a typical vehicle choice model, the ratio of willingness-to-pay for greater fuel estimated coefficients on fuel economy—or more 220 See e.g., Comment by Alliance of Automobile commonly, fuel cost per mile driven—and purchase Manufacturers, Docket ID EPA–HQ–OAR–2015– economy, and whether it implies over- price is used to infer the dollar value buyers attach 0827–4089 and NHTSA–2016–0068–0072. , under- or full-valuation of the expected to slightly higher fuel economy.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00087 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43072 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

analyses relates changes over time in large proportion—and perhaps even As Table 1 indicates, Allcott & Wozny individual models’ selling prices to all—of the future savings that models (2014) find that consumers incorporate fluctuations in fuel prices, differences in with higher fuel economy offer.223 55% of future fuel costs into vehicle their fuel economy, and increases in Because they rely on estimates of fuel purchase decisions at a six percent their age and accumulated use, which costs over vehicles’ expected remaining discount rate, when their expectations affects their expected remaining life, lifetimes, these studies’ estimates of for future gasoline prices are assumed to and thus their market value. Because a how buyers value fuel economy are reflect prevailing prices at the time of vehicle’s future fuel costs are a function sensitive to the strategies they use to their purchases. With the same of both its fuel economy and expected isolate differences among individual expectation about future fuel prices, the gasoline prices, changes in fuel prices models’ fuel economy, as well as to authors report that consumers would have different effects on the market their assumptions about buyers’ fully value fuel costs only if they apply values of vehicles with different fuel discount rates and gasoline price discount rates of 24% or higher. economy; comparing these effects over expectations, among others. Since However, these authors’ estimates are time and among vehicle models reveals Anderson et al. (2013) find evidence closer to full valuation when using the fraction of changes in fuel costs that that consumers expect future gasoline gasoline price forecasts that mirror oil is reflected in changes in their selling prices to resemble current prices, we futures markets because the petroleum prices (Allcott & Wozny, 2014). Using use this assumption to compare the market expected prices to fall during very large samples of sales enables these findings of the three studies and this period (this outlook reduces the studies to define vehicle models at an examine how their findings vary with discounted value of a vehicle’s expected extremely disaggregated level, which the discount rates buyers apply to future remaining lifetime fuel costs). With this enables their authors to isolate fuel savings.224 expectation, Allcott & Wozny (2014) differences in their fuel economy from find that buyers value 76% of future the many other attributes, including 223 Killian & Sims (2006) and Sawhill (2008) rely cost savings (discounted at six percent) those that are difficult to observe or on similar longitudinal approaches to examine from choosing a model that offers higher 222 consumer valuation of fuel economy except that measure, that affect their sale prices. they use average values or list prices instead of fuel economy, and that a discount rate These studies point to a somewhat actual transaction prices. Since these studies of 15% would imply that they fully narrower range of estimates than remain unpublished, their empirical results are value future cost savings. Sallee et al. subject to change, and they are excluded from this suggested by previous cross-sectional (2016) begin with the perspective that studies; more importantly, they discussion. 224 Each of the studies makes slightly different buyers fully internalize future fuel costs consistently suggest that buyers value a assumptions about appropriate discount rates. into vehicles’ purchase prices and Sallee et al. (2016) use five percent in their base cannot reliably reject that hypothesis; 222 These studies rely on individual vehicle specification, while Allcott & Wozny (2014) rely on transaction data from dealer sales and wholesale six percent. As some authors note, a five to six their base specification suggests that auctions, which includes actual sale prices and percent discount rate is consistent with current changes in vehicle prices incorporate allows their authors to define vehicle models at a interest rates on car loans, but they also slightly more than 100% of changes in highly disaggregated level. For instance, Allcott & acknowledge that borrowing rates could be higher future fuel costs. For discount rates of Wozny (2014) differentiate vehicles by in some cases, which could be justify higher manufacturer, model or nameplate, trim level, body discount rates. Rather than assuming a specific five to six percent, the Busse et al. type, fuel economy, engine displacement, number discount rate, Busse et al. (2013) directly estimate (2013) results imply that vehicle prices of cylinders, and ‘‘generation’’ (a group of implicit discount rates at which future fuel costs reflect 60 to 100% of future fuel costs. successive model years during which a model’s would be fully internalized; they find discount rates As Table II–31 suggests, higher private design remains largely unchanged). All three of six to 21% for used cars and one to 13% for new studies include transactions only through mid-2008 cars at assumed demand elasticities ranging from discount rates move all of the estimates to limit the effect of the recession on vehicle prices. ¥2 to ¥3. Their estimates can be translated into closer to full valuation or to over- To ensure that the vehicle choice set consists of true the percent of fuel costs internalized by consumers, valuation, while lower discount rates substitutes, Allcott & Wozny (2014) define the assuming a particular discount rate. To make these imply less complete valuation in all choice set as all gasoline-fueled light-duty cars, results more directly comparable to the other two trucks, SUVs, and minivans that are less than 25 studies, we assume a range of discount rates and three studies. years old (i.e., they exclude vehicles where the uses the authors’ spreadsheet tool to translate their substitution elasticity is expected to be small). results into the percent of fuel costs internalized quartile, these results depend on which quartiles of Sallee et al. (2016) exclude diesels, hybrids, and into the purchase price at each rate. Because Busse the fuel economy distribution are compared; our used vehicles with less than 10,000 or more than et al. (2013) estimate the effects of future fuel costs summary shows results using the full range of 100,000 miles. on vehicle prices separately by fuel economy quartile comparisons.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00088 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43073

The studies also explore the repeat sales, most of the valuation Perhaps more important in assessing the sensitivity of the results to other estimates reported in these studies case for regulating fuel economy, one parameters that could influence their apply most directly to buyers of used study suggests that buyers of new cars results. Busse et al. (2013) and Allcott vehicles. Only Busse et al. (2013) and light trucks value three-quarters or & Wozny (2014) find that relying on examine new vehicle sales; they find more of the savings in future fuel costs data that suggest lower annual vehicle that consumers value between 75 to they anticipate from purchasing higher- use or survival probabilities, which 133% of future fuel costs for new mpg models, although this result is imply that vehicles will not last as long, vehicles, a higher range than they based on more limited information. moves their estimates closer to full estimate for used vehicles. Allcott & In contrast, previous regulatory valuation, an unsurprising result Wozny (2014) examine how their analyses of fuel economy standards because both reduce the changes in estimates vary by vehicle age and find implicitly assumed that buyers expected future fuel costs caused by fuel that fluctuations in purchase prices of undervalue even more of the benefits price fluctuations. Allcott & Wozny’s younger vehicles imply that buyers they would experience from purchasing (2014) base results rely on an whose fuel price expectations mirror the models with higher fuel economy so instrumental variables estimator that petroleum futures market value a higher that without increases in fuel economy groups miles-per-gallon (MPG) into two fraction of future fuel costs: 93% for standards little improvement would quantiles to mitigate potential one- to three-year-old vehicles, occur, and the entire value of fuel attenuation bias due to measurement compared to their estimate of 76% for savings from raising CAFE standards error in fuel economy, but they find that all used vehicles assuming the same represented private benefits to car and greater disaggregation of the MPG price expectation.225 light truck buyers themselves. For groups implies greater undervaluation Accounting for differences in their instance, in the EPA analysis of the (for example, it reduces the 55% data and estimation procedures, the 2017–2025 model year greenhouse gas estimated reported in Table 1 to 49%). three studies described here suggest that emission standards, fuel savings alone Busse et al. (2013) allow gasoline prices car buyers who use discount rates of added up to $475 billion (at three to vary across local markets in their five to six percent value at least half— percent discount rate) over the lifetime main specification; using national and perhaps all—of the savings in future of the vehicles, far outweighing the average gasoline prices, an approach fuel costs they expect from choosing compliance costs: $150 billion). The more directly comparable to the other models that offer higher fuel economy. assertion that buyers were unwilling to studies, results in estimates that are take voluntary advantage of this closer to or above full valuation. Sallee 225 Allcott & Wozny (2014) and Sallee, et al. opportunity implies that collectively, et al. (2016) find modest undervaluation (2016) also find that future fuel costs for older they must have valued less than a third vehicles are substantially undervalued (26–30%). by vehicle fleet operators or ($150 billion/$475 billion = 32%) of the The pattern of Allcott and Wozny’s results for fuel savings that would have resulted manufacturers making large-scale different vehicle ages is similar when they use retail 226 purchases, compared to retail dealer transaction prices (adjusted for customer cash from those standards. The evidence sales (i.e., 70 to 86%). rebates and trade-in values) instead of wholesale auction prices, although the degree of valuation 226 In fact, those earlier analyses assumed that Since they rely predominantly on falls substantially in all age cohorts with the new car and light truck buyers attach relatively changes in vehicles’ prices between smaller, retail price based sample. Continued

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00089 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.046 43074 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

reviewed here makes that perspective tends to move in parallel with changes to purchase new cars and light trucks in extremely difficult to justify and would in economic growth; that is, average conjunction with the total number sold, call into question any analysis that new vehicle prices tend to be higher and other approaches. However, none of claims to show large private net benefits when the total number of new vehicles these methods offered a significant for vehicle buyers. sold is increasing and lower when the improvement over estimating the total What analysts assume about total number of new sales decreases number of vehicles sold directly from its consumers’ vehicle purchasing (typically during periods of low historical relationship to directly behavior, particularly about potential economic growth or recessions). Finally, measurable factors such as their average buyers’ perspectives on the value of counts of the total number of new cars sales price, macroeconomic variables increased fuel economy, clearly matters and light trucks that are sold do not such as GDP or Personal Disposable a great deal in the context of benefit-cost capture shifts in demand among vehicle Income, U.S. labor force participation, analysis for fuel economy regulation. In size classes or body styles (‘‘market and regularly published surveys of light of recent evidence on this segments’’); nor do they measure consumer sentiment or confidence. question, a more nuanced approach changes in the durability, safety, fuel Quarterly, rather than annual data on than assuming that buyers drastically economy, carrying capacity, comfort, or total sales of new cars and light trucks, undervalue benefits from higher fuel other aspects of vehicles’ quality. their average selling price, and economy, and that as a consequence, The historical series of new light-duty macroeconomic variables was used to these benefits are unlikely to be realized vehicle sales exhibits cyclic behavior develop an econometric model of sales, without stringent fuel economy over time that is most responsive to in order to increase the number of standards, seems warranted. One larger cycles in the macro economy— observations and more accurately possible approach would be to use a but has not increased over time in the capture the causal effects of individual baseline scenario where fuel economy same way the population, for example, explanatory variables. Applying levels of new cars and light trucks has. While U.S. population has grown conventional data diagnostics for time- reflected full (or nearly so) valuation of over 35 percent since 1980, the series economic data revealed that most fuel savings by potential buyers in order registered vehicle population has grown variables were non-stationary (i.e., they to reveal whether setting fuel economy at an even faster pace—nearly doubling reflected strong underlying time trends) 227 standards above market-determined between 1980 and 2015. But annual and displayed unit roots, and statistical levels could produce net social benefits. vehicle sales did not grow at a similar tests revealed co-integration between Another might be to assume that, unlike pace –even accounting for the cyclical the total vehicle sales—the model’s in the agencies’ previous analyses, nature of the industry. Total new light- dependent variable—and most where buyers were assumed to greatly duty sales prior to the 2008 recession candidate explanatory variables. climbed as high as 16 million, though undervalue higher fuel economy under (c) Current Estimation of Sales Impacts the baseline but to value it fully under similarly high sales years occurred in the 1980’s and 1990’s as well. In fact, To address the complications of the the proposed standards, buyers value time series data, the analysis estimated improved fuel economy identically when considering a 10-year moving average to smooth out the effect of an autoregressive distributed-lag (ARDL) under both the baseline scenario and model that employs a combination of with stricter CAFE standards in place. cycles, most 10-year averages between 1992 and 2015 are within a few percent lagged values of its dependent The agencies ask for comment on these variable—in this case, last year’s and the and any alternative approaches they of the 10-year average in 1992. And although average transaction prices for prior year’s vehicle sales—and the should consider for valuing fuel savings, new vehicles have been rising steadily change in average vehicle price, new peer-reviewed evidence on vehicle since the recession ended, prices are not quarterly changes in the U.S. GDP buyers’ behavior that casts light on how yet at historical highs when adjusted for growth rate, as well as current and they value improved fuel economy, the inflation. The period of highest lagged values of quarterly estimates of appropriate private discount rate to inflation-adjusted transaction prices U.S. labor force participation. The apply to future fuel savings, and thus occurred from 1996–2006, when the number of lagged values of each the degree to which private fuel savings average transaction price for a new explanatory variable to include was should be considered as private benefits light-duty vehicle was consistently determined empirically (using the of increasing fuel economy standards. higher than the price in 2015. Bayesian information criterion), by (b) Sales Data and Relevant In an attempt to overcome these examining the effects of including Macroeconomic Factors analytical challenges, various different combinations of their lagged approaches were experimented with to values on how well the model Developing a procedure to predict the predict the response of new vehicle ‘‘explained’’ historical variation in car effects of changes in prices and sales to the changes in prices, fuel and light truck sales. attributes of new vehicles is economy, and other features. These The results of this approach were complicated by the fact that their sales included treating new vehicle demand encouraging: The model’s predictions fit are highly pro-cyclical—that is, they are as a product of changes in total demand the historical data on sales well, each of very sensitive to changes in for vehicle ownership and demand its explanatory variables displayed the macroeconomic conditions—and also necessary to replace used vehicles that expected effect on sales, and analysis of statistically ‘‘noisy,’’ because they are retired, analyzing total expenditures its unexplained residual terms revealed reflect the transient effects of other little evidence of autocorrelation or factors such as consumers’ confidence 227 There are two measurements of the size of the other indications of statistical problems. in the future, which can be difficult to registered vehicle population that are considered to The model coefficients suggest that observe and measure accurately. At the be authoritative. One is produced by the Federal Highway Adminstration, and the other by R.L. Polk positive GDP growth rates and increases same time, their average sales price (now part of IHS). The Polk measurement shows in labor force participation are both fleet growth between 1980 and 2015 of about 85%, indicators of increases in new vehicle little value to higher fuel economy, since their while the FHWA measurement shows a slower sales, while positive changes in average baseline scenarios assumed that fuel economy growth rate over that period; only about 60%. Both levels would not increase in the absence of are still considerably larger than the growth in new new vehicle price reduce new sales. progressively tighter standards. vehicle sales over the same period. However, the magnitude of the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00090 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43075

coefficient on change in average price is buyer is (at least theoretically) choosing price increases, fuel economy, and new not as determinative of total sales as the between a more or less efficient version vehicle sales, as well as methods to other variables. of a pickup truck (for example) that is appropriately account for these Based on the model, a $1,000 increase otherwise identical. In an aggregate relationships. in the average new vehicle price causes sense, the average is not comparable to (d) Projecting New Vehicle Sales and approximately 170,000 lost units in the the decision an individual consumer Comparisons to Other Forecasts first year, followed by a reduction of faces. another 600,000 units over the next ten Estimating the sales response at the The purpose of the sales response years as the initial sales decrease level of total new vehicle sales likely model is to allow the CAFE model to propagates over time through the lagged fails to address valid concerns about simulate new vehicle sales in a given variables and their coefficients. The changes to the quality or attributes of future model year, accounting for the price elasticity of new car and light new vehicles sold—both over time and impact of a regulatory alternative’s truck sales implied by alternative in response to price increases resulting stringency on new vehicle prices (in a estimates of the model’s coefficients from CAFE standards. However, macro-economic context that is ¥ ¥ ranged from 0.2 to 0.3—meaning attempts to address such concerns identical across alternatives). In order to that changes in their prices have would require significant additional accomplish this, it is important that the moderate effects on total sales—which data, new statistical approaches, and model of sales response be dynamically contrasts with estimates of higher structural changes to the CAFE model stable, meaning that it responds to sensitivity to prices implied by some over several years. It is also the case that shocks not by ‘‘exploding,’’ increasing 228 models. The analysis was unable to using absolute changes in the average or decreasing in a way that is incorporate any measure of new car and price may be more limited than another unbounded, but rather returns to a light truck fuel economy in the model characterization of price that relies on stable path, allowing the shock to that added to its ability to explain distributions of household income over dissipate. The CAFE model uses the historical variation in sales, even after time or percentage change in the new sales model described above to experimenting with alternative vehicle price. The former would require dynamically project future sales; after measures of such as the unweighted and forecasting a deeply uncertain quantity the first year of the simulation, lagged sales-weighted averages fuel economy of many years into the future, and the values of new vehicle sales are those models sold in each quarter, the level of latter only become relevant once the that were produced by the model itself fuel economy they were required to simulation moves beyond the rather than observed. The sales response achieve, and the change in their fuel magnitude of observed price changes in model constructed here uses two lagged economy from previous periods. the historical series. Future versions of dependent variables and simple Despite the evidence in the literature, this model may use a different econometric conditions determine if the summarized above, that consumers model is dynamically stable. The value most, if not all, of the fuel characterization of cost that accounts for some of these factors if their inclusion coefficients of the one-year lag and the economy improvements when b b improves the model estimation and two-year lag, 1 and 2, respectively purchasing new vehicles, the model must satisfy three conditions. Their sum described here operates at too high a corresponding forecast projections are must be less than one, b ¥ b <1, and level of aggregation to capture these available. 2 1 the absolute value of b must be less preferences. By modeling the total The changes in selling prices, fuel 2 than one. The coefficients of this model number of new vehicles sold in a given economy, and other features of cars and satisfy all three conditions. year, it is necessary to quantify light trucks produced during future important measures, like sales price or model years that result from Using the Augural CAFE standards as fuel economy, by averages. Our model manufacturers’ responses to lower CAFE the baseline, it is possible to produce a operates at a high level of aggregation, and GHG emission standards are likely series of future total sales as shown in where the average fuel economy to affect both sales of individual models Table–II–32. For comparison, the table represents an average across many and the total number of new vehicles includes the calculated total light-duty vehicle types, usage profiles, and fuel sold. Because the values of changes in sales of a proprietary forecast purchased economy levels. In this context, the fuel economy and other features to to support the 2016 Draft TAR analysis, average fuel economy was not a potential buyers are not completely the total new light-duty sales in EIA’s meaningful value with respect to its understood; however, the magnitude, 2017 Annual Energy Outlook, and a influence on the total number of new and possibly even the direction, of their (short) forecast published in the Center vehicles sold. A number of recent effect on sales of new vehicles is for Automotive Research’s Q4 2017 studies have indeed shown that difficult to anticipate. On balance, it is Automotive Outlook. All of the forecasts consumers value fuel savings (almost) reasonable to assume that the changes in in Table–II–32 assume the Augural fully. Those studies are frequently based prices, fuel economy, and other Standards are in place through MY on large datasets that are able to control attributes expected to result from their 2025, though assumptions about the for all other vehicle attributes through a proposed action to amend and establish costs required to comply with them variety of econometric techniques. They fuel economy and GHG emission likely differ. As the table shows, despite represent micro-level decisions, where a standards are likely to increase total differences among them, the sales of new cars and light trucks during dynamically produced sales projection 228 Effects on the used car market are accounted future model years. Please provide from the CAFE model is not for separately. comment on the relationship between qualitatively different from the others.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00091 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43076 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

While this forecast projects a vehicles during the analysis period, Outlook; this integration continues that relatively high, but flat, level of new though it would continue the trend at a approach in a way that ensures greater vehicle sales into the future, it is worth slower rate. internal consistency when simulating noting that it continues another trend In addition to the statistical model multiple regulatory alternatives (and observed in the historical data. The time that estimates the response of total new conducting sensitivity analysis on any series of annual new vehicle sales is vehicle sales to changes in the average of the factors that influence fleet share). volatile from year to year, but multi-year new vehicle price, the CAFE model (e) Vehicle Choice Models as an averages are less so being sufficient to incorporates a dynamic fleet share Alternative Method To Estimate New wash out the variation associated with model that modifies the light truck (and, Vehicle Sales them peaks and valleys of the series. symmetrically, passenger car) share of Despite the fact that the moving average the new vehicle market. A version of Another potential option to estimate annual new vehicle sales has been this model first appeared in the 2012 future new vehicle sales would be to use growing over the last four decades, it final rule, when this fleet share a full consumer choice model. The has not kept pace with U.S. population component was introduced to ensure agencies simulate compliance with growth. Data from the Federal Reserve greater internal consistency within CAFE and CO2 standards for each Bank of St. Louis shows that the per- inputs in the uncertainty analysis. For manufacturer using a disaggregated capita sales of new vehicles peaked in today’s analysis, this dynamic fleet representation of its regulated vehicle 1986 and has declined more than 25% share is enabled throughout the analysis fleets. This means that each from this peak to today’s level.231 While of alternatives. manufacturer may have hundreds of the sales projection in Table–II–32 The dynamic fleet share model is a vehicle model variants (e.g., the Honda would represent a historically high series of difference equations that Civic with the 6-cylinder engine, and average of new vehicle sales over the determine the relative share of light the Honda Civic with the 4-cylinder analysis period, it would not be trucks and passenger cars based on the engine would each be treated as sufficient to reverse the trend of average fuel economy of each, the fuel different, in some ways, during the declining per-capita sales of new price, and average vehicle attributes like compliance simulation).232 While the horsepower and vehicle mass (the latter analysis accounts for a wide variety of 229 Out of necessity, the analysis in today’s rule of which explicitly evolves as a result of attributes across these vehicles, only a conflates production year (or ‘‘model year’’) and the compliance simulation). While this few of them change during the calendar year. The volumes cited in the CAFE model was taken from EIA’s National compliance simulation. However, all of model forecast represent forecasted production Energy Modeling System (NEMS), it is volumes for those model years, while the other those attributes are relevant in the represent calendar year sales (rather than applied at a different level. Rather than context of consumer choice models. production)—during which two, or possibly three, apply the shares based on the regulatory Aside from the computational different model year vehicles are sold. In the long class distinction, the CAFE model intensity of simulating new vehicle run, the difference is not important. In the early applies the shares to body-style. This is sales at the level of individual models— years, there are likely to be discrepancies. done to account for the large-scale shift 230 U.S. Total Sales by Make, Automotive News, for all manufacturers, under each http://www.autonews.com/section/datalist18 (last in recent years to crossover utility regulatory alternative, over the next visited June 22, 2018). vehicles that have model variants in decade or more—it would be necessary 231 Mislinski, J. Light Vehicle Sales Per Capita: both the passenger car and light truck to include additional relationships Our Latest Look at the Long-Term Trend, Advisor regulatory fleets. The agencies have Perspectives (June 1, 2018), https:// www.advisorperspectives.com/dshort/updates/ always modified their static forecasts of 232 For more detail about the compliance 2018/05/01/light-vehicle-sales-per-capita-our-latest- new vehicle sales to reflect the PC/LT simulation and manufacturer fleet representation, look-at-the-long-term-trend. split present in the Annual Energy see Section II.G.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00092 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.047 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43077

about how consumers trade off among quality and reliability, interior volume, branding. While the marketing budget of vehicle attributes, which types of or comfort. However, other factors also individual manufacturers may help a consumers prefer which types of influence customers’ purchase decision, consumer choice model estimate market attributes (and how much), and how and some of these can be challenging to share for a given brand, estimating the manufacturers might strategically price model. Consumer proximity to impact of a given campaign on new these modified vehicles. This requires a dealerships, quality of service and sales is more challenging as consumers strategic pricing model, which each customer experience at dealerships, make purchasing decisions based upon manufacturer has and would likely be availability and terms of financing, and their own needs and desires. unwilling to share. Some of this basic product awareness may Modelers must understand how strategic pricing behavior occurs on significantly factor into sales success. small time-scale through the use of Manufacturers’ marketing choices consumers and commercial buyers dealer incentives, rebates on specific may significantly and unpredictably select vehicles in order to effectively models, and creative financing offers. affect sales. Ad campaigns may increase develop and implement a consumer When simulating compliance at the awareness in the market, and campaigns choice model in a compliance annual scale, it is effectively impossible may reposition consumers’ perception simulation. Consumers purchase to account for these types of strategic of the brands and products. For vehicles for a variety of reasons such as decisions. example, in 2011 the Volkswagen Passat family need, need for more space, new It is also true consumers have featured an ad with a child in a Darth technology, changes to income and heterogeneous preferences that change Vader costume (and showcased remote affordability of a new vehicle, improved over time and determine willingness-to- start technology on the Passat). In MY fuel economy, operating costs of current pay for a variety of vehicle attributes. 2012, Kia established the Kia Soul with vehicles, and others. Once committed to These preferences change in response to party rocking, hip-hop hamster buying a vehicle, consumers use marketing, distribution, pricing, and commercials showcasing push-button different processes to narrow down their product strategies that manufacturers ignition, a roomy interior, and design shopping list. Consumer choice decision may change over time. With enough features in the brake lights. Both attributes include factors both related data, a consumer choice model could commercials raised awareness and and not related to the vehicle design. stratify new vehicle buyers into types highlighted basic product features. Each The vehicle’s utility for those attributes and attempt to measure the strength of commercial also impressed is researched across many different each type’s preference for fuel economy, demographic groups with pop culture information sources as listed in the table acceleration, safety rating, perceived references, product placement, and co- below.

An objective, attribute-based projections of future sales volumes at not so stringent as to’’ lead to ‘‘adverse consumer choice model could lead to the model, segment, or manufacturer economic consequences, such as a projected swings in manufacturer level. significant loss of jobs or unreasonable market shares and individual model Comment is sought on the elimination of consumer choice.’’ 233 volumes. The current approach development and use of potential EPA similarly conducted an industry simulates compliance for each consumer choice model in compliance employment analysis under the broad manufacturer assuming that it produces simulations. Comment is also sought on authority granted to the agency under the same set of vehicles that it produced the appropriate breadth, depth, and the Clean Air Act.234 Both agencies in the initial year of the simulation (MY complexity of considerations in a recognized the uncertainties inherent in 2016 in today’s analysis). If a consumer consumer choice model. estimating industry employment choice model were to drive projected impacts; in fact, both agencies dedicated sales of a given vehicle model below (f) Industry Employment Baseline a substantial amount of discussion to some threshold, as consumers have (Including Multiplier Effect) and Data uncertainty in industry employment done in the real market, the simulation Description analyses in the 2012 final rule for MYs 235 currently has no way to generate a new In the first two joint CAFE/CO2 2017 and beyond. Notwithstanding vehicle model to take its place. As rulemakings, the agencies considered an these uncertainties, CAFE and CO2 demand changes across specific market analysis of industry employment standards do impact industry labor segments and models, manufacturers impacts in some form in setting both hours, and providing the best analysis adapt by supplying new vehicle CAFE and emissions standards; NHTSA practicable better informs stakeholders nameplates and models (e.g., the conducted an industry employment proliferation of crossover utility analysis in part to determine whether 233 67 FR 77015, 77021 (Dec. 16, 2002). vehicles in recent years). Absent that the standards the agency set were 234 See George E. Warren Corp. v. EPA, 159 F.3d 616, 623–624 (D.C. Cir. 1998) (ordinarily flexibility in the compliance simulation, economically practicable, that is, permissible for EPA to consider factors not even the more accurate consumer choice whether the standards were ‘‘within the specifically enumerated in the Act). model may produce unrealistic financial capability of the industry, but 235 See 77 FR 62624, 62952, 63102 (Oct. 15, 2012).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00093 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.048 43078 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

and the public about the standards’ The analysis also accounts for sales consider changes in labor at retail gas impact than would omitting any projections in response to the different stations. The analysis did not consider estimates of potential labor impacts. regulatory alternatives; the labor possible labor changes due to raw Today many of the effects that were analysis considers changes in new material production, such as production previously qualitatively identified, but vehicle prices and new vehicle sales (for of aluminum, steel, copper and lithium, not considered, are quantified. For further discussion of the sales model, nor did the agencies consider possible instance, in the PRIA for the 2017–2025 see Section 2.E). As vehicle prices rise, labor impacts due to changes in rule EPA identified ‘‘demand effects,’’ the analysis expected consumers to production of oil and gas, ethanol, and ‘‘cost effects,’’ and ‘‘factor shift effects’’ purchase fewer vehicles than they electricity. The analysis did not analyze as important considerations for industry would have at lower prices. As effects of how consumers could spend labor, but the analysis did not attempt manufacturers sell fewer vehicles, the money saved due to improved fuel to quantify either the demand effect or manufacturers may need less labor to economy, nor did the analysis assess the the factor shift effect.236 Today’s produce the vehicles and less labor to effects of how consumers would pay for industry labor analysis quantifies direct sell the vehicles. However, as more expensive fuel savings labor changes that were qualitatively manufacturers add equipment to each technologies at the time of purchase; discussed previously. new vehicle, the manufacturers will either could affect consumption of other Previous analyses and new require human resources to develop, goods and services, and hence affect methodologies to consider direct labor sell, and produce additional fuel-saving labor in other industries. The effects of effects on the automotive sector in the technologies. The analysis also accounts increased usage of car-sharing, ride- United States were improved upon and for the potential that new standards sharing, and automated vehicles were developed. Potential changes that were could shift the relative shares of not analyzed. The analysis did not evaluated include (1) dealership labor passenger cars and light trucks in the estimate how changes in labor from any related to new light duty vehicle unit overall fleet (see Section 2.E); insofar as industry could affect gross domestic sales; (2) changes in assembly labor for different vehicles involved different product and possibly affect other vehicles, for engines and for amounts of labor, this shifting impacts industries as a result. transmissions related to new vehicle the quantity of estimated labor. The Finally, no assumptions were made CAFE model automotive labor analysis unit sales; and (3) changes in industry about full-employment or not full- takes into account reduction in vehicle labor related to additional fuel savings employment and the availability of sales, shifts in the mix of passenger cars technologies, accounting for new human resources to fill positions. When and light trucks, and addition of fuel- vehicle unit sales. All automotive labor the economy is at full employment, a savings technologies. effects were estimated and reported at a fuel economy regulation is unlikely to 237 For today’s analysis, it was assumed national level, in job-years, assuming have much impact on net overall U.S. 2,000 hours of labor per job-year. that some observations about the production of MY 2016 vehicles would employment; instead, labor would The analysis estimated labor effects primarily be shifted from one sector to from the forecasted CAFE model carry forward, unchanged into the future. For instance, assembly plants another. These shifts in employment technology costs and from review of impose an opportunity cost on society, automotive labor for the MY 2016 fleet. would remain the same as MY 2016 for all products now, and in the future. The approximated by the wages of the For each vehicle in the CAFE model employees, as regulation diverts analysis, the locations for vehicle analysis assumed percent U.S. content would remain constant, even as workers from other activities in the assembly, engine assembly, and economy. In this situation, any effects transmission assembly and estimated manufacturers updated vehicles and introduced new fuel-saving on net employment are likely to be labor in MY 2016 were recorded. The transitory as workers change jobs (e.g., percent U.S. content for each vehicle technologies. It was assumed that assembly labor hours per unit would some workers may need to be retrained was also recorded. Not all parts are or require time to search for new jobs, made in the United States, so the remain at estimated MY 2016 levels for vehicles, engines, and transmissions, while shortages in some sectors or analysis also took into account the regions could bid up wages to attract percent U.S. content for each vehicle as and the factor between direct assembly labor and parts production jobs would workers). On the other hand, if a manufacturers add fuel-savings regulation comes into effect during a technologies. As manufacturers added remain the same. When considering shifts from one technology to another, period of high unemployment, a change fuel-economy technologies in the CAFE in labor demand due to regulation may model simulations, the analysis the analysis assumed revenue per employee at suppliers and original affect net overall U.S. employment assumed percent U.S. content would because the labor market is not in remain constant in the future, and that equipment manufacturers would remain in line with MY 2016 levels, even as equilibrium. Schmalansee and Stavins the U.S. labor added would be point out that net positive employment proportional to U.S. content. From this manufacturers added fuel-saving technologies and realized cost effects are possible in the near term foundation, the analysis forecasted when the economy is at less than full automotive labor effects as the CAFE reductions from learning. The analysis focused on automotive employment due to the potential hiring model added fuel economy technology labor because adjacent employment of idle labor resources by the regulated and adjusted future sales for each factors and consumer spending factors sector to meet new requirements (e.g., to vehicle. for other goods and services are install new equipment) and new uncertain and difficult to predict. The economic activity in sectors related to 236 Regulatory Impact Analysis: Final Rulemaking for 2017–2025 Light-Duty Vehicle Greenhouse Gas analysis did not consider how direct the regulated sector longer run, the net Emission Standards and Corporate Average Fuel labor changes may affect the macro effect on employment is more difficult Economy Standards, U.S. EPA at 8–24 to 8–32 economy and possibly change to predict and will depend on the way (Aug. 2012). employment in adjacent industries. For in which the related industries respond 237 The agencies recognize a few local production to the regulatory requirements. For that facilities may contribute meaningfully to local instance, the analysis did not consider economies, but the analysis reported only on possible labor changes in vehicle reason, this analysis does not include national effects. maintenance and repair, nor did it multiplier effects but instead focuses on

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00094 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43079

labor impacts in the most directly assembly plant employment and analysis looked at financial reports from affected industries. Those sectors are production statistics and other publicly publicly traded automotive likely to face the most concentrated available information. The analysis businesses.240 Based on recent figures, it labor impacts. recognizes that some plants may use was estimated that OEMs would add Comment is sought on these less labor than the analysis estimates to one labor year per $633,066 revenue 241 assumptions and approaches in the produce the vehicle, the engine, or the and that suppliers would add one labor labor analysis. transmission, and other plants may have year per $247,648 in revenue.242 These 4. Estimating Labor for Fuel Economy used more labor. The analysis used the global estimates are applied to all Technologies, Vehicle Components, assembly locations and industry revenues, and U.S. content is applied as Final Assembly, and Retailers averages for labor per unit to estimate a later adjustment. In today’s analysis, it U.S. assembly labor hours for each was assumed these ratios would remain The following sections discuss the vehicle. U.S. assembly labor hours per constant for all technologies rather than approaches to estimating factors related vehicle ranged from as high as 39 hours that the increased labor costs would be to dealership labor, final assembly labor if the manufacturer assembled the shifted toward foreign countries. and parts production, and fuel economy vehicle, engine, and transmission at Comment is sought on the realism of technology labor. U.S. plants, to as low as zero hours if this assumption. the manufacturer imported the vehicle, (a) Dealership Labor (d) Labor Calculations engine, and transmission. The analysis evaluated dealership The analysis also considered labor for The analysis estimated the total labor labor related to new light-duty vehicle part production in addition to labor for as the sum of three components: sales, and estimated the labor hours per final assembly. Motor vehicle and Dealership hours, final assembly and new vehicle sold at dealerships, equipment manufacturing labor parts production, and labor for fuel- including labor from sales, finance, statistics from the U.S. Census Bureau, economy technologies (at OEM’s and insurance, and management. The effect the Bureau of Labor Statistics,239 and suppliers). The CAFE model calculated of new car sales on the maintenance, other publicly available sources were additional labor hours for each vehicle, repair, and parts department labor is surveyed. Based on these sources, the based on current vehicle manufacturing expected to be limited, as this need is analysis noted that the historical locations and simulation outputs for based on the vehicle miles traveled of average ratio of vehicle assembly additional technologies, and sales the total fleet. To estimate the labor manufacturing employment to changes. The analysis applied some hours at dealerships per new vehicle employment for total motor vehicle and constants to all vehicles,243 but other sold, the National Automobile Dealers equipment manufacturing for new constants were vehicle specific,244 or Association 2016 Annual Report, which vehicles remained roughly constant over year specific for a vehicle.245 provides franchise dealer employment the period from 2001 through 2013, at While a multiplier effect of all U.S. by department and function, was a ratio of 5.26. Observations from 2001– automotive related jobs on non-auto referenced.238 The analysis estimated 2013 spanned many years, many related U.S. jobs was not considered for that slightly less than 20% of dealership combinations of technologies and today’s analysis, the analysis did employees’ work relates to new car sales technology trends, and many economic program a ‘‘global multiplier’’ that can (versus approximately 80% in service, conditions, yet the ratio remained about be used to scale up or scale down the parts, and used car sales), and that on the same. Accordingly, the analysis total labor hours. This multiplier exists average dealership employees working scaled up estimated U.S. assembly labor in the parameters file, and for today’s on new vehicle sales labor for 27.8 hours by a factor of 5.26 to consider U.S. analysis the analysis set the value at hours per new vehicle sold. parts production labor in addition to 1.00. (b) Final Assembly Labor and Parts assembly labor for each vehicle. 5. Additional Costs and Benefits Production The industry estimates for vehicle Incurred by New Vehicle Buyers assembly labor and parts production How the quantity of assembly labor labor for each vehicle scaled up or down Some costs of purchasing and owning and parts production labor for MY 2016 as unit sales scaled up or down over a new or used vehicle scale with the vehicles would increase or decrease in time in the CAFE model. the future as new vehicle unit sales 240 The analysis considered suppliers that won increased or decreased was estimated. (c) Fuel Economy Technology Labor the Automotive News ‘‘PACE Award’’ from 2013– 2017, covering more than 40 suppliers, more than Specific assembly locations for final As manufacturers spend additional 30 of which are publicly traded companies. vehicle assembly, engine assembly, and dollars on fuel-saving technologies, Automotive News gives ‘‘PACE Awards’’ to transmission assembly for each MY parts suppliers and manufacturers innovative manufacturers, with most recent 2016 vehicle were identified. In some winners earning awards for new fuel-savings require human resources to bring those technologies. cases, manufacturers assembled technologies to market. Manufacturers 241 The analysis assumed incremental OEM products in more than one location, and may add, shift, or replace employees in revenue as the retail price equivalent for the analysis identified such products ways that are difficult for the agencies technologies, adjusting for changes in sales volume. and considered parallel production in 242 The analysis assumed incremental supplier to predict in response to adding fuel- revenue as the technology cost for technologies the labor analysis. savings technologies; however, it is before retail price equivalent mark-up, adjusting for The analysis estimated industry expected that the revenue per labor hour changes in sales volume. average direct assembly labor per at original equipment manufacturers 243 The analysis applied the same assumptions to vehicle (30 hours), per engine (four (OEMs) and suppliers will remain about all manufacturers for annual labor hours per hours), and per transmission (five employee, dealership hours per unit sold, OEM the same as in MY 2016 even as revenue per employee, supplier revenue per hours) based on a sample of U.S. industry includes additional fuel-saving employee, and factor for the jobs multiplier. technology. 244 The analysis made vehicle specific 238 NADA Data 2016: Annaul Financial Profile of To estimate the average revenue per assumptions about percent U.S. content and U.S. America’s Franchised New-Car Dealerships, assembly employment hours. National Automobile Dealers Association, https:// labor hour at OEMs and suppliers, the 245 The analysis estimated technology cost for www.nada.org/2016NADAdata/ (last visited June each vehicle, for each year based on the technology 22, 2018). 239 NAICS Code 3361, 3363. content applied in the CAFE model, year-by-year.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00095 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43080 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

value of the vehicle. Where fuel national weighted-average sales tax rate average new vehicle MSRP. Using the economy standards increase the almost identical to that resulting from assumption that 85% of new vehicle transaction price of vehicles, they will the use of Census population estimates purchases involve some financing, the affect both the absolute amount paid in as weights, just slightly above 5.5%. The average share of the MSRP financed for sales tax and the average amount of analysis opted to utilize Census all vehicles purchased, including non- financing required to purchase the population rather than the registration- financed transactions, rather than only vehicle. Further, where they increase based proxy of new vehicle sales as the those that are financed, was computed. the MSRP, they increase the appraised basis for computing this weighted Table–II–34 shows that this share ranges value upon which both value-related average, as the end results were between 70 and 72%. From this, the registration fees and a portion of negligibly different and the analytical analysis assumed that on an aggregate insurance premiums are based. The approach involving new vehicle level, including all new vehicle analysis assumes that the transaction registrations had not been as thoroughly price is a set share of the MSRP, which reviewed. Note: Sales taxes and purchases, 70% of the value of all allows calculation of these factors as registration fees are transfer payments vehicles’ MSRP is financed. It is likely shares of MSRP. Below the assumptions between consumers and the Federal that the share financed is correlated made about how each of these government and are therefore not with the MSRP of the new vehicle additional costs of vehicle purchase and considered a cost in the societal purchased, but for simplification ownership scale with the MSRP and perspective. However, these costs are purposes, it is assumed that 70% of all how the analysis arrived at these considered as additional costs in the vehicle costs are financed, regardless of assumptions are discussed. private consumer perspective. the MSRP of the vehicle. In measurements of the impacts on the (a) Sales Taxes (b) Financing Costs average consumer, this assumption will The analysis took auto sales taxes by The analysis assumes 85% of not affect the outcome of our state 246 and weighted them by automobiles are financed based on calculation, though this assumption will population by state to determine a Experian’s quarter 4, 2016 ‘‘State of the matter for any discussions about how national weighted-average sales tax of Automotive Finance Market,’’ which many, or which, consumers bear the 5.46%. The analysis sought to weight notes that 85.2% of 2016 new vehicles brunt of the additional cost of owning sales taxes by new vehicle sales by state; were financed, as were 85.9% of 2015 more expensive new vehicles. For sake however, such data were unavailable. It new vehicle purchases.247 The analysis of simplicity, the model also assumes is recognized that for this purpose, new used data from Wards Automotive and vehicle sales by state is a superior JD Power on the average transaction that increasing the cost of new vehicles weighting mechanism to Census price of new vehicle purchases, average will not change the share of new vehicle population; in effort to approximate financed new auto beginning principal, MSRP that is financed; the relatively new vehicle sales by state, a study of the and the average incentive as a percent constant share from 2011–2016 when change in new vehicle registrations of MSRP to compute the ratio of the the average MSRP of a vehicle increased (using R.L. Polk data) by state across average financed new auto principal to 10% supports this assumption. It is recent years was conducted, resulting in the average new vehicle MSRP for recognized that this is not indicative of a corresponding set of weights. Use of calendar years 2011–2016. Table–II–34 average individual consumer the weights derived from the study of shows that the average financed auto transactions but provides a useful tool vehicle registration data resulted in a principal is between 82 and 84% of the to analyze the aggregate marketplace.

From Wards Auto data, the average average finance term length for new include higher interest rates. The share 48- and 60-month new auto interest autos was 68 months. It is recognized financed, interest rate, and finance term rates were 4.25% in 2016, and the that longer financing terms generally length are added as inputs in the

246 See Car Tax by State, locality rates, and lack of availability of weights to the effect of the exclusion of municipality-specific FactoryWarrantyList.com, http://www.factory apply to locality taxes complicate the ability to taxes from this analysis. warrantylist.com/car-tax-by-state.html (last visited reliably analyze the subject at this level of detail. 247 Zabritski, M. State of the Automotive Finance June 22, 2018). Note: County, city, and other Localities with relatively high automobile sales Market: A look at loans and leases in Q4 2016, municipality-specific taxes were excluded from taxes may have relatively fewer auto dealerships, as Experian, https://www.experian.com/assets/ weighted averages, as the variation in locality taxes consumers would endeavor to purchase vehicles in automotive/quarterly-webinars/2016-Q4-SAFM- within states, lack of accessible documentation of areas with lower locality taxes, therefore reducing revised.pdf (last visited June 22, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00096 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.049 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43081

parameters file so that they are easier to financing payments paid for the average update in the future. Using these inputs financed purchases as the following: the model computes the stream of

Note: The above assumes the interest this reason, it is not included in the new vehicle based upon increased new is distributed evenly over the period, societal cost and benefit analysis. vehicle prices in response to increased when in reality more of the interest is However, because it is an additional CAFE standard stringency. paid during the beginning of the term. cost paid by the consumer, it is However, the incremental amount calculated as a part of the private (c) Insurance Costs calculated as attributable to the standard consumer welfare analysis. More expensive vehicles will require will represent the difference in the It is recognized that increased finance more expensive collision and annual payments at the time that they terms, combined with rising interest comprehensive (e.g., fire and theft) car are paid, assuming that a consumer does rates, lead to a longer period of time not repay early. This will represent the before a consumer will have positive insurance. Actuarially fair insurance expected change in the stream of equity in the vehicle to trade in toward premiums for these components of financing payments at the time of the purchase of a newer vehicle. This value-based insurance will be the financing. has impacts in terms of consumers amount an insurance company will pay The above stream does not equate to either trading vehicles with negative out in the case of an incident type the average amount paid to finance the equity (thereby increasing the amount weighted by the risk of that type of purchase of a new vehicle. In order to financed and potentially subjecting the incident occurring. It is expected that compute this amount, the share of consumer to higher interest rates and/or the same driver in the same vehicle type financed transactions at each interest rendering the consumer unable to will have the same risk of occurrence for rate and term combination would have obtaining financing) or delaying the the entirety of a vehicle’s life so that the to be known. Without having replacement of the vehicle until they share of the value of a vehicle paid out projections of the full distribution of the achieve suitably positive equity to allow should be constant over the life of a auto finance market into the future, the for a trade. Comment is sought on the vehicle. However, the value of vehicles above methodology reasonably accounts effect these developments will have on will decline at some depreciation rate so for the increased amount of financing the new vehicle market, both in general, that the absolute amount paid in value- costs due to the purchase of a more and in light of increased stringency of related insurance will decline as the expensive vehicle, on an average basis fuel economy and GHG emission vehicle depreciates. This is represented taking into account non-financed standards. Comment is also sought on in the model as the following stream of transactions. Financing payments are whether and how the model should expected collision and comprehensive also assumed to be an intertemporal account for consumer decisions to insurance payments: transfer of wealth for a consumer; for purchase a used vehicle instead of a

To utilize the above framework, and comprehensive coverage was computed. Then each regulatory class in estimates of the share of MSRP paid on computed for 1979–2003. For cars the the fleet is weighted by share to estimate collision and comprehensive insurance share ranges from 49 to 55%, with the the overall average amount paid for and of annual vehicle depreciations are share tending to be largest towards the collision and comprehensive insurance needed to implement the above end of the series. For trucks the share by model year as shown in Table–II–35. equation. Wards has data on the average ranges from 43 to 61%, again, with the The average share of the initial MSRP annual amount paid by model year for share increasing towards the end of the paid in collision and comprehensive new light trucks and passenger cars on series. It is assumed that for model years insurance by model year is then collision, comprehensive and damage 2004–2016, 60% of insurance premiums computed. The average share paid for and liability insurance for model years for trucks, and 55% for cars, is paid for model years 2010–2016 is 1.83% of the 1992–2003; for model years 2004–2016, collision and comprehensive. Using initial MSRP. This is used as the share they only offer the total amount paid for these shares the absolute amount paid of the value of a new vehicle paid for insurance premiums. The share of total for collision and comprehensive collision and comprehensive in the insurance premiums paid for collision coverage for cars and trucks is future.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00097 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.050 EP24AU18.051 43082 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

2017 data from Fitch Black Book was cumulative share of the initial MSRP of If a vehicle lives to 10 years, 9.9% of the used as a source for vehicle depreciation a vehicle assumed to be paid in initial MSRP is expected to be paid in rates; two- to six-year-old vehicles in collision and comprehensive insurance collision and comprehensive payments; 2016 had an average annual in five-year age increments under this by 20 years 11.9% of the initial MSRP; depreciation rate of 17.3%.248 It is depreciation assumption, conditional on finally, if a vehicle lives to age 40, assumed that future depreciation rates a vehicle surviving to that age—that is, 12.4% of the initial MSRP. As can be will be like recent depreciation, and the the expected insurance payments at the seen, the majority of collision and analysis used the same assumed time of purchase will be weighted by comprehensive payments are paid by depreciation. Table–II–36 shows the the probability of surviving to that age. the time the vehicle is 10 years old.

The increase in insurance premiums model a 75-mile EV for early adopters, the economic value of the difference resulting from an increase in the average who we assume would not be concerned between the useful life of a conventional value of a vehicle is a result of an with the lower range, and a 150-mile EV ICE and the 150-mile EV when it increase in the expected amount for the broader market. The initial five reaches a 55% battery capacity (as a insurance companies will have to pay percent of EV sales were assumed to go result of battery deteroriation).250 In out in the case of damage occurring to to early adopters, with the remainder subsequent analyses (the 2016 Draft the driver’s vehicle. In this way, it is a being 150-mile EVs. The broader market TAR analysis and today’s analysis), cost to the private consumer, was assumed to have some lower utility NHTSA removed the low-range EVs attributable to the CAFE standard that for the 150-mile EV, due to the lower from its technology set due to both weak caused the price increase. driving range between refueling events consumer demand for low-range EVs in (d) Consumer Acceptance of Specific relative to a conventional vehicle. Thus, the marketplace and subsequent Technologies an additional social cost of about $3,500 technology advances that make 200-mile per vehicle was assigned to the EV150 EVs a more practical option for new EVs In previous rulemaking analyses, to capture the lost utility to produced in future model years. The NHTSA imposed an economic cost of consumers.249 Additionally, NHTSA exclusion of low-range EVs in the lost welfare to buyers of advanced imposed a ‘‘relative value loss’’ of technology set reduced the need to electric vehicles. NHTSA chose to 1.94% of the vehicle’s MSRP to reflect account for consumer welfare losses

248 Fitch Ratings Vehicle Depreciation Report 249 Based on Michael K. Hidrue, George R. 250 The vehicle was assumed to be retired once February 2017, Black Book, http:// Parsons, Willett Kempton, Meryl P. Gardner, the capacity reached 55 percent of its initial www.blackbook.com/wp-content/uploads/2017/02/ Willingness to pay for electric vehicles and their capacity, and the residual lifetime miles from that Final-February-Fitch-Report.pdf (last visited June attributes, Resource and Energy Economics,Volume point forward were valued, discounted, and 22, 2018). 33, Issue 3, 2011, Pages 686–705. expressed as a fraction of initial MSRP.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00098 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.052 EP24AU18.053 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43083

attributable to reduced driving range. confidential business information about versus some amount of electrification). While the sensitivity analysis explores the costs, revenues, and profitability of For instance, the dataset contained used some potential for continuing consumer their electrified vehicle lines. The CBI vehicle prices for the Honda Accord and value loss, even in the improved provided a valuable check on the Honda Accord Hybrid, Toyota Camry electrified powertrain vehicles, the empirical work described below. As a and Toyota Camry Hybrid, Ford Fusion central analysis assumes that no value result of this examination, we no longer and Ford Fusion Hybrid, Kia Soul and loss exists for electrified powertrains. assume manufacturers can pass on the Kia Soul EV, and so on for several However, ongoing low sales volumes entire incremental cost of hybrid, plug- model years. In some cases, the and a growing body of literature suggest in hybrid, and battery electric vehicles manufacturer produced no identically that consumer welfare losses may still to buyers of those vehicles. The equivalent internal combustion engine exist if manufacturers are forced to difference between the buyer’s vehicle, so a similar internal produce electric vehicles in place of willingness-to-pay for those combustion vehicle produced by the vehicles with internal combustion technologies, and the cost to produce same manufacturer was used as the engines (forcing sacrifices to cargo them, must be recovered from buyers of point of comparison. For example, the capacity or driving range) in order to other vehicles in a manufacturer’s Nissan Leaf was paired with the Nissan comply with standards. This topic will product portfolio or sacrificed from its Versa, as well as the Toyota Prius and receive ongoing investigation and profits, or sacrificed from dealership Toyota Corolla. Only vehicles available revision before the publication of the profits, or supplemented with State or for private sale, and in good vehicle final rule. Please provide comments and Federal incentives (or, some condition were included in the any relevant data that would help to combination of the four). analysis.254 The dataset contains fewer inform the estimation of Using data from the used vehicle observations for PHEVs and BEVs implementation of any value loss market, statistical models were fit to because manufacturers have produced related to sacrificed attributes in electric estimate consumer willingness to pay fewer examples of vehicles with these vehicles. for new vehicles with varying levels of technologies, compared to HEV and ICE One reason it was necessary to electrification relative to comparable vehicles. In all of these cases, trim level account for welfare losses from reduced internal combustion engine vehicles and options packages were matched driving range in this way is that, in was evaluated in four steps. The between ICE and electric powertrains to previous rulemakings, the agencies analysis (1) gathered used car fair minimize the degree of non-powertrain implicitly assumed that every vehicle in market value for select vehicles; (2) difference between vehicle pairs. The the forecast would be produced and developed regression models to estimate resale price data spanned many model purchased and that manufacturers the portion of vehicle depreciation rate years, but most observations in the would pass on the entire incremental attributable to the vehicle nameplate dataset represent MY 2013 through MY cost of fuel-saving technologies to new and the portion attributable to the 2016. car (and truck) buyers. However, many vehicle’s technology content at each age The regression models used to stakeholders commented that (using fixed effects for nameplates and estimate the transaction price (or consumers are not willing to pay the full specific electrification technologies); (3) ‘‘Value’’) as a function of age, control for incremental costs for hybrids, plug-in estimated the value of vehicles at age the type of powertrain (ICE, HEV, PHEV, 251 hybrids, and battery electric vehicles. zero (i.e., when the vehicles were new); and BEV) and nameplate to account for For this analysis, consumer willingness and (4) compared new vehicle values for their impact on the value of the vehicle to pay for HEVs, PHEVs, BEVs relative comparable vehicles across different as it ages.255 The regression takes the to comparable ICE vehicles was electrification levels (i.e., internal following form, with ICE, HEV, PHEV, investigated. The analysis compared the combustion, HEV, PHEV, and BEV) to and BEV binary variables (0, or 1), and estimated price premium the electrified estimate willingness-to-pay for the age defined as 2017 minus the model vehicles command in the used car electric technology relative to an ICE. year was used: market and estimated the willingness to The dataset used for estimation 1n(Value = ,b1(ICE * Age) + b2(HEV * pay premium for new vehicles with consisted of vehicle attribute data from b b electrification technologies at age zero Edmunds and transaction data from Age) + 3(PHEV * Age) + 4(BEV * Age) + b5(HEV) + b6(PHEV) + relative to their internal combustion Kelley Blue Book published online in b engine counterparts. For the analysis, June and July of 2017 for select vehicles 7(BEV) + FENameplate the willingness to pay was compared of interest.252 253 The dataset was For each observation in the dataset, with the expected incremental cost to constructed to contain pairs of vehicles the ‘‘Value’’ at age zero is determined by produce electrification technologies. that were nearly the same, except for setting the age variable to zero and Manufacturers also contributed type of powertrain (internal combustion solving.

251 See e.g., Comment by Alliance of Automobile 254 It is possible ‘‘good’’ vehicles for all ages may equivalent, and the used vehicle market for Tesla Manufacturers, Docket ID EPA–HQ–OAR–2015– have inadvertently introduced a small bias in the has not cleared in the same way because of a variety 0827–4089 and NHTSA–2016–0068–0072. sample, as a ‘‘good’’ conditioning rating on a of unique business factors (previously, Tesla 252 See Edmunds, https://www.edmunds.com/ vehicle just a year or two old may not be in average guaranteed resale value prices for their products, condition relative to other vehicles of the vintage, (last visited June 22, 2018). Edmunds publishes which was a factory incentive program that only but a ‘‘good’’ rating for a much older car may reflect automotive data, reviews, and advice. an impeccably maintained vehicle. recently ended, no longer applying to vehicles sold 253 See Kelley Blue Book, https://www.kbb.com/ 255 In the case of electrified vehicles with no after July 1, 2016). These two factors impaired the (last visited June 22, 2018). Kelley Blue Book, part internal combustion engine equivalent, the analysis quality of used Tesla data for the purposes of the of Cox Automotive’s Autotrader brand, provides grouped like vehicle pairs together under the same analysis, so the agencies excluded Tesla vehicles automotive research, reviews, and advice, including nameplate fixed effects (or FENameplate). Tesla from today’s analysis on customer willingness-to- estimated market values of new and used vehicles. vehicles have no internal combustion engine pay for electrified vehicles.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00099 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.333 43084 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The estimated willingness-to-pay for price premium for new vehicles with incremental cost to produce these electrified powertrain packages over an HEV, PHEV, and BEV technologies. This vehicles. Specifically, the analysis internal combustion engine in an analysis suggests that consumers are estimated consumers are willing to pay otherwise similar vehicle is computed willing to pay more for new electrified between $2,000 and $3,000 more for the as the difference between their vehicles than their new internal engine electrified powertrains considered here estimated initial values, using the combustion counterparts, but only a than their internal combustion engine functions above. These pair-wise little more, and not necessarily enough counterparts. differences are averaged to estimate a to cover the relatively large projected

Table–II–37 illustrates the variation in luxury vehicles in the set retained value the statistical model, we estimate that willingness-to-pay by electrification in a way that skewed the average. The strong hybrids hold less than $100 of level (although the statistical model did CBI acquired from manufacturers was the initial price premium by age eight not distinguish between PHEV30 and more consistent with the mean than (on average). While the battery electric PHEV50 due to the small number of median value (except for the PHEVs). vehicles appear to be worth less than available operations for plug-in Additionally, the Kelley Blue Book their ICE counterparts by age eight, hybrids). As the table demonstrates, the data suggest that the used electrified there is limited data about this emerging difference between the median and vehicles were often worth less than their segment of the new vehicle market. mean predicted price premium for used internal combustion engine These independently-produced results PHEVs is significant. The limited counterpart vehicles after a few years of using publicly available data were in number of PHEV observations were not use.256 As Table–II–38 illustrates, the line with manufacturers’ reported uniformly distributed among the value of the price premium shrinks as confidential business information. nameplates present, and some of the the vehicles age and depreciate. Using

The ‘‘technology cost burden’’ business information received from equivalent markups (with a large numbers used in today’s analysis manufacturers any tax credits that can portion of those costs being batteries), represent the amount of a given be passed through in the price, and the but the estimated willingness-to-pay is technology’s incremental cost that cost of the technology. In general, the only about $3,000. While tax credits manufacturers are unable to pass along incremental willingness-to-pay falls offset some, if not most of that to the buyer of a given vehicle at the well short of the costs currently difference for PHEVs and BEVs, there is time of purchase. The burden is defined projected for HEVs, PHEVs, and BEVs; some residual amount that buyers of as the difference between estimated for example, BEV technology can add new electrified vehicles are currently willingness-to-pay, itself a combination roughly $18,000 in equipment costs to unwilling to cover, and that must either of the estimated values and confidential the vehicle after standard retail price come from forgone profits or be passed

256 The analysis did not identify an underlying interaction between maintenance costs and batteries superior to previous generation vehicles, and new reason for this observation, but the agencies posit or maintenance costs and low volume vehicles. electrified vehicles may be offered at lower prices for discussion purposes there could be some Alternatively, new electrified vehicles may be still because of a variety of market conditions.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00100 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.054 EP24AU18.055 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43085

along to buyers of other vehicles in a both the initial purchase price of the not include ICE vehicles with no similar manufacturer’s portfolio. vehicle and its later value in the used HEV, PHEV, or BEV nameplate or Manufacturers may be able to recover vehicle market. As such, the analysis counterpart, so the analysis presented some or all of these costs by charging would continue to be relevant even if here looks at a small portion of all higher prices for their other models, in incentive programs for vehicle transactions and is more likely to which case it will represent a welfare electrification change or phase out in include fuel efficient models where loss to buyers of other vehicles (even if the future. By considering a variety of market demand for hybrid (or higher) not to buyers of HEVs, PHEVs, or BEVs nameplates and body styles produced versions may exist. One possible themselves). To the extent that they are by several manufacturers, this analysis alternative is to rely on new vehicle unable to do so and must absorb part or produces average willingness-to-pay transaction prices to estimate consumer all of these costs, their profits will estimates that can be applied to the willingness-to-pay for new vehicles decline, and in effect this cost will be whole industry. By evaluating matched with certain attributes. However, new borne by their investors. In practice, the pairs of vehicles from the same vehicle transaction data is highly analysis estimates benefits and costs to manufacturer, the analysis accounts for proprietary and difficult to obtain in a car and light truck manufacturers and many additional factors that may be tied form that may be disclosed to the buyers under the assumption that each to the brand, rather than the technology, public. manufacturer recovers all technology and influence the fair market price of While estimating willingness-to-pay costs and civil penalties it incurs from vehicles. In particular, the data for electrification technologies from buyers via higher average prices for the inherently include customer valuations depreciation and MSRP data is models it produces and sells, although for fuel-savings and vehicle appealing, many manufacturers handle sufficient information to support maintenance schedules, as well as other MSRP and pricing strategies differently, specific assumptions about price factors like noise-vibration-and- with some preferring to deviate only a increases for individual models is not harshness, interior space,257 and fueling little from sticker price and others present. In effect, this means that any convenience in the context of the preferring to offer high discounts. There part of a manufacturer’s costs to convert vehicles considered. is evidence of large differences between specific models to electric drive There are some limitations to this MSRP and effective market prices to technologies that it cannot recover by approach. There are currently few consumers for many vehicles, especially charging higher prices to their buyers observations of PHEV and BEV BEVs. Please provide comments on methods will be borne collectively by buyers of technologies in the data, and most of the and data used to evaluate consumer the other models they produce. Each of observations for BEVs are sedans and willingness-to-pay for electrification those buyers is in effect assumed to pay small cars, the values for which are technologies. a slight premium (or ‘‘markup’’) over the extrapolated to other market segments. manufacturer’s cost to produce the Additionally, the used vehicle data (e) Refueling Surplus models they purchase (including the supporting these estimates inherently Direct estimates of the value of cost of any technology used to improve includes both older and newer extended vehicle range are not available its fuel economy), this premium on generations of technology, so the in the literature, so the reduction in the average is modeled to recover the full historical regression may be slow to required annual number of refueling cost of technology applied to all react to rapid changes in the new cycles due to improved fuel economy vehicles to improve the fuel economy of vehicle marketplace. As new vehicle was calculated and the economic value the fleet. So, even though electrified nameplates emerge, and existing of the resulting benefits assessed. Chief vehicles are modeled as if their buyers nameplates improve their among these benefits is the time that are unwilling to pay the full cost of the implementation of electrification owners save by spending less time both technology associated with their fuel technologies, this model will require re- in search of fueling stations and in the economy improvement, the price borne estimation to determine how these new act of pumping and paying for fuel. by the average new vehicle buyer entrants impact the estimated industry The economic value of refueling time represents the average incremental average willingness-to-pay. savings was calculated by applying technology cost for all applied Additionally, the willingness-to-pay DOT-recommended valuations for travel technology, the sum of all technology analysis does not consider electric time savings to estimates of how much costs divided by the number of units vehicles with no direct ICE counterpart. time is saved.258 The value of travel sold, across all classes, for each For example, today’s evaluation does time depends on average hourly manufacturer. not consider Tesla because the Tesla valuations of personal and business The willingness-to-pay analysis brand has no ICE equivalent, and time, which are functions of total hourly described above relies on used vehicle because the free-market prices for used compensation costs to employers. The data that is widely available to the Tesla vehicles have been difficult (if not total hourly compensation cost to public. Market tracking services update impossible) to obtain, primarily due to employers, inclusive of benefits, in used vehicle price estimates regularly as factory guaranteed resale values (which 2010$ is $29.68.259 Table–II–39 below fuel prices and other market conditions is a program that still affects the used demonstrates the approach to estimating change, making the data easy to update market for many Tesla vehicles). Still, the value of travel time ($/hour) for both in the future as market conditions Tesla vehicles have a large share of the urban and rural (intercity) driving. This change. The used vehicle data also BEV market by both unit sales and approach relies on the use of DOT- account for consumer willingness-to- dollar sales, it may be possible to recommended weights that assign a pay absent State and Federal rebates at include Tesla data in a future update to lesser valuation to personal travel time the time of sale, which are reflected in this analysis. Similarly, the analysis did than to business travel time, as well as

257 Often HEVs and PHEVs place batteries in 258 See https://www.transportation.gov/sites/ 259 Total hourly employer compensation costs for functional storage space, such as the trunk or floor dot.gov/files/docs/ValueofTravelTime 2010 (average of quarterly observations across all storage bins, thereby forcing consumers to trade-off Memorandum.pdf (last accessed July 3, 2018). occupations for all civilians). See https:// fuel-savings with other functional vehicle www.bls.gov/ncs/ect/tables.htm (last accessed July attributes. 3, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00101 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43086 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

weights that adjust for the distribution between personal and business travel.

The estimates of the hourly value of time—independent of urban or rural adjustment is performed separately for urban and rural travel time ($15.67 and status—may be produced. passenger cars and for light trucks, yielding $21.93, respectively) shown in Table–II– occupancy-adjusted valuations of vehicle Note: The calculations above assume only travel time during refueling trips for each 39 above must be adjusted to account one adult occupant per vehicle. To fully fleet. for the nationwide ratio of urban to rural estimate the average value of vehicle travel driving. By applying this adjustment (as time, the presence of additional adult Note: Children (persons under age 16) are shown in Table–II–40 below), an overall passengers during refueling trips must be excluded from average vehicle occupancy estimate of the hourly value of travel accounted for. The analysis applies such an counts, as it is assumed that the opportunity adjustment as shown in Table–II–40; this cost of children’s time is zero.

260 Time spent on personal travel during rural the divergence in these values: (1) Time is scarcer food, and entertainment because time at the (intercity) travel is valued at a greater rate than that on a long trip; (2) a long trip involves destination is worth such high costs. of urban travel. There are several reasons behind complementary expenditures on travel, lodging,

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00102 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.056 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43087

The analysis estimated the amount of Among these characteristics of fueling which interview respondents indicated refueling time saved using (preliminary) station visits is the total amount of time the primary reason was due to a low survey data gathered as part of our spent pumping and paying for fuel. reading on the gas gauge.265 This 2010–2011 National Automotive From a separate sample (also part of the restriction was imposed so as to exclude Sampling System’s Tire Pressure TPMS study), researchers conducted drivers who refuel on a fixed (e.g., Monitoring System (TPMS) study.263 interviews at the pump to gauge the weekly) schedule and may be unlikely The study was conducted at fueling distances that drivers travel in transit to to alter refueling patterns as a result of stations nationwide, and researchers and from fueling stations, how long that increased driving range. The relevant made observations regarding a variety of transit takes, and how many gallons of TPMS survey data on average refueling characteristics of thousands of fuel are being purchased. trip characteristics are presented below individual fueling station visits from This analysis of refueling benefits in Table–II–41. August 2010 through April 2011.264 considers only those refueling trips

As an illustration of how the value of level in the absence of a higher CAFE their tanks are 35% full (i.e. as shown extended refueling range was estimated, standard for the given model year). in Table–II–41, with 7.0 gallons in assume a small light truck model has an TPMS survey data indicate that drivers reserve, and the consumer purchases 13 average fuel tank size of approximately who indicated the primary reason for gallons). By this measure, a typical 20 gallons and a baseline actual on-road their refueling trips was a low reading driver would have an effective driving fuel economy of 24 mpg (its assumed on the gas gauge typically refuel when range of 312 miles (= 13.0 gallons × 24

261 See Travel Monitoring, Traffic Volume Trends, on the TPMS study. TPMS data are preliminary at vehicles. For more information on the National U.S. Department of Transportation Federal Highway this time, and rates are subject to change pending Automotive Sampling System and to access TPMS Administration, https://www.fhwa.dot.gov/policy availability of finalized TPMS data. Average data when they are made available, see http:// l information/travel monitoring/tvt.cfm (last visited occupancy rates shown here are specific to www.nhtsa.gov/NASS. June 22, 2018). Weights used for urban versus rural refueling trips and do not include children under 264 The data collection period for the TPMS study travel are computed using cumulative 2011 16 years of age. ranged from October 10, 2010, through April 15, estimates of urban versus rural miles driven 263 TPMS data are preliminary and not yet provided by the Federal Highway Administration. published. Estimates derived from TPMS data are 2011. 262 Source: National Automotive Sampling therefore preliminary and subject to change. 265 Approximately 60% of respondents indicated System 2010–2011 Tire Pressure Monitoring System Observational and interview data are from distinct ‘‘gas tank low’’ as the primary reason for the (TPMS) study. See next page for further background subsamples, each consisting of approximately 7,000 refueling trip in question.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00103 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.057 EP24AU18.058 43088 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

mpg) before he or she is likely to refuel. will continue to do. Based on peer the number of miles (VMT) driven by Increasing this model’s actual on-road reviewer comments to NHTSA’s initial those that stay in service, and for the fuel economy from 24 to 25 mpg would implementation of refueling time percentage of refueling trips that occur therefore extend its effective driving savings (subsequent to the CAFE NPRM for reasons other than a low reading on range to 325 miles (= 13.0 gallons × 25 issued in 2011), the analysis of refueling the gas gauge; a discount rate is also mpg). Assuming that the truck is driven time savings was updated for the final applied in the valuation of future 12,000 miles/year,266 this one mpg rule to reflect peer reviewer benefits. Using this direct estimation improvement in actual on-road fuel suggestions.269 Beyond updating time approach to quantify the value of this economy reduces the expected number values to current dollars, that analysis benefit by model year was considered; of refueling trips per year from 38.5 (= has been used, unchanged, in today’s however, it was concluded that the 12,000 miles per year/312 miles per analysis as well. value of this benefit is implicitly refueling) to 36.9 (= 12,000 miles per Because a reduction in the expected captured in the separate measure of year/325 miles per refueling), or by 1.6 number of annual refueling trips leads overall valuation of fuel savings. refuelings per year. If a typical fueling to a decrease in miles driven to and Therefore, direct estimates of this cycle for a light truck requires a total of from fueling stations, the value of benefit are not added to net benefits 6.83 minutes, then the annual value of consumers’ fuel savings associated with calculations. It is noted that there are time saved due to that one mpg this decrease can also be calculated. As other benefits resulting from the improvement would amount to $3.97 (= shown in Table–II–41, the typical reduction in miles driven to and from (6.83/60) × $21.81 × 1.6). incremental round-trip mileage per fueling stations, such as a reduction in In the central analysis, this refueling cycle is 1.08 miles for light greenhouse gas emissions—CO2 in calculation was repeated for each future trucks and 0.97 miles for passenger cars. particular—which, as per the case of calendar year that light-duty vehicles of Going back to the earlier example of a fuel savings discussed in the preceding each model year affected by the light truck model, a decrease of 1.6 in paragraph, are implicitly accounted for standards considered in this rule would the number of refuelings per year leads elsewhere. remain in service. The resulting to a reduction of 1.73 miles driven per Special mention must be made with cumulative lifetime valuations of time year (= 1.6 refuelings × 1.08 miles regard to the value of refueling time savings account for both the reduction driven per refueling). Again, if this savings benefits to owners of electric over time in the number of vehicles of model’s actual on-road fuel economy and plug-in electric (both referred to a given model year that remain in was 24 mpg, the reduction in miles here as EV) vehicles. EV owners who service and the reduction in the number driven yields an annual savings of routinely drive daily distances that do of miles (VMT) driven by those that stay approximately 0.07 gallons of fuel (= not require recharging on-the-go may in service. The analysis also adjusts the 1.73 miles/24 mpg), which at $3.25/ eliminate the need for trips to fueling or value of time savings that will occur in gallon 270 results in a savings of $0.23 charging stations. It is likely that early future years both to account for per year to the owner. adopters of EVs will factor this benefit expected annual growth in real into their purchasing decisions and Note: This example is illustrative only of wages 267 and to apply a discount rate to maintain driving patterns that require the approach used to quantify this benefit. In once-daily at-home recharging (a determine the net present value of time practice, the societal value of this benefit saved.268 A further adjustment is made process which generally takes five to excludes fuel taxes (as they are transfer 271 to account for evidence from the payments) from the calculation and is eleven hours for a full charge) for interview-based portion of the TPMS modeled using fuel price forecasts specific to those EV owners who have purchased study which suggests that 40% of each year the given fleet will remain in and installed a Level Two charging refueling trips are for reasons other than service. station to a high-voltage outlet at their a low reading on the gas gauge. It is The annual savings to each consumer home or parking place. However, EV therefore assumed that only 60% of the shown in the above example may seem owners who regularly or periodically theoretical refueling time savings will like a small amount, but the reader need to drive distances further than the be realized, as it was assumed that should recognize that the valuation of fully-charged EV range may need to owners who refuel on a fixed schedule the cumulative lifetime benefit of this recharge at fixed locations. A savings to owners is determined distributed network of charging stations 266 2009 National Household Travel Survey separately for passenger car and light (e.g., in parking lots, at parking meters) (NHTS), U.S Department of Transportation Federal truck fleets and then aggregated to show may allow some EV owners to recharge Highway Administration at 48 (June 2011), the net benefit across all light-duty their vehicles while at work or while available at http://nhts.ornl.gov/2009/pub/stt.pdf. shopping, yet the lengthy charging 12,000 miles/year is an approximation of a light vehicles, which is much more duty vehicle’s annual mileage during its initial significant at the macro level. cycles of current charging technology decade of use (the period in which the bulk of Calculations of benefits realized in may pose a cost to owners due to the benefits are realized). The CAFE model estimates future years are adjusted for expected value of time spent waiting for EVs to VMT by model year and vehicle age, taking into real growth in the price of gasoline, for charge and potential EV shoppers who account the rebound effect, secular growth rates in do not have access to charging at home VMT, and fleet survivability; these complexities are the decline in the number of vehicles of omitted in the above example for simplicity. a given model year that remain in (e.g., because they live in an apartment 267 See The Economics Daily, The compensation- service as they age, for the decrease in without a vehicle charging station, only productivity gap, U.S. Department of Labor Bureau of Labor Statistics (Feb. 24, 2011), http:// 271 269 Peer review materials, peer reviewer See generally All-New Nissan Leaf Range & www.bls.gov/opub/ted/2011/ted_20110224.htm. A backgrounds, comments, and NHTSA responses for Charging, Nissan USA, https://www.nissanusa.com/ 1.1% annual rate of growth in real wages is used this prior assessment are available at Docket vehicles/electric-cars/leaf/range-charging.html (last to adjust the value of travel time per vehicle NHTSA–2012–0001. visited June 22, 2018); Home Charging Calculator, ($/hour) for future years for which a given model Tesla, https://www.tesla.com/support/home- is expected to remain in service. This rate is 270 Estimate of $3.25/gallon is the forecasted cost charging-calculator (last visited June 22, 2018); supported by a BLS analysis of growth in real wages per gallon (including taxes, as individual 2018 Chevrolet Bolt EV, GM, https://media.gm.com/ from 2000–2009. consumers consider reduced tax expenditures to be content/media/us/en/chevrolet/vehicles/bolt-ev/ 268 Note: Here, as elsewhere in the analysis, savings) for motor gasoline in 2025. Source of price 2018/_jcr_content/iconrow/textfile/file.res/2018- discounting is applied on an annual basis from CY projections: U.S. Energy Information Chevrolet-Bolt-EV-Product-Guide.pdf (last visited 2017. Administration, Annual Energy Outlook Early 2018. June 22, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00104 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43089

have street parking, or have garages with during the Great Recession. The 2017 to calculate each average odometer insufficient voltage). Moreover, EV NHTS was not available at the time of reading by make, model, and model year owners who primarily recharge their this rulemaking. Because of the age of are representative of the total vehicles at home will still experience the last available NHTS and the unusual population of vehicles of that type. some level of inconvenience due to their economic conditions under which it Several indicators suggest that this is a vehicle being either unavailable for was collected, NHTSA built the new reasonable assumption. unplanned use or to its range being schedule using a similar method from a First, the majority of vehicle make/ limited during this time should they proprietary dataset collected in the fall models is well-represented in the interrupt the charging process. of 2015. This new data source has the sample. For more than 85% of make/ Therefore, at present EVs hold potential advantages of both being newer, a larger model combinations, the average in offering significant time savings but sample, and collected by a third party. odometer readings are collected for 20% only to owners with driving patterns or more of the total population. Most (1) Data Sources and Estimation (Polk optimally suited for EV characteristics. make/model observations have Odometer Data) If fast-charging technologies emerge and sufficient sample sizes, relative to their a widespread network of fast-charging To develop new mileage representation in the vehicle stations is established, it is expected accumulation schedules for vehicles population, to produce meaningful that a larger segment of EV vehicle regulated under the CAFE program average odometer totals at that level. owners will fully realize the potential (classes 1–3), NHTSA purchased a data Second, we considered whether the refueling time savings benefits that EVs set of vehicle odometer readings from representativeness of the odometer offer. This is an area of significant IHS/Polk (Polk). Polk collects odometer sample varies by vehicle age because uncertainty. readings from registered vehicles when VMT schedules in the CAFE model are they encounter maintenance facilities, specific to each age. It is possible that, 6. Vehicle Use and Survival state inspection programs, or for some of those models, an insufficient To properly account for the average interactions with dealerships and number of odometer readings is value of consumer and societal costs OEMs—these readings are more likely to recorded to create an average that is and benefits associated with vehicle be precise than the self-reported likely to be representative of all of those usage under various CAFE and GHG odometer readings collected in the models in operation for a given year. For alternatives, it is necessary to estimate NHTS. The average odometer readings all model years other than 2015, the portion of these costs and benefits in the data set NHTSA purchased are approximately 95% or more of vehicles that will occur at each age (or calendar based on more than 74 million unique types are represented by at least five year) for each model year cohort. Doing odometer readings across 16 model percent of their population. For this so requires some estimate of how many years (2000–2015) and vehicle classes reason, observations from all model miles the average vehicle of a given present in the data purchase (all years, other than 2015, were included in type 272 is expected to drive at each age registered vehicles less than 14,000 lbs. the estimation of the new VMT and what share of the initial model year GVW). This sample represents schedules. cohort is expected to remain at each age. approximately 28% of the light-duty Because model years are sold in in the The first estimates are referred to as the vehicles registered in 2015, and thus has Fall of the previous calendar year, vehicle miles travelled (VMT) schedules the benefit of not only being a newer, throughout the same calendar year, and and the second as the survival rate but also, a larger, sample. even into the following calendar year— schedules. In this section the data Comparably to the NHTS, the Polk not all registered vehicles of a make/ sources and general methodologies used data provide a measure of the model/model year will have been to develop these two essential inputs are cumulative lifetime vehicle miles registered for at least a year (or more) briefly discussed. More complete traveled (VMT) for vehicles, at the time until age three. The result is that some discussions of the development of both of measurement, aggregated by the MY 2014 vehicles may have been driven the VMT schedules and the survival rate following parameters: Make, model, for longer than one year, and some less, schedules are present in the PRIA model year, fuel type, drive type, door at the time the odometer was observed. Chapter 8. count, and ownership type (commercial In order to consider this in the or personal). Within each of these definition of age, an age of a vehicle is (a) Updates to Vehicle Miles Traveled subcategories they provide the average assigned to be the difference between Schedules Since 2012 FR odometer reading, the number of the average reading date of a make/ The MY 2017–2021 FRM built odometer readings in the sample from model and the average first registration estimates of average lifetime mileage which Polk calculated the averages, and date of that make/model. The result is accumulation by body style and age the total number of that subcategory of that the continuous age variable reflects using the 2009 National Household vehicles in operation. the amount of time that a car has been Travel Survey (NHTS), which surveys In estimating the VMT models, each registered at the time of odometer odometer readings of the vehicles data point was weighted (make/model reading and presumably the time span present from the approximately 113,000 classification) by the share of each that the car has accumulated the miles. households sampled. Approximately make/model in the total population of After creating the ‘‘age’’ variable, the 210,000 vehicles were in the sample of the corresponding vehicle body style. analysis fits the make/model lifetime self-reported odometer readings This weighting ensures that the VMT data points to a weighted quartic collected between April 2008 and April predicted odometer readings, by body polynomial regression of the age of the 2009. This represents a sample size of style and model year, represent each vehicle (stratified by vehicle body less than one percent of the more than vehicle classification among observed styles). The predicted values of the 250 million light-duty vehicles vehicles (i.e., the vehicles for which quartic regressions are used to calculate registered in 2008 and 2009. The NHTS Polk has odometer readings), based on the marginal annual VMT by age for sample is now 10 years old and taken each vehicles’ representation in the each body style by calculating registered vehicle population of its body differences in estimated lifetime mileage 272 Type here refers to the following body styles: style. Implicit in this weighting scheme accumulation by age. However, the Polk Pickups, vans/SUVs, and other cars. is the assumption that the samples used data acquired by NHTSA only contains

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00105 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43090 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

observations for vehicles newer than 16 (2) Using New Schedules in the CAFE vehicles analyzed within the CAFE years of age. In order to estimate the Model/Analysis Model simulation. schedule for vehicles older than the age As discussed in Section 0, recent While the Polk registration data set 15 vehicles in the Polk data, information literature supports a 20% ‘‘rebound contains odometer readings for about that portion of the schedule from effect’’ for light-duty vehicle use, which individual vehicles, the CAFE model represents an elasticity of annual use the VMT schedules used in both the tabulates ‘‘mileage accumulation’’ 2017–2021 Final Light Duty Rule and with respect to fuel cost per mile of schedules, which relate average annual ¥0.2. Because fuel cost per mile is 2019–2025 Medium-Duty NPRM was miles driven to vehicle age, based on combined. The light-duty schedules calculated as fuel price per gallon vehicles’ body style. For the purposes of divided by fuel economy (in miles per were derived from the survey data VMT accounting, the CAFE model contained in the 2009 National gallon), this same elasticity applies to classifies vehicles in the analysis fleet as changes in fuel cost per mile that result Household Travel Survey (NHTS). being one of the following: Passenger from variation in fuel prices or From the old schedules, the annual car, SUV, pickup truck, passenger van, differences in fuel economy. It suggests 273 VMT is expected to be decreasing for all or medium-duty pickup/van. In order that a five percent reduction in the cost ages. Towards the end of the sample, the to use the Polk data to develop VMT per mile of travel for vehicles of a predictions for annual VMT increase. In schedules for each of these vehicle certain body style will result in a one order to force the expected classes in the CAFE model, a mapping percent increase in the average number monotonicity, a triangular smoothing between the classification of each model of miles they are driven annually. algorithm is performed until the in the Polk data and the classes in the The average cost per mile (CPM) of a schedule is monotonic. This performs a CAFE model was first constructed. This vehicle of a given age and vehicle style mapping enabled separate tabulations of weighted average which weights the in CY 2016 (the first analysis year of the average annual miles driven at each age observations close to the observation simulation) was used as the reference for each of the vehicle classes included point to calculate the rebound effect more than those farther from it. The in the CAFE model. result is a monotonic function, that within the CAFE model. However, this The only revision made to the predicts similar lifetime VMT for the does not perfectly align with the time of mappings used to construct the new sample span as the original function. the collection of the Polk dataset. The VMT schedules was to merge the SUV Because the analysis does not have data Polk data were collected in 2015 (so that and passenger van body styles into a 2014 fuel prices were the last to beyond 15 years of age, it is not able to single class. These body styles were correctly capture that part of the annual influence sampled vehicles’ odometer merged because there were very few readings), and represents the average VMT curve using only the new dataset. examples of vans—only 38 models were odometer reading at a single point in For this reason, trends in the old data in use during 2014, where every other time for age (model year) included in to extrapolate the new schedule for ages body style had at least three times as the cross-section. We use the difference beyond the sample range are used. many models. Further, as shown in the in the average odometer reading for each To use the VMT information from the PRIA Chapter 8, there was not a vintage during 2014 to calculate the newer data source for ages outside of the significant difference between the 2009 number of miles vehicles are driven at sample, final in-sample age (15 years) NHTS van and SUV mileage schedules, each age (see PRIA Chapter 8 for are used as a seed and then applied to nor was there a significant difference specific details on the analysis). For the proportional trend from the old between the schedules built with the example, we interpret the difference in schedules to extrapolate the new two body styles merged or kept separate the average odometer reading between schedules out to age 40. To do this, the using the 2015 Polk data. Merging these the five- and six-year-old vehicles of a annual percentage difference in VMT of body styles does not change the given body style as the average number the old schedule for ages 15–40 is workings of the CAFE model in any of miles they are driven during the year way, and the merged schedule is simply calculated. The same annual percentage when they were five years old. entered as an input for both vans and difference in VMT is applied to the new However, vehicles produced during SUVs. schedule to extend beyond the final in- different model years do not have the sample value. This assumes that the Although there is a single VMT by age same average fuel economy, so it is overall proportional trend in the outer schedule used as an input for each body important to consider the average fuel style, the assumptions about the economy of each vintage (or model year) years is correctly modeled in the old rebound effect require that this schedule used to measure mileage accumulation VMT schedule and imposes this same be scaled for future analysis years to at a given age when scaling VMT for the trend for the outer years of the new reflect changes in the cost of travel from rebound calculation. schedule. The extrapolated schedules the time the Polk sample was originally The first step in doing so is to adjust are the final input for the VMT collected. These changes result from for any change in average annual use schedules in the CAFE model. PRIA both changes in fuel prices between the that would have been caused by Chapter 8 contains a lengthier time the sample was collected and any differences in fuel prices between CYs discussion of both the data source and future analysis year and differences in 2014 and 2016. This is done by scaling the methodology used to create the new fuel economy between the vehicles the original schedules of annual VMT schedules. included in the sample used to build the by age tabulated from the Polk sample mileage schedules and the future-year using the following equation:

273 Though not included in today’s analysis, and vans were developed using the same corresponding schedules for heavy-duty pickups methodology.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00106 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43091

Here, the average fuel economy for (2) the difference between the average differences in fuel cost per mile that vehicles of a given body style and age fuel economy for vehicles of that body result from changes in fuel prices or refers to a different MY in 2016 than it type and age during a future calendar from differences in fuel economy. The did in 2014; for example, a MY 2014 year and the average fuel economy for model then repeats this calculation for vehicle had reached age two vehicle vehicles of that same body type and age each calendar year during the lifetimes during CY 2016, whereas a 2012 model during 2016. These two factors combine of vehicles of other body types, and year vehicle was age two during CY to determine the average fuel cost per subsequently repeats this entire set of 2014. mile for vehicles of that body type and calculations for each regulatory To estimate the average annual use of age during each future calendar year alternative under consideration. The vehicles of a specified body type and and the average fuel cost per mile for resulting differences in average annual age during future calendar years under vehicles of that same body type and age use of vehicles of each body type at each a specific regulatory alternative, the during 2016. age interact with the number estimated CAFE model adjusts the resulting The elasticity of annual vehicle use to remain in use at that age to determine estimates of vehicle use by age for that with respect to fuel cost per mile is total annual VMT by vehicles of each body type during CY 2016 to reflect (1) applied to the difference between these body type. the projected change in fuel prices from two values because vehicle use is This adjustment is defined by the 2016 to each future calendar year; and assumed to respond identically to equation below:

This equation uses the observed cost vehicles of each given type and age The mileage estimates described per mile of a vehicle of each age and relative to CPM resulting from the above are a crucial input in the CAFE style in CY 2016 as the reference point average fuel economy of vehicles of that model’s calculation of fuel consumption for all future calendar years. That is, the type and age and observed fuel prices and savings, energy security benefits, reference fuel price is fixed at 2016 during the year when the mileage consumer surplus from cheaper travel, levels, and the reference fuel economy accumulation schedules were originally recovered refueling time, tailpipe of vehicles of each age is fixed to the measured. This is consistent with a emissions, and changes in crashes, average fuel economy of the vintage that definition of the rebound effect as the fatalities, noise and congestion. had reached that age in 2016. For elasticity of annual vehicle use with (3) Comparison to other VMT example, the reference CPM for a one- respect to changes in the fuel cost per projections (2012 FR, AEO average year-old SUV is always the CPM of the mile of travel, regardless of the source lifetime miles, totals?) average MY 2015 SUV in CY 2016, and of changes in fuel cost per mile. the CPM for a two-year-old SUV is Alternative forms of referencing are Across all body styles and ages, the always the CPM of the average possible, but none can guarantee that previous VMT schedules estimate MYv2014 SUV in CY 2016. projected future vehicle use will higher average annual VMT than the This referencing ensures that the respond to both projected changes in updated schedules. Table–II—42 model’s estimates of annual mileage fuel prices and differences in individual compares the lifetime VMT under the accumulation for future calendar years models’ fuel economy among regulatory 2009 NHTS and the 2015 Polk dataset. reflect differences in the CPM of alternatives. The 40-year lifetime VMT gives the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00107 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.059 EP24AU18.060 43092 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

expected lifetime VMT of a vehicle The dynamic survival-weighted lifetime proportional reduction than the conditional on surviving to age 40. The VMT represents the expected number of unweighted lifetime assumptions. Using new schedules predict between 24 and miles driven by each body style, the updated schedules, the expected 31% fewer miles for a 40-year old weighting by the dynamic survival lifetime mileage accumulation is vehicle depending on the body style. schedules under baseline between approximately 150k and 170k The new schedules predict that the assumptions.274 There is a similar miles depending on the body style, average 40-year old vehicle will drive proportional reduction in expected rather than the approximately 180k to between approximately 260k and 280k lifetime VMT under both survival 210k miles under the previous miles depending on the body style assumptions, with the dynamic schedules. For more detail on when the versus between approximately 350k and scrappage model predicting lifetime mileage and survival rates occur, mileage accumulation within 10,000 380k for the previous schedules. chapter 8 of the PRIA gives the full VMT The static survival-weighted lifetime miles of the previous static model under schedules by age. The section below VMT represents the expected number of both VMT schedules. The expected miles the average vehicle of each body lifetime mileage accumulation reduces gives further estimates of how lifetime style will drive, weighting by the between 13 and 15% under the current VMT estimates vary under different likelihood it survives to each age using VMT schedules when compared to the assumptions within the dynamic the previous static scrappage schedules. previous schedules—a smaller scrappage model.

We have several reasons for preferring reported information). Further, the 2009 next VMT schedule iteration, NHTSA the new VMT schedules over the prior NHTS represents approximately one intends to use panel data to test the iterations. Before discussing these percent of the sample of vehicles magnitude of any attrition effect that reasons, it is important to note that registered in 2008/2009, while the 2015 may exist. While this caveat is NHTSA uses the same general Polk dataset represents approximately important, all previous iterations were methodology in developing both 30% of all registered light-duty vehicles; also built from a single calendar year schedules. We consider data on average it is a much larger dataset, and less cross-section and contain the same odometer readings by age and body style likely to oversample certain vehicles. inherent bias. collected once during a given window Additionally, while the NHTS may be a (b) How does CAFE affect vehicle of time; we then estimate a weighted representative sample of households, it retirement rates? polynomial function between vehicle is less likely to be a representative age and lifetime accumulation for a sample of vehicles. However, by Lightly used vehicles are a close given vehicle style. As with the properly accounting for vehicle substitute for new vehicles; thus, there previous schedules, we use the inter- population weights in the new averages is relationship between the two markets. annual differences as the estimate of and models, we corrected for this issue As the price for new vehicles increases, annual miles traveled for a given age. in the derivation of the new schedules. there is an upward shift in the demand The primary advantage of the current Importantly, this methodology treats for used vehicles. As a result of the schedules is the data source. The the cross-section of ages in a single upward shift in the demand curve, the previous schedules are based on data calendar year as a panel of the same equilibrium price and quantity of used that is outdated and self-reported, while model year vehicle, when in reality each vehicles both increase; the value of used the observations from Polk are between age represents a single model year, and vehicles increases as a result. The five and seven years newer than those not a true panel. We have some concern decision to scrap or maintain a used in the NHTS and represent valid that where the most heavily driven vehicle is closely linked with the value odometer readings (rather than self- vehicles drop out of the sample that the of the vehicle; when the value is lesser lifetime odometer readings will be lower than the cost to maintain the vehicle, it 274 In estimating the dynamic survival rate to than they would be if the scrapped will be scrapped. In general, as a result weight the annual VMT schedules, we make the vehicles had been left in the dataset of new vehicle price increases, the following input assumptions: The reference vehicle is MY 2016, GDP growth rates and fuel prices are without additional mileage scrappage rate, or the proportion of our central estimates, and the future average new accumulation. This would bias our vehicles remaining on the road vehicle fuel economies by body style and overall estimates of inter-annual mileage unregistered in a given year, of used average new vehicle prices are those simulated by accumulation downward and may result vehicles will decline. Because older the CAFE model when CAFE standards are omitted (by setting standards at 1 mpg), such that only in an undervaluation of costs and vehicles are on average less efficient and technologies that pay back within 30 months are benefits associated with additional less safe, this will have important applied. travel for vehicles of older ages. For the implications for the evaluations of costs

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00108 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.061 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43093

and benefits of fuel economy standards, in incremental fatalities will decrease reduced form to approximate used car which increase the cost of new vehicles the societal net benefits of increasing scrappage in response to increasing fuel and reduce the average cost per mile of new vehicle fuel economy standards. economy standards. fuel costs. Our previous estimates of vehicle Greenspan & Cohen (1996) offer Fuel economy standards result in the scrappage did not include a dynamic additional foundations from which to application of more fuel saving response to new vehicle price, but think about vehicle stock and scrappage. technologies for at least some models, recent literature has continued to Their work identifies two types of which result in a higher cost for illustrate that this an omission which scrappage: Engineering scrappage and manufacturers to produce otherwise could rival the rebound effect in cyclical scrappage. Engineering identical vehicles. This increase in magnitude (Jacobsen & van Bentham, scrappage represents the physical wear production cost amounts to an upward 2015). For this reason, we worked to on vehicles, which results in their being shift in the supply curve for new develop an econometric survival model scrapped. Cyclical scrappage represents vehicles. This increases the equilibrium which captures the effect of increasing the effects of macroeconomic conditions price and reduces the quantity of the price of new vehicles on the survival on the relative value of new and used vehicles demanded. While the cost of rate of used vehicles discussed in the vehicles; under economic growth the new vehicles increases under increased following sections and in more detail in demand for new vehicles increases and fuel economy standards, the fuel cost the PRIA Chapter 8. We discuss the the value of used vehicles declines, per mile of travel declines. Consumers literature on vehicle scrappage rate and resulting in increased scrappage. In will place some value on the fuel discuss in the succeeding section. A addition to allowing new vehicle prices savings associated with the additional brief explanation of why we develop our to affect cyclical vehicle scrappage a` la technology, to the extent that they value own models and the data sources and the Gruenspecht effect, Greenspan and reduced operating expenses against the econometric estimations we use to do Cohen also note that engineering increased price of a new vehicle, so, follows. We conclude the discussion scrappage seems to increase where EPA increased financing costs (and of the updates to vehicle survival emission standards also increase; as impediments to obtaining financing), estimates with a summary of the results, more costs goes towards compliance and increased insurance costs. a description of how we use them in the technologies, it becomes more There is a trade-off between fuel CAFE model, and finally, how the expensive to maintain and repair more economy and other attributes that updated schedules compare with the complicated parts, and scrappage consumers value such as: Vehicle previous static scrappage schedules. increases. In this way, Greenspan and performance, interior volume, etc. Cohen identify two ways that fuel Where the additional value of fuel (1) What does the literature say about economy standards could affect vehicle savings associated with a technology is the relationship? scrappage: (1) Through increasing new greater than any loss of value from (a) How Fuel Economy Standards vehicle prices, thereby increasing used trade-offs with other attributes, the Impact Vehicle Scrappage vehicle prices, and finally, reducing on- demand for new vehicles will also shift road vehicle scrappage, and (2) by The effects of differentiated upwards. Where the additional shifting resources towards fuel-saving regulation 275 in the context of fuel evaluation of fuel savings is lesser than technologies—potentially reducing the economy (particularly, emission any loss of value from changes to other durability of new vehicles by making standards only affecting new vehicles) attributes, the demand will shift them more complex. downwards. Thus, the direction of the was discussed in detail in Gruenspecht demand shift is unknown. However, if (1981) and (1982), and has since been (b) Aggregate vs. Atomic Data Source in we assume that manufacturers pass all coined the ‘‘Gruenspecht effect.’’ the Literature costs associated with a model off to the Gruenspecht recognized that because One important distinction between consumer of that vehicle, then the per fuel economy standards affect only new the literatures on vehicles scrappage is vehicle profit remains constant. If we vehicles, any increase in price (net of between those that use atomic vehicle also assume that manufacturers are good the portion of reduced fuel savings data, data following specific individual predictors of the valuation and elasticity valued by consumers) will increase the vehicles, and those that use some level of certain vehicle attributes, then we can expected life of used vehicles and of aggregated data, data that counts the assume that even if there is some reduce the number of new vehicles total number of vehicles of a given type. positive demand shift, it is not enough entering the fleet. In this way, increased The decision to scrap a vehicle is an to increase demand above the original fuel economy standards slow the atomic one—that is, made on an equilibrium levels, or manufacturers turnover of the fleet and the entrance of individual vehicle basis. The decision would apply those technologies even in any regulated attributes tied only to new relates to the cost of maintaining a the absence of regulation. vehicles. Although Gruenspecht vehicle, and the value of the vehicle As noted above, the increase in the acknowledges that a structural model both on the used car market, and as price of new vehicles will result in which allows new vehicle prices to scrap metal. Generally, a used car owner increased demand for used vehicles as affect used vehicle scrappage only will decide to scrap a vehicle where the substitutes, extending the expected age through their effect on used vehicle value of the vehicle is less than the and lifetime vehicle miles travelled of prices would be preferable, the data value of the vehicle as scrap metal plus less efficient, and generally, less safe available on used vehicle prices was the cost to maintain or repair the vehicles. The additional usage of older (and still is) limited. Instead he tested vehicle. In other words, the owner gets vehicles will result in fewer gallons his hypothesis in his 1981 dissertation more value from scrapping the vehicle saved and more total on-road fatalities using new vehicle price and other than continuing to drive it or from under more stringent CAFE alternatives. determinants of used car prices as a selling it. For more on the topic of safety, the Recent work is able to model relative safety of specific model year 275 Differentiated regulations are regulations scrappage as an atomic decision due to affecting segments of the market differently; here, vehicles is discussed in Section 0 of the it references the fact that emission and fuel the availability of a large database of preamble and PRIA Chapter 11. Both the economy standards have largely only applied to used vehicle transactions. Following erosion of fuel savings and the increase new and not used vehicles. works by other authors including:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00109 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43094 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Busse, Knittel, & Zettelmeyer (2013); repair prices to impute some measure of ‘durability’ factor as a function of model Sallee, West, & Fan (2010); Alcott & the effect of new vehicle prices on year vintage is included. Wozny (2013); and Li, Timmins, & von vehicle scrappage. (d) Models of the Gruenspecht Effect Haefen (2009)—Jacobsen & van Benthem (c) Historical Trends in Vehicle Used in Other Policy Analyses (2015) considers the impact of changes Durability in gasoline prices on used vehicle This is not the first estimation of the values and scrappage rates. In turn, they Waker (1968); Park (1977); Feeney & ‘Gruenspecht Effect’ for policy consider the impact of an increase in Cardebring (1988); Hamilton & considerations. In their Technical used vehicle values on the scrappage Macauley (1999); and Bento, Ruth, & Support Document (TSD) for the 2004 rate of those vehicles. They find that Zhuo (2016) all note that vehicles proposal to reduce greenhouse gas increases in gasoline price result in a change in durability over time. Walker emissions from motor vehicles, reduction in the scrappage rate of the (1968) simply notes a significant California Air Resources Board (CARB) most fuel efficient vehicles and an distinction in expected vehicle lifetimes outlines how they utilized the CARBITS increase in the scrappage rate of the pre- and post-World War I. Park (1977) vehicle transaction choice model in an least fuel efficient vehicles. This has discusses a ‘durability factor’ set by the attempt to capture the effect of important implications for the validity producer for each year so that different increasing new vehicle prices on vehicle of the average fuel economy values vintages and makes will have varying replacement rates. They consider data linked to model years and assumed to expected lifecycles. Feeney & from the National Personal be constant over the life of that model Cardebring (1988) show that durability Transportation Survey (NPTS) as a year fleet within this study. Future of vehicles appears to have generally source of revealed preferences and a iterations of this study could further increased over time both in the U.S. and University of California (UC) study as a investigate the relationship between fuel Swedish fleets using registration data source of stated preferences for the economy, vehicle usage, and scrappage, from each country. They also note that purchase and sale of household fleets as noted in other places in this the changes in median lifetime between under different prices and attributes discussion. the Swedish and U.S. fleet track well, (including fuel economy) of new While the decision to scrap a vehicle with a 1.5 year lag in the U.S. fleet. This vehicles. is made atomically, the data available to lag is likely due to variation in how the The transaction choice model NHTSA on scrappage rates and data is collected—the Swedish vehicle represents the addition and deletion of variables that influence these scrappage registry requires a title to unregister a a vehicle from a household fleet within rates are aggregate measures. This vehicle, and therefore gets immediate a short period of time as a influences the best available methods to responses, where the U.S. vehicle ‘‘replacement’’ of a vehicle, rather than measure the impacts of new vehicle registry requires re-registration, which as two separate actions. Their final data prices on existing vehicle scrappage. creates a lag in reporting. set consists of 790 vehicle replacements, The result is that this study models Hamilton & Macauley (1999) argue for 292 additions, and 213 deletions; they aggregate trends in vehicle scrappage a clear distinction between embodied do not include the deletions, but assume and not the atomic decisions that make versus disembodied impacts on vehicle any vehicle over 19 years old that is up these trends. Many other works longevity. They define embodied sold is scrapped. This allows them to within the literature use the same data impacts as inherent durability similar to capture a slowing of vehicle source and general scrappage construct, Park’s producer supplied ‘durability replacement under higher new vehicle such as: Walker (1968); Park (1977), factor’ and Greenspan’s ‘engineering prices, but because their model does not Greene & Chen (1981); Gruenspecht scrappage’ and disembodied effects include deletions, does not explicitly (1981); Gruenspecht (1982); Feeney & those which are environmental, not model vehicle scrappage, but assumes Cardebring (1988); Greenspan & Cohen unlike Greenspan and Cohen’s ‘cyclical all vehicles aged 20 and older are (1996); Jacobsen & van Bentham (2015); scrappage.’ They use calendar year and scrapped rather than resold. They and Bento, Roth, & Zhuo (2016) all use vintage dummy variables to isolate the calibrate the model so that the overall the same aggregate vehicle registration effects—concluding that the fleet size is benchmarked to Emissions data as the source to compute vehicle environmental factors are greater than FACtors (EMFAC) fleet predictions for scrappage. any pre-defined ‘durability factor.’ Some the starting year; the simulation then Walker (1968) and Bento, Roth, & of their results could be due to some produces estimates that match the Zhuo (2016) use aggregate data to inflexibility of assuming model year EMFAC predictions without further directly compute the elasticity of coefficients are constant over the life of calibration. scrappage from measures of used a vehicle, and there may be some The CARB study captures the effect vehicle prices. Walker (1968) uses the correlation between the observed life of on new vehicle prices on the fleet ratio of used vehicle Consumer Price the later model years of their sample replacement rates and offers some Index (CPI) to repair and maintenance and the ‘stagflation’ 276 of the 1970’s. precedence for including some estimate CPI. Bento, Roth, & Zhuo (2016) use Bento, Ruth, & Zhuo (2016) find that the of the Gruenspecht Effect. One used vehicle prices directly. While the average vehicle lifetime has increased important thing to note is that because direct measurement of the elasticity of 27% from 1969 to 2014 by sub-setting vehicles that exited the fleet without scrappage is preferable in a theoretical their data into three model year cohorts. replacement were excluded, the effect of sense, the CAFE model does not predict To implement these findings in the new vehicle prices on scrappage rates future values of used vehicles, only scrappage model incorporated into the where the scrapped vehicle is not future prices of new vehicles. For this CAFE model, this study takes pains to replaced is not captured. Because new reason, any model compatible with the estimate the effect of durability changes and used vehicles are substitutes, it is current CAFE model must estimate a in such a way that the historical expected that used vehicle prices will reduced form similar to Park (1977); durability trend can be projected into increase with new vehicle prices. Gruenspecht (1981); Greenspan & Cohen the future; for this reason, a continuous Because higher used vehicle prices will (1996), who use some form of new lower the number of vehicles whose vehicle prices or the ratio of new 276 Continued high inflation combined with high cost of maintenance is higher than their vehicle prices to maintenance and unemployment and slow economic growth. value, it is expected that not only will

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00110 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43095

replacements of used vehicles slow, but represent the number of vehicles of that preference for more fuel efficient and also, that some vehicles that would have vintage and type scrapped in a given expensive new vehicles, are not used. been scrapped without replacement year. There were a couple other As further justification for using the under lower new vehicle prices will important data considerations for the NADA price series over the BLS New now remain on the road because their calculations of the historical scrappage Vehicle CPI, Park (1977) cites a value will have increased. Aggregate rates not discussed here but discussed discontinuity found in the amount of measures of the Gruenspecht effect will in detail in the PRIA Chapter 8.277 quality adjustments made to the series include changes to scrappage rates both For historical data on vehicle so that more adjustments are made over from slower replacement rates, and transaction prices, the models use data time. This could further limit the ability slower non-replacement scrappage rates. from the National Automobile Dealers for the BLS New Vehicle CPI to predict changes in vehicle scrappage. (2) Description of Data Sources Association (NADA), which records the average transaction price of all light- Vehicle scrappage rates are also NHTSA purchases proprietary data on duty vehicles. These transaction prices influenced by fuel economy and fuel the registered vehicle population from represent the prices consumers paid for prices. Historical data on the fuel IHS/Polk for safety analyses. IHS/Polk new vehicles but do not include any economy by vehicle style from model has annual snapshots of registered value of vehicles that may have been years 1979–2016 was obtained from the vehicle counts beginning in calendar traded in to dealers. Importantly, these 2016 EPA Motor Trends Report.279 The year (CY) 1975 and continuing until transaction prices were not available by van/SUV fuel economy values represent calendar year 2015. The data includes vehicle body styles; thus, the models a sales-weighted harmonic average of the following regulatory classes as will miss any unique trends that may the individual body styles. Fuel prices defined by NHTSA: Passenger cars, light have occurred for a particular vehicle were obtained from Department of trucks (classes 1 and 2a), and medium body style. This may be particularly Energy (DOE) historical values, and and heavy-duty trucks (classes 2b and relevant for pickup trucks, which future fuel prices within the CAFE 3). Polk separates these vehicles into observed considerable average price model use the Annual Energy Outlook 280 another classification scheme: Cars and increases as luxury and high option (AEO) future oil price projections. trucks. Under their schema, pickups, pickups entered the market. Future From these values the average cost per vans, and SUVs are treated as trucks, models will further consider 100 miles of travel for the cohort of new and all other body styles are included as incorporating price series that consider vehicles in a given calendar year and cars. In order to build scrappage models the price trends for cars, SUVs and vans, the average cost per 100 miles of travel to support the model year (MY) 2021– and pickups separately.278 for each used model year cohort in that 2026 light duty vehicle (LDV) standards, 281 The models use the NADA price same calendar year are computed. It it was important to separate these series rather than the Bureau of Labor is expected that as the new vehicle fleet vehicle types in a way compatible with Statistics (BLS) New Vehicle Consumer becomes more efficient (holding all the existing CAFE model. other attributes constant) that it will be There were two compatible choices to Price Index (CPI), used by Park (1977) and Greenspan & Cohen (1997), because more desirable, and the demand for aggregate scrappage rates: (1) By used vehicles should decrease regulatory class or (2) by body style. the BLS New Vehicle CPI makes quality adjustments to the new vehicle prices. (increasing their scrappage). As a given Because for NHTSA’s purposes vans/ model year cohort becomes more SUVs are sometimes classified as BLS assumes that additions of safety and fuel economy equipment are a expensive to operate due to increases in passenger cars and sometimes as light fuel prices, it is expected the scrappage trucks, and there was no quick way to quality adjustment to a vehicle model, which changes the good and should not of that model year will increase. It is reclassify some SUVs as passenger cars perhaps worth noting that more efficient within the Polk dataset, NHTSA chose be represented as an increase in its price. While this is good for some model year vintages will be less to aggregate survival schedules by body susceptible to changes in fuel prices, as style. This approach is also preferable purposes, it presumes consumers fully value technologies that improve fuel because NHTSA uses body style specific 279 Light-Duty Automotive Technology, Carbon lifetime VMT schedules. Vehicles economy. Because it is the purpose to Dioxide Emissions, and Fuel Economy Trends: 1975 experience increased wear with use; this study to measure whether this is Through 2016, U.S. EPA (Nov. 2016), available at many maintenance and repair events are true, it is important that vehicle prices https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey= closely tied to the number of miles on adjusted to fully value fuel economy P100PKK8.pdf. improving technologies, which would 280 Note: The central analysis uses the AEO a vehicle. The current version of the reference fuel price case, but sensitivity analysis CAFE model considers separate lifetime obscure the ability to measure the also considers the possibility of AEO’s low and high VMT schedules for cars, vans/SUVs, fuel price cases. 277 pickups and classes 2b and 3 vehicles. The first is any discontinuity caused by a 281 Work by Jacobsen and van Bentham suggests change in how Polk collected their data beginning that these initial average fuel economy values may These vehicles are assumed to serve in calendar year 2010, and the second is the use of not represent the average fuel economy of a model different purposes and, as a result, are the adjustment described in Greenspan & Cohen year cohort as it ages—mainly, they find that the modelled to have different average (1996). most fuel efficient vehicles scrap earlier than the lifetime VMT patterns. These different 278 Note: Using historical data aggregated by body least fuel efficient models in a given cohort. This styles to capture differences in price trends by body may be an important consideration in future uses likely also result in different style does not require the assertion technology costs endeavors that work to link fuel economy, vehicle lifetime scrappage patterns. are or are not borne by the body style to which they miles travelled (VMT), and scrappage. Studies on Once stratified into body style level are applied. If the body-style level average price ‘‘the rebound effect’’ suggest that lowering the fuel buckets, the data can be aggregated into change is used, then the assumption is cost per driven mile increases the demand for VMT. manufacturers do not cross-subsidize across body With more miles, a vehicle will be worth less as its population counts by vintage and age. styles, whereas if the average price change is used perceived remaining useful life will be shorter; this These counts represent the population then the assumption is they would proportion costs will result in the vehicle being more likely to be of vehicles of a given body style and equally for each vehicle. These are implementation scrapped. A rebound effect is included in the CAFE vintage in a given calendar year. The questions to be worked out once NHTSA has a model, but because reliable data on how average historical data source separating price series by VMT by age has varied over calendar year and difference between the counts of a given body styles, but these do not matter in the current model year vintage is not available, expected vintage and vehicle type from one model which only considers the average price of all lifetime VMT is not included within the current calendar year to the next is assumed to light-duty vehicles. dynamic scrappage model.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00111 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43096 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

absolute changes in their cost per mile and Cohen deem ‘engineering consistent trend in new vehicle prices. will be smaller. The functional forms of scrappage.’ This source of scrappage is The optimal number of lags is three so the cost per mile measures are further largely determined by the age of a that the price trend from the current discussed in the model specification vehicle and the durability of a specific year and the three prior years influences subsection 3 below. model year vintage. Vehicle scrappage the demand for and scrappage of used Aggregate measures that cyclically typically follows a roughly logistic vehicles. Note: The average lease length affect the value of used vehicles include function with age—that is, is three years 283 so that the price of an macroeconomic factors like the real instantaneous scrappage increases to average vehicle coming off lease is interest rate, the GDP growth rate, some peak, and then declines, with age estimated to affect the scrappage rate of unemployment rates, cost of as noted in Walker (1968); Park (1977); used vehicles—this is a major source of maintenance and repairs, and the value Greene & Chen (1981); Gruenspecht the newest used vehicles that enter the of a vehicle as scrap metal or as parts. (1981); Feeney & Cardebring (1988); used car fleet. Further, because Here only the GDP growth rate is Greenspan & Cohen (1996); Hamilton & increases in new vehicle prices due to discussed, as this is the only measure Macauley (1999); and Bento, Roth, & increased stringency of CAFE standards included in the final model. Extended Zhuo (2016). Thus, this analysis also is the primary mechanism through reasoning as to why other variables are uses a logistic function to capture this which CAFE standards influence not included in the final model in the trend of vehicle scrappage with age but vehicle scrappage and the CAFE Model PRIA Chapter 8 is offered, but the allows non-linear terms to capture any assumes that usage, efficiency, and discussion was omitted here for brevity skew to the logistic relationship. safety vary with the age of the vehicle, in describing only the final model. Specific details about the final and particular attention is paid to the form Generally economic growth will result considered forms of engineering of this effect. It is important to know the in a higher demand for new vehicles— scrappage by body styles is presented in likelihood of scrappage by the age of the cars in aggregate are normal goods—and the PRIA Chapter 8. vehicle to correctly account for the The final and considered independent a reduction in the value of used additional costs of additional fatalities variables intended to capture cyclical vehicles. The result should be an and increased fuel consumption from elements of vehicle scrappage and the increase in the scrappage rate of existing deferred scrappage. Thus, the influence considered forms of each are discussed vehicles so that we expect the GDP of increasing new vehicle prices is growth rate to be an important predictor in PRIA Chapter 8; here only inclusion of the GDP growth rate is discussed. The allowed to influence the demand for of vehicle scrappage rates. used vehicles (and reduce their NHTSA sourced the GDP growth rate GDP growth rate is not a single-period scrappage) differently for different ages from the 2017 OASDI Trustees effect; both the current and previous of vehicles in the scrappage model. We Report.282 The Trustees Report offers GDP growth rates will affect vehicle discuss both how we determined the credible projections beyond 2032. scrappage rates. A single year increase correct form and number of lags for each Because the purpose of building this will affect scrappage differently than a body style in PRIA Chapter 8. scrappage model is to project vehicle multi-period trend. For this reason, an survival rates under different fuel optimal number of lagged terms are The final cyclical factor affecting economy alternatives and the current included: The within-period GDP vehicle scrappage in the preferred fuel economy projections go as far growth rate, the previous period GDP model is the cost per 100 miles of travel forward as calendar year 2032, using a growth rate, and the growth rate from both of new vehicles and of the vehicle data set that encompasses projections at two prior years for the car model, while which is the subject of the decision to least through 2032 is an essential for vans/SUVs, and pickups, the current scrap or not to scrap. The new vehicle characteristic of any source used for this and previous period GDP growth rate cost per 100 miles is defined as the ratio analysis. are sufficient. of the average fuel price faced by new Similarly, the considered model vehicles in a given calendar year and (3) Summary of Model Estimation allows that one-period changes in new the average new vehicle fuel economy The most predictive element of vehicle prices will affect the used for 100 miles in the same calendar year, vehicle scrappage is what Greenspan vehicle market differently than a and varies only with calendar year:

The cost per 100 miles of the price faced by that model year vintage that model year when it was new, and potentially scrapped vehicle is in a given calendar year and the average varies both with calendar year and described as the ratio of the average fuel fuel economy for 100 miles of travel for model year:

282 The 2017 Annual Report of the Board of available at https://www.ssa.gov/oact/tr/2017/ dealers.edmunds.com/static/assets/articles/lease- Trustees of the Federal Old-Age and Survivors tr2017.pdf. report-jan-2017.pdf. Insurance and Federal Disability Insurance Trust 283 See e.g., Edmunds January 2017 Lease Market Funds, Social Security Administration (2017), Report, Edmunds (Jan. 2017), https://

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00112 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.062 EP24AU18.063 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43097

The average per-gallon fuel price model year fleet for the given calendar of fuel types present in a given model faced by a model year vintage in a given year, weighted by the share of each fuel year vintage: calendar year is the annual average fuel type in that model year fleet. Or the price of all fuel types present in that following, where FT represents the set

For these variables, the best fit model section VII.H.1 of this document. In increase. However, given the includes the cost per mile of both the order to isolate the effect, the price of distribution of the mileage new and the used vehicle for the current new vehicles is held constant at CY accumulation schedule by age, this and prior year. This is congruent with 2016 levels. The specific methodology amounts only to a two percent increase research that suggests consumers used to do so is described in detail in in the expected lifetime mileage respond to current fuel prices and fuel PRIA Chapter 8, as is the leakage accumulation of an individual vehicle. price changes. The selection process of implied by comparing the reference and This range is consistent with DOT this form for the cost per mile and the no Gruenspecht effect sensitivity cases. expectations in terms of direction and implications is discussed in PRIA It is important to note here that the magnitude. Chapter 8. leakage calculated ranges between 12 The use of a static retirement There are a couple other controlling and 18% across regulatory alternatives. schedule, while deemed a reasonable factors considered in our final model. This is in line with Jacobsen & van approach in the past, is a limited The 2009 Car Allowance Rebate System Bentham (2015) estimates which put representation of scrappage behavior. It (CARS) is not outlined here but is leakage for their central case between 13 fails to account for increasing vehicle outlined in PRIA Chapter 8. This and 16%. Their high gasoline price case durability—occurring for the last several program aimed to accelerate the is more in line this analysis’ central decades—and the resulting increase in retirement of less fuel efficient vehicles case—with fuel prices of $3/gallon—and average vehicle age in the on-road fleet, and replace them with more fuel predicts leakage of 21%. This further which has nearly doubled since 1980.284 efficient vehicles. Further discussion of validates the scrappage model effects Thus, turning off the dynamic scrappage how this is controlled for is located in against examples in the literature. model described above would not PRIA Chapter 8. Finally, evidence of The models used for this analysis are impose a perspective on the analysis autocorrelation was found, and able to capture the relationship for that is neutral with respect to observed including three lagged values of the vehicle scrappage as it varies with age scrappage behavior but would instead dependent variable addresses the and how this relationship changes with represent a strong assumption that concern. Treatment of autocorrelation is increases to new vehicle price, the cost asserts important trends in the historical discussed in PRIA Chapter 8. per mile of travel of new and used record will abruptly cease or change One additional issue encountered in vehicles, and how the rate varies direction. the estimations of scrappage rates is that cyclically with the GDP growth rate. It As discussed above, the dynamic the models predict too many vehicles also controls for the CARS program and scrappage model implemented to remain on the road in the later years. checks the influence of a change in support this proposal affects total fleet This issue occurs because the data Polk’s data collection procedures. The size through several mechanisms. beyond age 15 are progressively more goodness of fit measures and the Although the model accounts for the sparsely populated; vehicles over 15 plausibility of the predictions of the influence of changes to average new years were not captured in the Polk data model are discussed at some length in vehicle price and U.S. GDP growth, the until 1994, when each successive PRIA Chapter 8. In the next section, the most influential mechanism, by far, is collection year added an additional age impacts of updating the static scrappage the observed trend of increasing vehicle of vehicles until 2005 when all ages models to the dynamic models on durability over successive model years. began to be collected. This means that average vehicle age and usage, by body This phenomenon is prominently for vehicles over the age of 25 there are styles, and across different regulatory discussed in the academic literature only 10 years of data. In order to correct assumptions are discussed. related to vehicle retirement, where for this issue the fact that the final fleet there is no disagreement about its share converges to roughly the same (c) What is the estimated effect on existence or direction.285 In fact, when share for most model years for a given vehicle retirement and how do results the CAFE model is exercised in a way vehicle type is used. The predicted compare to previously estimated fleets that keeps average new vehicle prices at versus historical relationships seem to and VMT? (approximately) MY 2016 levels, the on- deviate beginning around age 20; thus, The expected lifetime of a car road fleet grows from an initial level of for scrappage rates for vehicles beyond estimated using the static scrappage 228 million in 2016 to 340 million in age 20 an exponential decay function schedule from the 2012 final rule, both 2050, an increase of 49% over the 35- which guarantees that by age 40 the in years and miles, is between the year period from 2016 to 2050. final fleet share reaches the convergence expected lifetime of the dynamic The historical data show the size of level observed in the historical data is scrappage model in the absence of CAFE the registered vehicle population (i.e., applied. The application of the decay standards and under the baseline the on-road fleet) growing by about 60% function and mathematical definition is standards. Estimated by the dynamic in the 35 years between 1980 and further defended in PRIA Chapter 8. scrappage model, the average vehicle is A sensitivity case is also developed to expected to live 15.1 years under the 284 Based on data from FHWA and IHS/Polk. 285 Waker (1968); Park (1977); Feeney & isolate the magnitude of the influence of only market demand for Cardebring (1988); Hamilton & Macauley (1999); Greunspecht effect. The impacts on new technology, and 15.6 years under and Bento, Ruth, & Zhuo (2016) note that vehicles costs and benefits are presented in the baseline scenario, a four percent change in durability over time.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00113 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.064 43098 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

2015.286 In the 35 years between 2016 observed over the course of decades are model relies on its own estimated and 2050, our simulation shows the on- likely to persist. Analyzing fuel values as the previous scrappage rates at road fleet growing from about 230 economy standards requires an earlier ages, forcing any estimation million vehicles to about 345 million understanding of the mechanisms that errors to propagate through to future vehicles when the market adopts only influence new vehicle sales, the size of years. This exercise is discussed further the amount of fuel economy, which it the on-road fleet, and vehicle miles in PRIA Chapter VII. While the years of naturally demands. The simulated traveled. It is upon these mechanisms the recession represent a significant growth over this period is about 50% that the policy acts: Increasing/ shock to the size of the fleet, briefly from today’s level, rather than the 60% decreasing new vehicle prices changes reversing many years of annual growth, observed in the historical data over the the rate at which new vehicles are sold, the model recovers quickly and last 35 years. Under the baseline changing the attributes and prices of produces results within one percent of regulatory scenario, the growth over the these vehicles influences the rates at the actual fleet size, as it did prior to the next 35 years is simulated to be about which all used vehicles are retired, the recession. 54%—still short of the observed growth overall size of the on-road fleet In order to compare the magnitudes of over a comparable period of time. In determines the total amount of VMT, the sales and scrappage effects across fact, the simulated annual growth rate in which in turn affects total fuel different fuel economy standards the size of the on-road fleet in this consumption, fatalities, and other considered it is important to define analysis, about 1.3%, is lower than the externalities. The fact that DOT’s comparable measures. The sales effect long-term average annual growth rate of bottom-up approach produces results in in a single calendar year is simply the about two percent dating back to the line with historical trends is both difference in new vehicle sales across 1970s.287 expected and intended. alternatives. However, the scrappage Additionally, there are inherent This is not to say that all details of effect in a single calendar year is not precision limitations in measuring this new approach will be immediately simply the change in fleet size across something as vast and complex as the intuitive for reviewers accustomed to regulatory alternatives. The scrappage registered vehicle population. For results that do not include a dynamic model predicts the probability that a decades, the two authoritative sources sales model or dynamic scrappage vehicle will be scrapped in the next year for the size of the on-road fleet have model, much less results that combine conditional on surviving to that age; the been R.L. Polk (now IHS/Polk) and the two. For example, some reviewers absolute probability that a vehicle FHWA. For two decades these two may observe that today’s analysis shows survives to a given age is conditional on sources differed by more than 10% each that, compared to the baseline the scrappage effect for all previous year, only lately converging to within a standards, the proposed standards analysis years. In other words, if few percent of each other. These produce a somewhat smaller on-road successive calendar years observe lower discrepancies over the correct fleet (i.e., fewer vehicles in service) average new vehicle prices, the effect of interpretation of the data by each source despite somewhat increased sales of increased scrappage on fleet size will have consistently represented new vehicles (consistent with reduced accumulate with each successive differences of more than 10 million new vehicle prices) and decreased calendar year—because fewer vehicles vehicles. prices for used vehicles. While it might survived to previous ages, the same The total number of new vehicles be natural to assume that reduced prices probability of scrappage would result in of new vehicles and increased sales a smaller fleet size for the following year projected to enter the fleet is slightly should lead to a larger on-road fleet, in as well, though fewer vehicles will have higher than the historical trend (though our modelling, the increased sales are been scrapped than in the previous year. the impact of the great recession makes more than offset by the somewhat To isolate the number of vehicles not it hard to say by how much). More accelerated scrappage that accompanies scrapped in a single calendar year generally, the projections used in the the estimated decrease in new vehicle because of the change in standards, the analysis cover long periods of time prices. This outcome represents an on- first step is to calculate the number of without exhibiting the kinds of road fleet that is both smaller and a little vehicles scrapped in every calendar year fluctuation that are present in the younger on average (relative to the for both the proposed standards and the historical record. For example, the baseline) and ‘‘turns over’’ more baseline; this is calculated by the inter- forecast of GDP growth in our analysis quickly. annual change in the size of the used posits a world in which the United To further test the validity of the vehicle fleet (vehicles ages 1–39) for States sees uninterrupted positive scrappage model, a dynamic forecast each alternative. The difference in this annual growth in real GDP for four was constructed for calendar years 2005 measure across regulatory alternatives decades. The longest such period in the through 2015 to see how well it predicts represents the change in vehicle historical record is 17 years and still the fleet size for this period. The last scrappage because of a change in the included several years of low (but true population the scrappage model standards. The resulting scrappage positive) growth during that interval. ‘‘sees’’ is the 2005 registered vehicle effect for a single calendar year can be Over such a long period of time, in population. It then takes in known compared to the difference across the absence of deep insight into the production volumes for the new model regulatory alternatives in new vehicle future of the U.S. auto industry, it is year vehicles and dynamically estimates sales for the same calendar year as a sensible to assume that the trends instantaneous scrappage rates for all comparison of the relative magnitudes registered vehicles at each age for CYs of the two effects. In most years, under 286 There are two measurements of the size of the 2006–2015, based only on the observed the proposed standards relative to the registered vehicle population that are considered to be authoritative. One is produced by the Federal exogenous values that inform the model baseline standards, the analysis shows Highway Administration, and the other by R.L. Polk (GDP growth rate, observed new vehicle that for each additional new vehicles (now part of IHS). The Polk measurement shows prices, and cost per mile of operation), sold, two to four used vehicles are fleet growth between 1980 and 2015 of about 85%, fleet attributes of the vehicles (body removed from the fleet. Over the time while the FHWA measurement shows a slower growth rate over that period, only about 60%. style, age, cost per mile of operation), period of the analysis these predicted 287 Based on calculations using Polk’s National and estimated scrappage rates at earlier differences in the numbers of vehicles Vehicle Population Profile (NVPP). ages. Within this exercise, the scrappage accumulate, resulting in a maximum of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00114 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43099

seven million fewer vehicles by CY vehicle to the used car market if the scrappage rates does not respond to the 2033 for the proposed CAFE standards price of the vehicle is greater than the quantity of new vehicles sold. As one relative to the augural standards, and cost of supplying it. Lowering the potential option for development, the nine million fewer vehicles by CY 2035 equilibrium price of used vehicles will potential for an integrated model of for the proposed GHG standards relative lower the increase the number of sales and scrappage, or for a dynamic to the current GHG standards. Tables scrapped vehicles, lowering the supply connection between the two models will 11–29 and 11–30 in the PRIA show the of used vehicles, and decreasing the be considered. Comment is sought on difference in the fleet size by calendar equilibrium quantity. The change in both the sales and scrappage models, on year for the proposed standards relative new vehicle sales is related to demand potential alternatives, and on data and to the augural standards for the CAFE of new vehicles at a given price, but the methods that may enable practicable and GHG programs, respectively. change in used vehicle scrappage is integration of any alternative models To understand why the sales and related to the shift in the demand curve into the CAFE model. scrappage effects do not perfectly offset for used vehicles, and the resulting each other to produce a constant fleet change in the quantity current owners 7. Accounting for the Rebound Effect size across regulatory alternatives it is will supply; these effects are likely not Caused by Higher Fuel Economy important to remember that the decision exactly offsetting. (a) What is the rebound effect and how to buy a new vehicle and the decision Our models indicate that the ratio of to scrap a used vehicle are often not the magnitude of the scrappage effect to is it measured? made by the same household as a joint the sales effect is greater than one so Amending and establishing fuel decision. The average length of initial that the fleet grows under more economy and GHG standards at a lesser ownership for new vehicles is stringent scenarios. However, it is stringency than the augural standards approximately 6.5 years (and increasing important to remember that not all for future model years will lead to over time). Cumulative scrappage up to vehicles are driven equally; used comparatively lower fuel economy for age seven is typically less than 10%of vehicles are estimated to deliver new cars and light trucks, thus the initial fleet. This suggests that most considerably less annual travel than increasing the amount of fuel they vehicles belong to more than one new vehicles. Further, used vehicles consume in traveling each mile than household over the course of their only have a portion of their original life they would under the augural standard. lifetimes. The household that is left so that it will take more than one deciding whether or not to purchase a used vehicle to replace the full lifetime The resulting increase in their per-mile new vehicle is rarely the same of a new vehicle, at least in the long- fuel and total driving costs will lead to household deciding whether or not to run. The result of the lower annual VMT a reduction in the number of miles they scrap a vehicle. So a vehicle not and shorter remaining lifetimes of used are driven each year over their lifetimes, scrapped in a given year is seldom the vehicles, is that although the fleet is and example of the rebound effect that direct substitute for a new vehicle 1.5% bigger in CY 2050 for the augural is usually associated with energy purchased by that household. baseline than it is for the proposed efficiency improvements working in Considering this, it is not expected that standards, the total non-rebound VMT reverse. The fuel economy rebound for every additional vehicle scrapped, for CY 2050 is 0.4% larger in the effect—a specific example of the energy there is also an additional new one sold, augural baseline than in the proposed efficiency rebound effect for the case of under the proposed standards relative to standards. This small increase in VMT motor vehicles—refers to the well- the baseline standards. is consistent with a larger fleet size; if documented tendency of vehicles’ use Further, while sales and scrappage more used vehicles are supplied, there to increase when their fuel economy is decisions are both influenced by likely is some small resulting increase improved and the cost of driving each changes in new vehicle prices, the in VMT. mile declines as a result. mechanism through which these Our models face some limitations, decisions change are different for the and work will continue toward (b) What does the literature say about two effects. A decrease in average new developing methods for estimating the magnitude of this effect? vehicle prices will directly increase the vehicle sales, scrappage, and mileage Table–II–43 summarizes estimates of demand for new vehicles along the same accumulation. For example, our the fuel economy rebound effect for demand curve. This decrease in new scrappage model assumes that the light-duty vehicles from studies vehicle prices will cause a substitution average VMT for a vehicle of a conducted through 2008, when the towards new vehicles and away from particular vintage is fixed—that is, aside agencies originally surveyed research on used vehicles, shifting the entire from rebound effects, vehicles of a this subject.288 After summarizing all of demand curve for used vehicles particular vintage drive the same downwards. This will decrease both the amount annually, regardless of changes the estimates reported in published and equilibrium prices of used vehicles, as to the average expected lifetimes. The other publicly-available research shown in Figure 8–16 of the PRIA. Since agencies seek comment on ways to available at that time, it distinguishes the decision to scrap a vehicle in a given further integrate the survival and among estimates based on the type of year is closely related to the difference mileage accumulation schedules. Also, data used to develop them. As the table between the vehicle’s value and the cost our analysis uses sales and scrappage reports, estimates of the rebound effect to maintain it, if the value of a vehicle models that do not dynamically interact ranged from 6% to as high as 75%, and is lower than the cost to maintain it, the (though they are based on similar sets of the range spanned by published current owner will not choose to underlying factors); while both models estimates was nearly as wide (7–75%). maintain the vehicle for their own use are informed by new vehicle prices, the 288 or for resale in the used car market, and model of vehicle sales does not respond Complete references to the studies summarized in Table 8–2 are included in the PRIA, the vehicle will be scrapped. That is, a to the size and age profile of the on-road and many of the unpublished studies are available current owner will only supply a fleet, and the model of vehicle in the docket for this rulemaking.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00115 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43100 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Most studies reported more than one the single estimate in which they were estimates spanned only a slightly empirical estimate, and the authors of most confident; these preferred narrower range (9–75%). published studies typically identified

Despite their wide range, these Of these studies, a then recently- Table–II–44 summarizes estimates of estimates displayed a strong central published analysis by Small & Van the rebound effect reported in research tendency, as Table–II–43 also shows. Dender (2007), which reported that the that has become available since the The average values of all estimates, rebound effect appeared to be declining agencies’ original survey, which those that were published, and authors’ over time in response to increasing extended through 2008, and the preferred estimates from published income of drivers, was singled out. following discussion briefly summarizes studies were 22–23%, and the median These authors theorized that rising the approaches used by these more estimates in each category were close to income increased the opportunity cost recent studies. Bento et al. (2009) these values, indicating nearly of drivers’ time, leading them to be less combined demographic characteristics symmetric distributions. The estimates responsive over time to reductions in of more than 20,000 U.S. households, in each category also clustered fairly the fuel cost of driving each mile. Small the manufacturer and model of each tightly around their respective average and Van Dender reported that while the vehicle they owned, and their annual values, as shown by their standard rebound effect averaged 22% over the usage of each vehicle from the 2001 deviations in the table’s last column. entire time period they analyzed (1967– National Household Travel Survey with When classified by the type of data they 2001), its value had declined by half— detailed data on fuel economy and other relied on, U.S. aggregate time-series data or to 11%—during the last five years attributes for each vehicle model produced slightly smaller values they studied (1997–2001). Relying obtained from commercial publications. (averaging 18%) than did panel-type primarily on forecasts of its continued The authors aggregated vehicle models data for individual states (23%) or decline over time, the analysis reduced into 350 categories representing household survey data (25%). In each the 20% rebound effect that NHTSA combinations of manufacturer, vehicle category, the median estimate was again used to analyze the effects of CAFE type, and age, and use the resulting data quite close to the average reported standards for light trucks produced to estimate the parameters of a complex value, and comparing the standard during model years 2005–07 and 2008– model of households’ joint choices of deviations of estimates based on each 11 to 10% for their analysis of CAFE the number and types of vehicles to type of data again suggests a fairly tight and GHG standards for MY 2012–16 own, and their annual use of each scatter around their respective means. passenger cars and light trucks. vehicle.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00116 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.065 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43101

Bento et al. estimate the effect of and aggregate economic activity year when changes in economic activity vehicles’ operating costs per mile, (measured by gross product) using and fuel prices initially occurred, with including fuel costs, which depend in panel-type data for the 10 Canadian three to five years typically required for part on each vehicle’s fuel economy, as provinces over the period from 1990 behavioral adjustments to stabilize. well as maintenance and insurance through 2004. The authors estimated a After controlling for the joint expenses, on households’ annual use of system of equations for these three relationship among vehicle ownership, each vehicle they own. Combining the variables using statistical procedures driving demand, and the fuel efficiency authors’ estimates of the elasticity of appropriate for models where the of cars and light trucks, Barla et al. vehicle use with respect to per-mile variables of interest are simultaneously estimated elasticities of average vehicle operating costs with the reported determined (that is, where each variable use with respect to fuel efficiency that fraction of total operating costs is one of the factors explaining variation corresponded to a rebound effect of accounted for by fuel (slightly less than in the others). This procedure enabled eight percent in the short run, rising to one-half) yields estimates of the them to control for the potential nearly 20% within five years. A notable rebound effect. The resulting values ‘‘reverse influence’’ of households’ feature of their analysis was that vary by household composition, vehicle demand for vehicle travel on their variation in average fuel efficiency size and type, and vehicle age, ranging choices of how many vehicles to own among the individual Canadian from 21 to 38%, with a composite and their fuel efficiency levels when provinces and over the time period they estimate of 34% for all households, estimating the effect of variation in fuel studied was adequate to identify its vehicle models, and ages. The smallest efficiency on vehicle use. effect on vehicle use, without the need values apply to new luxury cars, while Their analysis found that provincial- to combine it with variation in fuel the largest estimates are for light trucks level aggregate economic activity had prices in order to identify its effect. and households with children, but the moderately strong effects on car and Wadud et al. (2009) combine data on implied rebound effects differ little by light truck ownership and use but that U.S. households’ demographic vehicle age. fuel prices had only modest effects on characteristics and expenditures on Barla et al. (2009) analyzed the driving and the average fuel efficiency gasoline over the period 1984–2003 responses of car and light truck of the light-duty vehicle fleet. Each of from the Consumer Expenditure Survey ownership, vehicle travel, and average these effects became considerably with data on gasoline prices and an fuel efficiency to variation in fuel prices stronger over the long term than in the estimate of the average fuel economy of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00117 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.066 43102 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

vehicles owned by individual vehicle use than did differences in annual use. Using variation in the fuel households (constructed from a variety vehicles’ fuel economy. West and economy and per-mile cost of and of sources). They employ these data to Pickrell also found that while the detailed controls for the demographic, explore variation in the sensitivity of rebound effect for households’ use of economic, and locational characteristics individual households’ gasoline passenger cars appeared to be quite of the households that owned them (e.g., consumption to differences in income, large—ranging from 17% to nearly twice road and population density) and each gasoline prices, the number of vehicles that value—it was difficult to detect a vehicle’s main driver (as identified by owned by each household, and their consistent rebound effect for SUVs. survey respondents), the author average fuel economy. Using an Anjovic & Haas (2012) examined employs specialized regression methods estimation procedure intended to variation in vehicle use and fuel to capture the variation in the rebound account for correlation among efficiency among six European nations effect across 10 different categories of unmeasured characteristics of over an extended period (1970–2006), vehicle use. households and among estimation errors using an elaborate model and estimation Su estimated the overall rebound for successive years, the authors explore procedure intended to account for the effect for all vehicles in the sample variation in the response of fuel existence of common underlying trends averaged 13%, and that its magnitude consumption to fuel economy and other among the variables analyzed and thus varied from 11–19% among the 10 variables among households in different avoid identifying spurious or different categories of annual vehicle income categories and between those misleading relationships among them. use. The smallest rebound effects were residing in urban and rural areas. The six nations included in their estimated for vehicles at the two Dividing U.S. households into five analysis were Austria, Germany, extremes of the distribution of annual equally-sized income categories, Wadud Denmark, France, Italy, and Sweden; the use—those driven comparatively little, et al. estimate rebound effects ranging authors also conducted similar analyses and those used most intensively—while from 1–25%, with the smallest estimates for the six nations combined. The the largest estimated effects applied to (8% and 1%) for the two lowest income authors focused on the effects of average vehicles that were driven slightly more categories, and significantly larger income levels, fuel prices, and the fuel than average. Controlling for the estimates for the middle (18%) and two efficiency of each nation’s fleet of cars possibility that high-mileage drivers highest income groups (18 and 25%). In on the total distance they were driven respond to the increased importance of a separate analysis, the authors estimate each year and their total fuel energy fuel costs by choosing vehicles that offer rebound effects of seven percent for consumption. They also tested whether higher fuel economy narrowed the range households of all income levels residing the responses of energy consumption to of Su’s estimated rebound effects in U.S. urban areas and 21% for rural rising and falling fuel prices appeared to slightly (to 11–17%), but did not alter households. be symmetric in the different nations. the finding that they are smallest for West & Pickrell (2011) analyzed data Anjovic and Haas report a long-run lightly- and heavily-driven vehicles and on more than 100,000 households and aggregate rebound effect of 44% for the largest for those with slightly above 300,000 vehicles from the 2009 six nations their study included, with average use. Nationwide Household Transportation corresponding values for individual Linn (2013) also uses the 2009 NHTS Survey to explore how households nations ranging from a low of 19% (for to develop a linear regression approach owning multiple vehicles chose which Austria) to as high as 56% (Italy). These to estimate the relationship between the of them to use and how much to drive estimates are based on the estimated VMT of vehicles belonging to each each one on the day the household was response of vehicle use to variation in household and a variety of different surveyed. Their study focused on how average fuel cost per kilometer driven in factors: Fuel costs, vehicle the type and fuel economy of each each of the six nations and for their characteristics other than fuel economy vehicle a household owned, as well as combined total. Other information (e.g., horsepower, the overall ‘‘quality’’ its demographic characteristics and reported in their study, however, of the vehicle), and household location, influenced household suggests lower rebound effects; their characteristics (e.g., age, income). Linn members’ decisions about whether and estimates of the response of total fuel reports a fuel economy rebound effect how much to drive each vehicle. They energy consumption to fuel efficiency with respect to VMT of between 20– also investigated whether fuel economy appear to imply an aggregate rebound 40%. and fuel prices exerted similar effect of 24% for the six nations, with One interesting result of the study is influences on vehicle use, and whether values ranging from as low as 0–3% (for that when the fuel efficiency of all households owning more than one Austria and Denmark) to as high as 70% vehicles increases, which would be the vehicle tended to substitute use of one (Sweden), although the latter is very long-run effect of rising fuel efficiency for another—or vary their use of all of uncertain. These results suggest that standards, two factors have opposing them similarly—in response to vehicle use in European nations may be effects on the VMT of a particular fluctuations in fuel prices and somewhat less sensitive to variation in vehicle. First, VMT increases when that differences in their vehicles’ fuel driving costs caused by changes in fuel vehicle’s fuel efficiency increases. But economy. efficiency than to changes in driving the increase in the fuel efficiency of the Their estimates of the fuel economy costs arising from variation in fuel household’s other vehicles causes the rebound effect ranged from as low as prices, but they find no evidence of vehicle’s own VMT to decrease. Because nine percent to as high as 34%, with asymmetric responses of total fuel the effect of a vehicle’s own fuel their lowest estimates typically applying consumption to rising and falling prices. efficiency is larger than the other to single-vehicle households and their Using data on household characteristics vehicles’ fuel efficiency, VMT increases highest values to households owning and vehicle use from the 2009 if the fuel efficiency of all vehicles three or more vehicles. They generally Nationwide Household Transportation increases proportionately. Linn also found that differences in fuel prices Survey (NHTS), Su (2012) analyzes the finds that VMT responds much more faced by households who were surveyed effects of locational and demographic strongly to vehicle fuel economy than to on different dates or who lived in factors on household vehicle use and gasoline prices, which is at variance different regions of the U.S. explained investigates how the magnitude of the with the Hymel et al. and Greene results more of the observed variation in daily rebound effect varies with vehicles’ discussed above.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00118 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43103

Like Su and Linn, Liu et al. (2014) of 90%. Although their analysis finds no vehicles are generally less fuel-efficient. employed the 2009 NHTS to develop an significant variation of the rebound The survey data they rely on includes elaborate model of an individual effect with household income, vehicle both owners’ estimates of their annual household’s choices about how many ownership, or urban versus rural use of each car and the distance it was vehicles to own, what types and ages of location, it concludes that the rebound actually driven on a specific day; vehicles to purchase, and how much effect is substantially larger for because they are not closely correlated, combined driving to do using all of households that drive less (90%) than the authors employ them as alternative them. Their analysis used a complex for those who use their vehicles most measures of vehicle use to estimate the mathematical formulation and statistical intensively (56%). rebound effect, but this restricts their methods to represent and measure the Gillingham (2014) analyzed variation sample to the roughly 8,100 cars for interdependence among households’ in the use of approximately five million which both measures are available. choices of the number, types, and ages new vehicles sold in California from Weber and Farsi’s estimates of the of vehicles to purchase, as well as how 2001 to 2003 during the first several rebound effect are extremely large: 75% intensively to use them. years after their purchase, focusing using estimated annual driving and 81% Liu et al. employed their model to particularly on how their use responded when they measure vehicle use by simulate variation in households’ total to geographic and temporal variation in actual daily driving. Excluding vehicle vehicle use to changes in their income fuel prices. His sample consisted size (weight) and limiting the choices levels, neighborhood characteristics, primarily of personal or household that households are assumed to consider and the per-mile fuel cost of driving vehicles (87%) but also included some simultaneously to just vehicles’ fuel averaged over all vehicles each that were purchased by businesses, efficiency and how much to drive household owns. The complexity of the rental car companies, and government approximately reverses these estimates, relationships among the number of agencies. Using county-level data, he but both are still very large. Using a vehicles owned, their specific types and analyzed the effect of differences in the simpler procedure that does not account ages, fuel economy levels, and use monthly average fuel price paid by their incorporated in their model required drivers on variation in their monthly for the potential effect of driving them to measure these effects by use and explored how that effect varied demand on households’ choices among introducing variation in income, with drivers’ demographic vehicle models of different size and fuel neighborhood attributes, and fuel costs, characteristics and household incomes. efficiency produces much smaller and observing the response of Gillingham’s analysis did not include values for the rebound effect: 37% using households’ annual driving. Their a measure of vehicles’ fuel economy or annual driving and 19% using daily results imply a rebound effect of fuel cost per mile driven, so he could travel. The authors interpret these latter approximately 40% in response to not measure the rebound effect directly, estimates as likely to be too low because significant (25–50%) variation in fuel but his estimates of the effect of fuel actual on-road fuel efficiency has not costs, with almost exactly symmetrical prices on vehicle use correspond to a improved as rapidly as suggested by the responses to increases and declines. rebound effect of 22–23% (depending manufacturer-reported measure they A study of the rebound effect by on whether he controlled for the employ. This introduces an error in Frondel et al. (2012) used data from potential effect of gasoline demand on their measure that may be related to a travel diaries recorded by more than its retail price). His estimation vehicle’s age, and their more complex 2,000 German households from 1997 procedure and results imply that vehicle estimation procedure may reduce its through 2009 to estimate alternative use requires nearly two years to adjust effect on their estimates. Nevertheless, measures of the rebound effect, and to fully to changes in fuel prices. He found even their lower estimates exceed those explore variation in their magnitude little variation in the sensitivity of from many other studies of the rebound among households. Each household vehicle use to fuel prices among car effect, as Table 8–2 shows. participating in the survey recorded its buyers with different demographic Hymel, Small, & Van Dender (2010)— automobile travel and fuel purchases characteristics, although his results and more recently, Hymel & Small over a period of one to three years and suggested that it increases with their (2015)—extended the simultaneous supplied information on its composition income levels. equations analysis of time-series and Weber & Farsi (2014) analyzed and the personal characteristics of each state-level variation in vehicle use variation in the use of more than 70,000 of its members. The authors converted originally reported in Small & Van individual cars owned by Swiss households’ travel and fuel Dender (2007) and to test the effect of consumption to a monthly basis, and households who were included in a including more recent data. As in the used specialized estimation procedures 2010 survey of travel behavior. Their original 2007 study, both subsequent (quantile and random-effects panel analysis focuses on the simultaneous extensions found that the fuel economy regression) to analyze monthly variation relationships among households’ rebound effect had declined over time in their travel and fuel use in relation choices of the fuel efficiency and size in response to increasing personal to differences in fuel prices, the fuel (weight) of the vehicles they own, and efficiency of each vehicle a household how much they drive each one, income and urbanization but had risen owned, and the fuel cost per mile of although they recognize that fuel during periods when fuel prices driving each vehicle. efficiency cannot be chosen increased. Because they rely on the Frondel et al. estimate four separate independently of vehicle weight. The response of vehicle use to fuel cost per measures of the rebound effect, three of authors employ a model specification mile to estimate the rebound effect, which capture the response of vehicle and statistical estimation procedures however, none of these three studies is use to variation in fuel efficiency, fuel that account for the likelihood that able to detect whether its apparent price, and fuel cost per mile traveled, households intending to drive more will decline in response to rising income and a fourth capturing the response of purchase more fuel-efficient cars but levels over time truly reflects its effect fuel consumption to changes in fuel may also choose more spacious and on drivers’ responses to changing fuel price. Their first three estimates range comfortable—and thus heavier— economy—the rebound effect itself—or from 42% to 57%, while their fourth models, which affects their fuel simply captures the effect of rising estimate corresponds to a rebound effect efficiency indirectly, since heavier income on their sensitivity to fuel

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00119 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43104 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

prices.289 These updated studies each above—made them barely ineligible for change in households’ vehicle use in revised Small and Van Dender’s original the program. response to differences in the fuel estimate of an 11% rebound effect for The authors reported that the higher economy between vehicles they had 1997–2011 upward when they included fuel economy of new models that owned previously and the new models more recent experience: To 13% for the eligible households purchased in they purchased to replace them, over period 2001–04, and subsequently to response to the generous financial and above any change in vehicle use 18% for 2000–2009. incentives offered under the ‘‘Cash for among households who did not buy In their 2015 update, Hymel and Clunkers’’ did not prompt their buyers new cars (and thus saw no change in Small hypothesized that the recent to use them more than the older, low- fuel economy). increase in the rebound effect could be MPG vehicles they replaced. They These authors’ data enabled them to traced to a combination of expanded attributed this apparent absence of a control for the effects of changes over media coverage of changing fuel prices, fuel economy rebound effect—which time in household characteristics and increased price volatility, and an they described as an ‘‘attribute- vehicle features other than fuel economy that were likely to have asymmetric response by drivers to adjusted’’ measure of its magnitude—to contributed to observed changes in variation in fuel costs. The authors the fact that eligible households chose vehicle use. They also employed estimated that about half of the apparent to buy less expensive, smaller, and complex statistical methods to account increase in the rebound effect for recent lower-performing models to replace for the fact that some households years could be attributed to greater those they retired. Because these replacing their vehicles may have done volatility in fuel prices and more media replacements offered lower-quality so in anticipation of changes in their coverage of sudden price changes. Their transportation service, their buyers did driving demands (rather than the results also suggest that households not drive them more than the vehicles reverse), as well as for the possibility curtail their vehicle use within the first they replaced. The applicability of this result to the that some households who replaced year following an increase in fuel prices proposal’s analysis is doubtful because their cars may have done so because and driving costs, while the increase in previous regulatory analyses assumed their driving behavior was more driving that occurs in response to that manufacturers could achieve sensitive to fuel prices than other declining fuel prices—and by required improvements in fuel economy households. Their estimates ranged implication, to improvements in fuel without compromising the performance, from 8–10%, varying only minimally economy—occurs more slowly. carrying and towing capacity, comfort, among alternative model specifications West et al. (2015) attempted to infer or safety of cars and light trucks from and statistical estimation procedures or the fuel economy rebound effect using recent model years.290 While this may in response to whether their sample was data from Texas households who be technically true, doing so would restricted to households that replaced replaced their vehicles with more fuel- come at a combined greater cost. If this their vehicles or also included efficient models under the 2009 ‘‘Cash argument is correct, then amending households that kept their original for Clunkers’’ program, which offered future standards at a reduced stringency vehicles throughout the period.291 sizeable financial incentives to do so. from their previously-adopted levels Finally, De Borger et al. found no Under the program, households that would lead to less driving attributable to evidence that the rebound effect is retired older vehicles with fuel economy rebound, and should therefore not lead smaller among lower-income levels of 18 miles per gallon (MPG) or to artificial constraints in new vehicles’ households than among their higher- less were eligible for cash incentives other features that offset the reduction income counterparts. ranging from $3,500–4,000, while those in their use stemming from lower fuel (c) What value have the agencies retiring vehicles with higher fuel economy. economy were ineligible for such Most recently, De Borger et al. (2017) assumed in this rule? rebates. The authors examined the fuel analyze the response of vehicle use to On the basis of all of the evidence economy, other features, and changes in fuel economy among a summarized here, a fuel economy subsequent use of new vehicles sample of nearly 350,000 Danish rebound effect of 20% has been chosen households in Texas purchased to households owning the same model to analyze the effects of the proposed replace older models that narrowly vehicle, of which almost one-third action. This is a departure from the 10% qualified for the program’s financial replaced it with a different model value used in regulatory analyses for incentives because their fuel economy sometime during the period from 2001 MYs 2012–2016 and previous analyses was only slightly below the 18 MPG to 2011. By comparing the changes in for MYs 2017–2025 CAFE and GHG threshold. They then compared these to households’ driving from the early years standards and represents a return to the the fuel economy, features, and use of of this period to its later years among value employed in the analyses for MYs new vehicles that demographically those who replaced their vehicles 2005–2011 CAFE standards. There are comparable households bought to during the intervening period to the several reasons the estimate of the fuel replace older models, but whose slightly changes in driving among households economy rebound effect for this analysis higher fuel economy—19 MPG or who kept their original vehicles, the has been increased. authors attempted to isolate the effect of First, the 10% value is inconsistent 289 DeBorger et al. (2016) analyze the separate changes in fuel economy on vehicle use with nearly all research on the effects of variation in household income on the from those of other factors. They magnitude of the rebound effect, as sensitivity of their vehicle use to fuel prices and the Table–II–43 and Table–II–44 indicate. fuel economy of vehicles they own. Their results measured the rebound effect as the imply the decline in the fuel economy rebound Instead, it is based almost exclusively effect with income reported in Small & Van Dender 290 As discussed, this does not mean attributes of (2007) and its subsequent extensions appears to future cars and light trucks will be anything close 291 This latter result suggests their estimates were result entirely from a reduction in drivers’ to those manufacturers could have offered if lower not biased by any tendency for households whose sensitivity to fuel prices as their incomes rise, standards had remained in effect. Instead, the demographic characteristics, economic rather than from any effect of rising income on the agencies asserted features other than fuel economy circumstances, or driving demands changed over sensitivity of vehicle use to improving fuel could be maintained at the levels offered in recent the period in ways that prompted them to replace economy; i.e., on the fuel economy rebound effect model years—that features will not likely be their vehicles with models offering different fuel itself. removed, but may not be improved. economy.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00120 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43105

on the finding of the 2007 study by In addition, buyers of new cars and light benefits to vehicle owners, which reflect Small and Van Dender that the rebound trucks belong disproportionately to the value to drivers and other vehicle effect had been declining over time in higher-income households that already occupants of the added (or more response to drivers’ rising incomes and own multiple vehicles, which further desirable) social and economic on extending that decline through future suggests that the higher values of the opportunities that become accessible years using an assumption of steady rebound effect estimated by many with additional travel. As evidenced by income growth. As indicated above, studies for such households are more the fact that they elect to make more however, subsequent extensions of relevant for analyzing use of newly- frequent or longer trips when the cost of Small and Van Dender’s original purchased cars and light trucks. driving declines, the benefits from this research have produced larger estimates Finally, research on the rebound added travel exceed drivers’ added of the rebound effect for recent years: effect conducted since the agencies’ outlays for the fuel it consumes While their original study estimated the original 2008 review of evidence almost (measured at the improved level of fuel rebound effect at 11% for 1997–2001, universally reports estimates in the 10– economy resulting from stricter CAFE the 2010 update by Hymel, Small, and 40% (and larger) range, as Table–II–43 standards). The amount by which the Van Dender reported a value of 13% for shows. Thus, the 20% rebound effect benefits from this increased driving 2004, and Hymel and Small’s 2015 used in this analysis more accurately travel exceed its increased fuel costs update estimated the rebound effect at represents the findings from both the measures the net benefits they receive 18% for 2003–09. Further, the issues studies considered in 2008 review and from the additional travel, usually are with state-level measures of vehicle use, the more recent analyses. referred to as increased consumer fuel consumption, and fuel economy (1) What are the implications of the surplus. identified previously raise some doubt rebound effect for VMT? NHTSA’s analysis estimates the about the reliability of these studies’ economic value of the decreased estimates of the rebound effect. The assumed rebound effect not only consumer surplus provided by reduced At the same time, the continued influences the use of new vehicles in driving using the conventional increases in income that were today’s analysis but also affects the approximation, which is one half of the anticipated to produce a continued response of the initial registered vehicle product of the increase in vehicle population to changes in fuel price decline in the rebound effect have not operating costs per vehicle-mile and the throughout their remaining useful lives. materialized. The income measure (real resulting decrease in the annual number The fuel prices used in this analysis are personal income per Capita) used in of miles driven. Because it depends on lower than the projections used to these analyses has grown only the extent of the change in fuel inform the 2012 Final Rule but generally approximately one percent annually economy, the value of economic increase from today’s level over time. As over the past two decades and is impacts from decreased vehicle use they do so, the rebound effect acts as a projected to grow at approximately changes by model year and varies price elasticity of demand for travel—as 1.5% for the next 30 years, in contrast among alternative CAFE standards. to the two to three percent annual the cost-per-mile of travel increases, growth assumed by the agencies when owners of all vehicles in the registered (d) Societal Externalities Associated developing earlier forecasts of the future population respond by driving less. In With CAFE Alternatives particular, they drive 20% less than the rebound effect. Further, another recent (1) Energy Security Externalities study by DeBorger et al. (2016) analyzed difference between the cost-per-mile of the separate effects of variation in travel when they were observed in Higher U.S. fuel consumption will household income on the sensitivity of calendar year 2016 and the relevant produce a corresponding increase in the their vehicle use to fuel prices and the cost-per-mile at any future age. For the nation’s demand for crude petroleum, fuel economy of vehicles they own. new vehicles subject to this proposal which is traded actively in a worldwide These authors’ results indicate that the (and explicitly simulated by the CAFE market. The U.S. accounts for a large decline in the fuel economy rebound model), fuel economies increase relative enough share of global oil consumption effect with income reported in Small & to MY 2016 levels, and generally that the resulting boost in global Van Dender (2007) and subsequent improve enough to offset the effect of demand will raise its worldwide price. research results entirely from a rising fuel prices—at least during the The increase in global petroleum prices reduction in drivers’ sensitivity to fuel years covered by the proposal. For those that results from higher U.S. demand prices as their incomes rise rather than vehicles, the difference between the causes a transfer of revenue to oil from any effect of rising income on the initial cost-per-mile of travel and future producers worldwide from not only sensitivity of vehicle use to fuel travel costs is negative. As the vehicles buyers of new cars and light trucks, but economy itself. This latter measure, become less expensive to operate, they also other consumers of petroleum which DeBorger et al. find has not are driven more (20% more than the products in the U.S. and throughout the changed significantly as incomes have difference between initial and present world, all of whom pay the higher price risen over time, is the correct measure travel costs, precisely). Of course, each that results. of the fuel economy rebound effect, so of the regulatory alternatives considered Although these effects will be their analysis calls into question its in the analysis would result in lower tempered by growing U.S. oil assumed sensitivity to income. fuel economy levels for vehicles production, uncertainty in the long-term Some studies of households’ use of produced in model year 2020 and later import-export balance makes it difficult individual vehicles also find that the than if the baseline standards remained to precisely project how these effects fuel economy rebound effect increases in effect, so total VMT is lower under might change in response to that with the number of vehicles they own. these alternatives than under the increased production. Growing U.S. Because vehicle ownership is strongly baseline. petroleum consumption will also associated with household income, this increase potential costs to all U.S. common finding suggests that the (2) What is the mobility benefit that petroleum users from possible overall value of the rebound effect is accrues to vehicle owners? interruptions in the global supply of unlikely to decline with rising incomes The increase in travel associated with petroleum or rapid increases in global as the agencies had previously assumed. the rebound effect produces additional oil prices, not all of which are borne by

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00121 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43106 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

the households or businesses who The social damage costs associated the baseline, which generates savings in increase their petroleum consumption with changes in the emissions of criteria congestion and road noise relative to the (that is, they are partly ‘‘external’’ to pollutants and CO2 was calculated, baseline. petroleum users). If U.S. demand for attributing benefits and costs to the F. Impact of CAFE Standards on Vehicle regulatory alternatives considered based imported petroleum increases, it is also Safety possible that increased military on the sign of the change in each spending to secure larger oil supplies pollutant. In previous rulemakings, the In past CAFE rulemakings, NHTSA from unstable regions of the globe will agencies have considered the social cost has examined the effect of CAFE be necessary. of CO2 emissions from a global standards on vehicle mass and the These three effects are often referred perspective, accumulating social costs subsequent effect mass changes will have on vehicle safety. While setting to collectively as ‘‘energy security for CO2 emissions based on adverse externalities’’ resulting from U.S. outcomes attributable to climate change standards based on vehicle footprint petroleum consumption, and increases in any country. In this analysis, helps reduce potential safety impacts associated with CAFE standards as in their magnitude are sometimes cited however, the costs of CO2 emissions and compared to setting standards based on as potential social costs of increased resulting climate damages from both some other vehicle attribute, footprint- U.S. demand for oil. To the extent that domestic and global perspectives were based standards cannot entirely they represent real economic costs that considered. Chapter 9 of the Regulatory eliminate those impacts. Although prior would rise incrementally with increases Impact Analysis provides a detailed analyses noted that there could also be in U.S. petroleum consumption of the discussion of how the agencies estimate impacts because of other factors besides magnitude likely to result from less changes in emissions of criteria air mass changes, those impacts were not stringent CAFE and GHG standards, pollutants and CO2 and reports the estimated quantitatively.292 In this these effects represent potential values the agencies use to estimate current analysis, the safety analysis has additional costs of this proposed action. benefits or costs associated with those changes in emissions. been expanded to include a broader and Chapter 7 of the Regulatory Impact more comprehensive measure of safety Analysis for this proposed action (3) Traffic Externalities (Congestion, impacts, as discussed below. A number defines each of these energy security Noise) of factors can influence motor vehicle externalities in detail, assesses whether Increased vehicle use associated with fatalities directly by influencing vehicle its magnitude is likely to change as a the rebound effect also contributes to design or indirectly by influencing consequence of this action, and increased traffic congestion and consumer behavior. These factors identifies whether that change highway noise. To estimate the include: represents a real economic cost or economic costs associated with these (1) Changes, which affect the benefit of this action. consequences of added driving, the crashworthiness of vehicles impact (2) Environmental Externalities estimates of per-mile congestion and other vehicles or roadside objects, in noise costs caused by increased use of vehicle mass made to reduce fuel The change in criteria pollutant automobiles and light trucks developed consumption. NHTSA’s statistical emissions that result from changes in previously by the Federal Highway analysis of historical crash data to vehicle usage and fuel consumption is Administration (FHWA) were applied. understand effects of vehicle mass and estimated as part of this analysis. These values are intended to measure size on safety indicates reducing mass Criteria air pollutants include carbon the increased costs resulting from added in light trucks generally improves monoxide (CO), hydrocarbon congestion and the delays it causes to safety, while reducing mass in compounds (usually referred to as other drivers and passengers and noise passenger cars generally reduces safety. ‘‘volatile organic compounds,’’ or VOC), levels contributed by automobiles and NHTSA’s crash simulation modeling of nitrogen oxides (NOX), fine particulate light trucks. NHTSA previously vehicle design concepts for reducing matter (PM2.5), and sulfur oxides (SOX). employed these estimates in its analysis mass revealed similar trends.293 These pollutants are emitted during accompanying the MY 2011 final CAFE (2) The delay in the pace of consumer vehicle storage and use, as well as rule as well as in its analysis of the acquisition of newer safer vehicles that throughout the fuel production and effects of higher CAFE standards for MY results from higher vehicle prices distribution system. While increases in 2012–16 and MY 2017–2021. After associated with technologies needed to domestic fuel refining, storage, and reviewing the procedures used by meet higher CAFE standards. Because of distribution that result from higher fuel FHWA to develop them and considering a combination of safety regulations and consumption will increase emissions of other available estimates of these values voluntary safety improvements, these pollutants, reduced vehicle use and recognizing that no commenters passenger vehicles have become safer associated with the fuel economy have addressed these costs directly in over time. Compared to prior decades, rebound effect will decrease their their comments on previous rules, the fatality rates have declined significantly emissions. The net effect of less values continue to be appropriate for stringent CAFE standards on total use in this proposal. For this analysis, 292 NHTSA included a quantification of rebound- emissions of each criteria pollutant FHWA’s estimates of per-mile costs are associated safety impacts in its Draft TAR analysis, but because the scrappage model is new for this depends on the relative magnitudes of multiplied by the annual increases in rulemaking, did not include safety impacts increases in its emissions during fuel automobile and light truck use from the associated with the effect of standards on new refining and distribution, and decreases rebound effect to yield the estimated vehicle prices and thus on fleet turnover. The fact in its emissions resulting from increases in total congestion and noise that the scrappage model did not exist previously does not mean that the effects that it aims to show additional vehicle use. Because the externality costs during each year over were not important considerations, simply that the relationship between emissions in fuel the lifetimes of the cars and light trucks agency was unable to account for them refining and vehicle use is different for in the on-road fleet. Due to the fact that quantitatively prior to the current analysis. each criteria pollutant, the net effect of this proposal represents a decrease in 293 DOT HS 812051a—Methodology for evaluating fleet protection of new vehicle designs increased fuel consumption from the stringency, the fuel economy rebound Application to lightweight vehicle designs, DOT HS proposed standards on total emissions effect results in fewer miles driven 812051b Methodology for evaluating fleet of each pollutant is likely to differ. under the action alternatives relative to protection of new vehicle designs_Appendices.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00122 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43107

because of technological safety higher vehicle prices are directly because older vehicle vintages are improvements as well as behavioral attributed to CAFE standards.294 This is associated with higher rates of shifts such as increased seat belt use. reflected monetarily by valuing extra involvement in fatal crashes than newer The results of this analysis project that rebound miles at the full value of their vehicles. Finally, a dynamic fleet share vehicle prices will be nearly $1,900 added driving cost plus the added safety model also predicts the effects of higher under the augural CAFE risk consumers experience, which changes in the standards on the share of standards compared to the preferred completely offsets the societal impact of light trucks and passenger cars in future alternative that would hold stringency any added fatalities from this voluntary model year light-duty vehicle fleets. at MY 2020 levels in MYs 2021–2026. consumer choice. Vehicles of different body styles have This will induce some consumers to The safety component of CAFE different rates of involvement in fatal delay or forgo the purchase of newer analysis has evolved over time. In the crashes, so that changing the share of safer vehicles and slow the transition of 2012 final rule, the analysis accounted each in the projected future fleet has the on-road fleet to one with the for the change in projected fatalities safety impacts; the implied safety effects improved safety available in newer attributable to mass reduction of new are captured in the current modelling. vehicles. This same factor can also shift vehicles. The model assumed that The agencies seek comment on changes the mix of passenger cars and light manufacturers would choose mass to the safety analysis made in this trucks. reduction as a compliance method proposal, they seek particular comment (3) Increased driving because of better across vehicle classes such that the net on the following changes: fuel economy. The ‘‘rebound effect’’ effect of mass reduction on fatalities was (1) The sales scrappage models as predicts consumers will drive more zero. However, in the 2016 draft independent models: Two separate models when the cost of driving declines. More Technical Assessment Report, DOT capture the effects of new vehicle prices on stringent CAFE standards reduce made two consequential changes to the new vehicle demand and used vehicle vehicle operating costs, and in response, analysis of fatalities associated with the retirement rates—the sales model and the some consumers may choose to drive CAFE standards. In particular, first, the scrappage model, respectively. We seek more. Driving more increases exposure modelling assumed that mass reduction public comment on the methods used for to risks associated with on-road each of these models, in particular we seek technology was available to all vehicles, comment on: transportation, and this added exposure regardless of net safety impact, and translates into higher fatalities. • The assumptions and variables included in second, it accounted for the incremental the independent models Although all three factors influence safety costs associated with additional • predicted fatality levels that may occur, The techniques and data used to estimate miles traveled due to the rebound effect. the independent models only two of them, the changes in vehicle The current analysis extends the • The structure and implementation of the mass and the changes in the acquisition analysis to report incremental fatality independent models of safer vehicles—are actually imposed impacts associated with additional (2) Integration of the sales and scrappage on consumers by CAFE standards. The miles traveled due to the rebound effect, models: The new sales and scrappage models safety of vehicles has improved over use many of the same predictors, but are not and identifies the increase in fatalities time and is expected to continue directly integrated. We seek public comment associated with additional driving improving in the future commensurate on, and data supporting whether integrating separately from changes in fatalities with the pace of safety technology the two models is appropriate. attributable other sources.295 (3) Integration of the scrappage rates and innovation and implementation and The current analysis adds another mileage accumulation: The current model motor vehicle safety regulation. Safety element: The effect that higher new assumes that annual mileage accumulation improvements will likely continue vehicle prices have on new vehicle sales and scrappage rates are independent of one regardless of changes to CAFE another. We seek public comment on the and on used vehicle scrappage, which standards. However, its pace may be appropriateness of this assumption, and data modified if manufacturers choose to influences total expected fatalities that would support developing an interaction delay or forgo investments in safety between scrappage rates and mileage 294 It could be argued fatalities resulting from accumulation, or testing whether such an technology because of the demand consumer’s decision to delay the purchase of newer CAFE standards impose on research, interaction is important to include. safer vehicles is also a market decision implying (4) Increased risk of older vehicles: The consumers fully accept the added safety risk development, and manufacturing observed increase in crash and injury risk budgets. Increased driving associated associated with this delay and value the time value of money saved by the delayed purchase more than associated with older vehicles is likely due with rebound is a consumer choice. this risk. This scenario is likely accurate for some to a combination of vehicle factors and driver Improved CAFE will reduce driving purchasers. For others, the added cost may factors. For example, older vehicles are less costs, but nothing in the higher CAFE represent a threshold price increase effectively crashworthy because in general they’re standards compels consumers to drive preventing them from being financially able to equipped with fewer or less modern safety purchase a new vehicle. Presently there is no way features, and drivers of older cars are on additional miles. If consumers choose to to determine the proportion of lost sales reflected do so, they are making a decision that average younger and may be less skilled by these two scenarios. The added driving from the drivers or less risk-averse than drivers of new the utility of more driving exceeds the rebound effect results from a positive benefit of CAFE, which reduces the cost of driving. By vehicles. We fit a model which includes both marginal operating costs as well as the an age and vintage affect, but assume that the added crash risk it entails. Thus, while contrast, the effect of retaining older vehicles longer results from costs imposed on consumers, which age effect is entirely a result of changes in the predicted fatality impacts with all potentially limit their purchase options. Thus, average driver demographics, and not three factors embedded into the model fatalities are attributed to retaining older vehicles impacted by changes in CAFE or GHG are measured, the fatalities associated due to CAFE but not those resulting from decisions standards. We seek comment on this with consumer choice decisions are to drive more. Comments are sought on this approach for attributing increased older assumption. vehicle risk. Is the analysis likely to accounted for separately from those 295 Drivers who travel additional miles are resulting from technologies overestimate or underestimate the safety assumed to experience benefits that at least offset benefits under the proposed alternative? the costs they incur in doing so, including the implemented in response to CAFE (5) Changes in the mix of light trucks and regulations or economic limitations increased safety risks they face. Thus while the number of additional fatalities resulting from passenger cars: The dynamic fleet share resulting from CAFE regulation. Only increased driving is reported, the associated costs model predicts changes in the future share of those safety impacts associated with are not included among the social costs of the light truck and passenger car vehicles. mass reduction and those resulting from proposal. Changes in the mix of vehicles may result in

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00123 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43108 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

increased or decreased fatalities. Does the response to the more stringent future safety analyses in the NPRM and final dynamic fleet share model reasonably CAFE and GHG standards. rule are based with newer vehicle data capture consumers’ decisions about how they Manufacturers stated they will reduce and create a common database that substitute between different types and sizes vehicle mass as one of the cost-effective could be made publicly available to of vehicles depending on changes in fuel address concerns that differences in economy, relative and absolute prices, and means of increasing fuel economy and other vehicle attributes? We seek comment reducing CO2 to meet standards, and data were leading to different results in on whether our safety analysis provides a this approach is incorporated this statistical analyses by different reasonable estimate of the effects of changes expectation into the modeling analysis researchers. in fleet mix on future fatalities. supporting the standards. Because the And third, to assess if the design of analysis discerns a historical recent model year vehicles 1. Impact of Weight Reduction on Safety relationship between vehicle mass, size, incorporating various mass reduction methods affect relationships among The primary goals of CAFE and CO and safety, it is reasonable to assume 2 vehicle mass, size, and safety, the standards are reducing fuel these relationships will continue in the agencies sought to identify vehicles consumption and CO emissions from future. 2 using material substitution and smart the on-road light-duty vehicle fleet; in (a) Historical Analyses of Vehicle Mass design and to assess if there is sufficient addition to these intended effects, the and Safety crash data involving those vehicles for potential of the standards to affect statistical analysis. If sufficient data vehicle safety is also considered.296 As Researchers have been using statistical analysis to examine the exists, statistical analysis would be a safety agency, NHTSA has long conducted to compare the relationship considered the potential for adverse relationship of vehicle mass and safety in historical crash data for many years among mass, size, and safety of these safety consequences when establishing smart design vehicles to vehicles of CAFE standards, and under the CAA, and continue to refine their techniques. In the MY 2012–2016 final rule, the similar size and mass with more EPA considers factors related to public traditional designs. health and human welfare, including agencies stated we would conduct further study and research into the By the time of the MY 2017–2025 safety, in regulating emissions of air final rule, significant progress was made pollutants from mobile sources. interaction of mass, size, and safety to assist future rulemakings and start to on these tasks: The independent review Safety trade-offs associated with fuel of recent and updated statistical economy increases have occurred in the work collaboratively by developing an interagency working group between analyses of the relationship between past, particularly before NHTSA CAFE vehicle mass, size, and crash fatality standards were attribute-based; past NHTSA, EPA, DOE, and CARB to evaluate all aspects of mass, size, and rates had been completed. NHTSA safety trade-offs may have occurred contracted with the University of because manufacturers chose at the safety. The team would seek to coordinate government-supported Michigan Transportation Research time, in response to CAFE standards, to Institute (UMTRI) to conduct this build smaller and lighter vehicles. studies and independent research to the greatest extent possible to ensure the review, and the UMTRI team led by Although the agency now uses attribute- Green evaluated more than 20 papers, based standards, in part to protect work is complementary to previous and ongoing research and to guide further including studies done by NHTSA’s against excessive vehicle downsizing, Kahane, Wenzel of the U.S. Department the agency must be mindful of the research in this area. The agencies also identified three of Energy’s Lawrence Berkeley National possibility of related safety trade-offs in Laboratory, Dynamic Research, Inc., and the future. In cases where fuel economy specific areas to direct research in preparation for future CAFE/CO others. UMTRI’s basic findings are improvements were achieved through 2 discussed in Chapter 11 of the PRIA rulemaking regarding statistical analysis reductions in vehicle size and mass, the accompanying this NPRM. of historical data. First, NHTSA would smaller, lighter vehicles did not fare as Some commenters in recent CAFE well in crashes as larger, heavier contract with an independent rulemakings, including some vehicle vehicles, on average. institution to review statistical methods manufacturers, suggested designs and Historically, as shown in FARS data NHTSA and DRI used to analyze materials of more recent model year analyzed by NHTSA, the safest cars historical data related to mass, size, and vehicles may have weakened the generally have been heavy and large, safety, and to provide recommendations historical statistical relationships while cars with the highest fatal-crash on whether existing or other methods between mass, size, and safety. It was rates have been light and small. The should be used for future statistical agreed that the statistical analysis would question, then, is whether past is analysis of historical data. This study be improved by using an updated necessarily a prologue when it comes to would include a consideration of database reflecting more recent safety potential changes in vehicle size (both potential near multicollinearity in the technologies, vehicle designs and footprint and ‘‘overhang’’) and mass in historical data and how best to address materials, and reflecting changes in the it in a regression analysis. The 2010 vehicle fleet. An updated database was 296 In this rulemaking document, ‘‘vehicle safety’’ NHTSA report (hereinafter 2010 Kahane created and employed for assessing is defined as societal fatality rates per vehicle mile report) was also peer reviewed by two safety effects for that final rule. The of travel (VMT), including fatalities to occupants of other experts in the safety field—Farmer agencies also believed, as UMTRI found, all vehicles involved in collisions, plus any (Insurance Institute for Highway Safety) pedestrians. Injuries and property damage are not different statistical analyses may have within the scope of the statistical models discussed and Lie (Swedish Transport produced different results because they 297 in this section because of data limitations (e.g., Administration). used slightly different datasets for their limited information on observed or potential Second, NHTSA and EPA, in analyses. relationships between safety standards and injury consultation with DOE, would update and property damage outcomes, consistency of To try to mitigate this issue and to reported injury severity levels). Rather, injuries and the MY 1991–1999 database where support the current rulemaking, NHTSA property damage are represented within the CAFE created a common, updated database for model through adjustment factors based on 297 All three peer reviews are available in Docket observed relationships between societal costs of No. NHTSA–2010–0152, Relationships Between statistical analysis consisting of crash fatalities and societal injury and property damage Fatality Risk, Mass, and Footprint, https:// data of model years 2000–2007 vehicles costs. www.regulations.gov/docket?D=NHTSA-2010-0152. in calendar years 2002–2008, as

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00124 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43109

compared to the database used in prior NHTSA’s baseline results and casualty 2021 passenger cars and light trucks NHTSA analyses, which was based on risk per VMT; included a statistical analysis of model years 1991–1999 vehicles in • the 2012 DRI reports on, among relationships between fatality risk, calendar years 1995–2000. The new other things, the effects of mass mass, and footprint in MY 2000–2007 database was the most up-to-date reduction on crash frequency and passenger cars and LTVs (light trucks possible, given the processing lead time fatality risk per crash; and vans), based on calendar year (CY) for crash data and the need for enough • LBNL’s subsequent review of DRI’s 2002–2008 crash and vehicle- crash cases to permit statistically study; registration data; 302 this analysis was meaningful analyses. NHTSA made the • the 2015 National Academy of also detailed in the 2012 Kahane report. preliminary version of the new Sciences Report; and The principal findings and database, which was the basis for • the 2017 NBER working paper conclusions of the 2012 Kahane report NHTSA’s 2011 preliminary report analyzing the relationships among were mass reduction in the lighter cars, (hereinafter 2011 Kahane report), traffic fatalities, CAFE standards, and even while holding footprint constant, available to the public in May 2011, and distributions of MY 1989–2005 light- would significantly increase fatality an updated version in April 2012 (used duty vehicle curb weights. risk, whereas mass reduction in the in NHTSA’s 2012 final report, A detailed discussion of each analysis heavier LTVs would reduce societal hereinafter 2012 Kahane report),298 is discussed in Chapter 11 of the PRIA fatality risk by reducing the fatality risk enabling other researchers to analyze accompanying this proposed rule. of occupants of lighter vehicles the same data and hopefully minimize colliding with those heavier LTVs. (b) Recent NHTSA Analysis Supporting discrepancies in results because of NHTSA concluded, as a result, any CAFE Rulemaking inconsistencies across databases.299 reasonable combination of mass Since the publication of the MYs As mentioned previously, NHTSA reductions that held footprint constant 2017–2025 final rule, NHTSA has and EPA’s 2012 joint final rule for MYs in MY 2017–2021 vehicles— sponsored, and is sponsoring, new 2017 and beyond set ‘‘footprint-based’’ concentrated, at least to some extent, in studies and research to inform the standards, with footprint being defined the heavier LTVs and limited in the lighter cars—would likely be current CAFE and CO2 rulemaking. In as roughly equal to the wheelbase addition, the National Academy of multiplied by the average of the front approximately safety-neutral; it would Sciences published a new report in this and rear track widths. Basing standards not significantly increase fatalities and area.300 Throughout the rulemaking on vehicle footprint ideally helps to might well decrease them. NHTSA released a preliminary report process, NHTSA’s goal is to publish as discourage vehicle manufacturers from (2016 Puckett and Kindelberger report) much of our research as possible. In downsizing their vehicles; the agencies on the relationship between fatality risk, establishing standards, all available set higher (more stringent) mile per mass, and footprint in June 2016 in data, studies, and information gallon (mpg) targets for smaller-footprint advance of the Draft TAR. The objectively without regard to whether vehicles but would not similarly preliminary report covered the same they were sponsored by the agencies, discourage mass reduction that scope as the 2012 Kahane report, will be considered. maintains footprint while potentially offering a detailed description of the Undertaking these tasks has helped improving fuel economy. Several technologies, such as substitution of databases, modeling approach, and come closer to resolving ongoing analytical results on relationships debates in statistical analysis research of light, high-strength materials for conventional materials during vehicle among vehicle size, mass, and fatalities historical crash data. It is intended that that informed the Draft TAR. Results in these conclusions will be applied going redesigns, have the potential to reduce weight and conserve fuel while the Draft TAR and the 2016 Puckett and forward in future rulemakings, and it is Kindelberger report are consistent with maintaining a vehicle’s footprint and believed the research will assist the results in the 2012 Kahane report; maintaining or possibly improving the public discussion of the issues. Specific chiefly, societal effects of mass vehicle’s structural strength and historical analyses (in addition to reduction are small, and mass reduction handling. NHTSA’s own analysis) on vehicle mass concentrated in larger vehicles is likely In considering what technologies are and safety used to support this to have a beneficial effect on fatalities, available for improving fuel economy, rulemaking include: while mass reduction concentrated in • including mass reduction, an important The 2011 and 2013 NHTSA smaller vehicles is likely to have a corollary issue for NHTSA to consider is Workshops on Vehicle Mass, Size, and detrimental effect on fatalities. Safety; the potential effect those technologies For the 2016 Puckett and • the University of Michigan may have on safety. NHTSA has thus far Kindelberger report and Draft TAR, Transportation Research Institute specifically considered the likely effect NHTSA, working closely with EPA and (UMTRI) independent review of a set of of mass reduction that maintains the DOE, performed an updated statistical relationships between vehicle footprint on fatal crashes. The statistical analysis of relationships curb weight, footprint variables (track relationship between a vehicle’s mass, between fatality rates, mass and width, wheelbase), and fatality rates size, and fatality risk is complex, and it footprint, updating the crash and from vehicle crashes; varies in different types of crashes. As exposure databases to the latest • the 2012 Lawrence Berkeley mentioned above, NHTSA, along with available model years. The agencies National Laboratory (LBNL) Phase 1 and others, has been examining this analyzed updated databases that 301 Phase 2 reports on the sensitivity of relationship for more than a decade. included MY 2003–2010 vehicles in CY The safety chapter of NHTSA’s April 2005–2011 crashes. For this proposed 298 Those databases are available at ftp:// 2012 final regulatory impact analysis ftp.nhtsa.dot.gov/CAFE/. (FRIA) of CAFE standards for MY 2017– 302 Kahane, C.J. Relationships Between Fatality 299 See 75 FR 25324, 25395–25396 (May 7, 2010) Risk, Mass, and Footprint in Model Year 2000–2007 (for a discussion of planned statistical analyses). 301 A complete discussion of the historical Passenger Cars and LTVs—Final Report, National 300 Cost, Effectiveness and Deployment of Fuel analysis of vehicle mass and safety is located in Highway Traffic Safety Administration (Aug. 2012), Economy Technologies for Light-Duty Vehicles, Chapter 10 of the PRIA accompanying this available at https://crashstats.nhtsa.dot.gov/Api/ National Academy of Sciences (2015). proposed rulemaking. Public/ViewPublication/811665.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00125 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43110 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

rule, databases are the most up-to-date minivans in fatal crashes in the The adjustment for ESC, a feature of the possible (MY 2004–2011 vehicles in CY database); car, CUV, or minivan ≥ 3,187 analysis added in 2012, takes into 2006–2012), given the processing time pounds; truck-based LTV < 4,360 account results will be used to analyze for crash data and the need for enough pounds (the median curb weight of effects of mass reduction in future crash cases to permit statistically other truck-based LTVs in fatal crashes vehicles, which will all be ESC- meaningful analyses. As in previous in the database); and truck-based LTV ≥ equipped, as required by NHTSA’s analyses, NHTSA has made the new 4,360 pounds. An additional crash type regulations. databases available to the public on its includes all other fatal crash types (e.g., Techniques developed in the 2011 website, enabling other researchers to collisions involving more than two (preliminary) and 2012 (final) Kahane analyze the same data and hopefully vehicles, animals, or trains). Splitting reports have been retained to test minimizing discrepancies in results that the ‘‘other’’ vehicles into a lighter and statistical significance and to estimate would have been because of a heavier group permits more accurate 95 percent confidence bounds (sampling inconsistencies across databases. analyses of the mass effect in collisions error) for mass effects and to estimate of two light vehicles. Grouping partner- the combined annual effect of removing (c) Updated Analysis for This vehicle CUVs and minivans with cars 100 pounds of mass from every vehicle Rulemaking rather than LTVs is more appropriate (or of removing different amounts of The basic analytical method used to because their front-end profile and mass from the various classes of analyze the impacts of weight reduction rigidity more closely resembles a car vehicles), while holding footprint on safety in this proposed rule is the than a typical truck-based LTV. constant. same as in NHTSA’s 2012 Kahane The curb weight of passenger cars is NHTSA considered the near report, 2016 Puckett and Kindelberger formulated, as in the 2012 Kahane multicollinearity of mass and footprint report, and the Draft TAR: The agency report, 2016 Puckett and Kindelberger to be a major issue in the 2010 Kahane analyzed cross sections of the societal report, and Draft TAR, as a two-piece report 303 and voiced concern about fatality rate per billion vehicle miles of linear variable to estimate one effect of inaccurately estimated regression travel (VMT) by mass and footprint, mass reduction in the lighter cars and coefficients.304 High correlations while controlling for driver age, gender, another effect in the heavier cars. The between mass and footprint and and other factors, in separate logistic boundary between ‘‘lighter’’ and variance inflation factors (VIF) have not regressions by vehicle class and crash ‘‘heavier’’ cars is 3,201 pounds (which changed from MY 1991–1999 to MY type. ‘‘Societal’’ fatality rates include is the median mass of MY 2004–2011 2004–2011; large vehicles continued to fatalities to occupants of all the vehicles cars in fatal crashes in CY 2006–2012, be, on the average, heavier than small involved in the collisions, plus any up from 3,106 for MY 2000–2007 cars in vehicles to the same extent as in the pedestrians. CY 2002–2008 in the 2012 NHTSA previous decade.305 The temporal range of the data is now safety database, and up from 3,197 for Nevertheless, multicollinearity MY 2004–2011 vehicles in CY 2006– MY 2003–2010 cars in CY 2005–2011 in appears to have become less of a 2012, updated from previous databases the 2016 NHTSA safety database). problem in the 2012 Kahane, 2016 of MY 2000–2007 vehicles in CY 2002– Likewise, for truck-based LTVs, curb Puckett and Kindelberger/Draft TAR, 2008 (2012 Kahane Report) and MY weight is a two-piece linear variable and current NHTSA analyses. 2003–2010 vehicles in CY 2005–2011 with the boundary at 5,014 pounds Ultimately, only three of the 27 core (2016 Puckett and Kindelberger report (again, the MY 2004–2011 median, models of fatality risk by vehicle type in and Draft TAR). NHTSA purchased a higher than the median of 4,594 for MY the current analysis indicate the file of odometer readings by make, 2000–2007 LTVs in CY 2002–2008 and potential presence of effects of model, and model year from Polk that the median of 4,947 for MY 2003–2010 multicollinearity, with estimated effects helped inform the agency’s improved LTVs in CY 2005–2011). Curb weight is of mass and footprint reduction greater VMT estimates. As in the 2012 Kahane formulated as a simple linear variable than two percent per 100-pound mass report, 2016 Puckett and Kindelberger for CUVs and minivans. Historically, reduction and one-square-foot footprint report, and the Draft TAR, the vehicles CUVs and minivans have accounted for reduction, respectively; these three are grouped into three classes: Passenger a relatively small share of new-vehicle models include passenger cars and cars (including both two-door and four- sales over the range of the data, CUVs in first-event rollovers, and CUVs door cars); CUVs and minivans; and resulting in less crash data available in collisions with LTVs greater than truck-based LTVs. than for cars or truck-based LTVs. 4,360 pounds. This result is consistent There are nine types of crashes For a given vehicle class and weight with the 2016 Puckett and Kindelberger specified in the analysis. Single-vehicle range (if applicable), regression report, which also found only three crashes include first-event rollovers, coefficients for mass (while holding cases out of 27 models with estimated collisions with fixed objects, and footprint constant) in the nine types of effects of mass and footprint reduction collisions with pedestrians, bicycles and crashes are averaged, weighted by the greater than two percent per 100-pound motorcycles. Two-vehicle crashes number of baseline fatalities that would mass reduction and one-square-foot include collisions with: heavy-duty have occurred for the subgroup MY footprint reduction. vehicles; car, CUV, or minivan < 3,187 2008–2011 vehicles in CY 2008–2012 if Table II–45 presents the estimated pounds (the median curb weight of these vehicles had all been equipped percent increase in U.S. societal fatality other, non-case, cars, CUVs and with electronic stability control (ESC). risk per 10 billion VMT for each 100-

303 Kahane, C. J. Relationships Between Fatality 304 Van Auken and Green also discussed the issue 305 Greene, W. H. Econometric Analysis 266–68 Risk, Mass, and Footprint in Model Year 1991–1999 in their presentations at the NHTSA Workshop on (Macmillan Publishing Company 2d ed. 1993); Paul and Other Passenger Cars and LTVs (Mar. 24, Vehicle Mass-Size-Safety in Washington, DC D. Allison, Logistic Regression Using the SAS 2010), in Final Regulatory Impact Analysis: February 25, 2011. More information on the NHTSA System 48–51 (SAS Institute Inc. 2001). VIF scores Corporate Average Fuel Economy for MY 2012–MY Workshop on Vehicle Mass-Size-Safety is available are in the 6–9 range for curb weight and footprint at https://one.nhtsa.gov/Laws-&-Regulations/CAFE- in NHTSA’s new database—i.e., in the somewhat 2016 Passenger Cars and Light Trucks, National %E2%80%93-Fuel-Economy/NHTSA-Workshop- unfavorable 2.5–10 range where near Highway Traffic Safety Administration (Mar. 2010) on-Vehicle-Mass%E2%80%93Size%E2%80%93 multicollinearity begins to become a concern in at 464–542. Safety. logistic regression analyses.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00126 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43111

pound reduction in vehicle mass, while holding footprint constant, for each of the five vehicle classes:

None of the estimated effects have 95- estimated increases in societal fatality components of the analysis, as percent confidence bounds that exclude risk for mass reduction in the heavier discussed at the end of this section. zero, and thus are not statistically cars and the lighter truck-based LTVs, However, this degree of uncertainty is significant at the 95-percent confidence and the estimated decrease in societal methodological in nature rather than level. Two estimated effects are fatality risk for mass reduction in CUVs statistical. statistically significant at the 85-percent and minivans are not significant, even at It is useful to compare the new results level. Societal fatality risk is estimated the 85-percent confidence level. in Table II–45 to results in the 2012 to: (1) Increase by 1.2 percent if mass is Confidence bounds estimate only the Kahane report (MY 2000–2007 vehicles reduced by 100 pounds in the lighter sampling error internal to the data used in CY 2002–2008) and the 2016 Puckett cars; and (2) decrease by 0.61 percent if in the specific analysis that generated and Kindelberger report and Draft TAR mass is reduced by 100 pounds in the the point estimate. Point estimates are (MY 2003–2010 vehicles in CY 2005– heavier truck-based LTVs. The also sensitive to the modification of 2011), presented in Table II–46 below:

New results are directionally the same passenger cars; for all reports, the while the corresponding estimate in the as in 2012; in the 2016 analysis, the estimate for lighter passenger cars is 2016 Puckett and Kindelberger report estimate for lighter LTVs was of statistically significant at the 85-percent was insignificant and negative. opposite sign (but small magnitude). confidence level, while the estimate for Vehicle mass continued an historical Consistent with the 2012 Kahane and heavier passenger cars is insignificant. upward trend across the MYs in the 2016 Puckett and Kindelberger reports, The estimated mass effect for heavier newest databases. The average (VMT- mass reductions in lighter cars are truck-based LTVs is stronger in this weighted) masses of passenger cars and estimated to lead to increases in analysis and in the 2016 Puckett and CUVs both increased by approximately Kindelberger report than in the 2012 three percent from MYs 2004 to 2011 fatalities, and mass reductions in Kahane report; both estimates are (3,184 pounds to 3,289 pounds for heavier LTVs are estimated to lead to statistically significant at the 85-percent passenger cars, and 3,821 pounds to decreases in fatalities. However, NHTSA confidence level, unlike the 3,924 pounds for CUVs). Over the same does not consider this conclusion to be corresponding insignificant estimate in period, the average mass of minivans definitive because of the relatively wide the 2012 Kahane report. The estimated increased by six percent (from 4,204 confidence bounds of the estimates. The mass effect for lighter truck-based LTVs pounds to 4,462 pounds), and the estimated mass effects are similar is insignificant and positive in this average mass of LTVs increased by 10% among analyses for both classes of analysis and the 2012 Kahane report, (from 4,819 pounds to 5,311 pounds).

306 Median curb weights in the 2012 Kahane truck-based LTVs. Median curb weights in the 2016 Puckett and Kindelberger report: 3,197 pounds for report: 3,106 pounds for cars, 4,594 pounds for cars, 4,947 pounds for truck-based LTVs.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00127 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.067 EP24AU18.068 43112 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Historical reasons for mass increases The principal difference between vehicle. Through conservation of within vehicle classes include: heavier vehicles, especially truck-based momentum, the degree to which the Manufacturers discontinuing lighter LTVs, and lighter vehicles, especially delta V in the lighter vehicle is greater models; manufacturers re-designing passenger cars, is mass reduction has a than in the heavier vehicle is models to be heavier and larger; and different effect in collisions with proportional to the ratio of mass in the shifting consumer preferences with another car or LTV. When two vehicles heavier vehicle to mass in the lighter respect to cabin size and overall vehicle of unequal mass collide, the change in vehicle: size. velocity (delta V) is greater in the lighter

Because fatality risk is a positive lower, they may more successfully The regression results are best suited function of delta V, the fatality risk in avoid crashes or reduce the severity of to predict the effect of a small change in the lighter vehicle in two-vehicle crashes. mass, leaving all other factors, including collisions is also higher. Removing some Farmer, Green, and Lie, who reviewed footprint, the same. With each mass from the heavy vehicle reduces the 2010 Kahane report, again peer- additional change from the current delta V in the lighter vehicle, where reviewed the 2011 Kahane report.307 In environment (e.g., the scale of mass fatality risk is higher, resulting in a large preparing his 2012 report (along with change, presence and prevalence of benefit, offset by a small penalty the 2016 Puckett and Kindelberger safety features, demographic because delta V increases in the heavy report and Draft TAR), Kahane also took characteristics), the model may become vehicle where fatality risk is low— into account Wenzel’s 308 assessment of less accurate. It is recognized that the adding up to a net societal benefit. the preliminary report and its peer light-duty vehicle fleet in the MY 2021– Removing some mass from the lighter reviews, DRI’s analyses published early 2026 timeframe will be different from vehicle results in a large penalty offset in 2012, and public comments such as the MY 20042011 fleet analyzed here. by a small benefit—adding up to net the International Council on Clean Nevertheless, one consideration harm. Transportation’s comments submitted provides some basis for confidence in applying regression results to estimate These considerations drive the overall on NHTSA and EPA’s 2010 notice of 309 effects of relatively large mass result: Mass reduction is associated with joint rulemaking. These comments prompted supplementary analyses, reductions or mass reductions over an increase in fatality risk in lighter longer periods. This is NHTSA’s sixth cars, a decrease in fatality risk in especially sensitivity tests, discussed at the end of this section. evaluation of effects of mass reduction heavier LTVs, CUVs, and minivans, and and/or downsizing,310 comprising has smaller effects in the intermediate 307 Items 0035 (Lie), 0036 (Farmer) and 0037 groups. Mass reduction may also be 310 As outlined throughout this section, NHTSA’s (Green) in Docket No. NHTSA–2010–0152. harmful in a crash with a movable six related studies include the new analysis 308 Wenzel, T. An Analysis of the Relationship object such as a small tree, which may supporting this rulemaking, and: Kahane, C. J. Between Casualty Risk Per Crash and Vehicle Mass Vehicle Weight, Fatality Risk and Crash break if hit by a high mass vehicle and Footprint for Model Year 2000–2007 Light Duty Compatibility of Model Year 1991–99 Passenger resulting in a lower delta V than may Vehicles, Lawrence Berkeley National Laboratory Cars and Light Trucks, National Highway Traffic occur if hit by a lower mass vehicle (Dec. 2011), available at http://eta- Safety Administration (Oct. 2003), available at publications.lbl.gov/sites/default/files/lbnl- which does not break the tree and https://crashstats.nhtsa.dot.gov/Api/Public/View 5695e.pdf; Tom Wenzel, Lawrence Berkeley Publication/809662; Kahane, C. J. Relationships therefore has a higher delta V. However, National Laboratory -Assessment of NHTSA Report Between Fatality Risk, Mass, and Footprint in in some types of crashes not involving Relationships Btw Fatality Risk Mass and Footprint Model Year 1991–1999 and Other Passenger Cars collisions between cars and LTVs, in MY 2000–2007 PC and LTV,’’ Docket NHTSA– and LTVs (Mar. 24, 2010), in Final Regulatory 2010–0131–0315; and a peer review of Wenzel’s especially first-event rollovers and Impact Analysis: Corporate Average Fuel Economy reports—Peer Review of LBNL Statistical Analysis for MY 2012–MY 2016 Passenger Cars and Light impacts with fixed objects, mass of the Effect of Vehicle Mass & Footprint Reduction Trucks, National Highway Traffic Safety reduction may not be harmful and may on Safety (LBNL Phase 1 and 2 Reports), prepared Administration (Mar. 2010) at 464–542; Kahane, C. be beneficial. To the extent lighter for U.S. EPA (Feb. 2012), available at Docket ID J. Relationships Between Fatality Risk, Mass, and NHTSA–2010–0131–0328. vehicles may respond more quickly to Footprint in Model Year 2000–2007 Passenger Cars 309 Comment by International Council on Clean and LTVs—Preliminary Report, National Highway braking and steering, or may be more Transportation, Docket ID NHTSA–2010–0131– Traffic Safety Administration (Nov. 2011), available stable because their center of gravity is 0258. at Docket ID NHTSA–2010–0152- 0023); Kahane, C.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00128 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.069 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43113

databases ranging from MYs 1985 to (d) Calculation of MY 2021–2026 Safety The agency assumed safety trends 2011. Impact will result in a reduction in the target Results of the six studies are not Neither CAFE standards nor this population of fatalities from which the identical, but they have been consistent analysis mandate mass reduction, or vehicle mass impacts are derived. Table to a point. During this time period, mandate mass reduction occur in any II–47 through Table II–52 show results many makes and models have increased specific manner. However, mass of NHTSA’s vehicle mass-size-safety substantially in mass, sometimes as reduction is one of the technology analysis over the cumulative lifetime of much as 30–40%.311 If the statistical applications available to manufacturers, MY 1977–2029 vehicles, for both the analysis has, over the past years, been and thus a degree of mass reduction is CAFE and GHG programs, based on the able to accommodate mass increases of allowed within the CAFE model to: (1) MY 2016 baseline fleet, accounting for this magnitude, perhaps it will also Determine capabilities of manufacturers; the projected safety baselines. The and (2) to predict cost and fuel succeed in modeling effects of mass reported fatality impacts are consumption effects of improved CAFE undiscounted, but the monetized safety reductions of approximately 10–20%, standards. should they occur in the future. impacts are discounted at three-percent The agency utilized the relationships and seven-percent discount rates. The between weight and safety from the new reported fatality impacts are estimated NHTSA analysis, expressed as percentage increases in fatalities per increases or decreases in fatalities over J. Relationships Between Fatality Risk, Mass, and the lifetime of the model year fleet. A Footprint in Model Year 2000–2007 Passenger Cars 100-pound weight reduction, and and LTVs: Final Report, NHTSA Technical Report. examined the weight impacts assumed positive number means that fatalities are Washington, DC: NHTSA, Report No. DOT–HS– in this CAFE analysis. The effects of projected to increase; a negative number 811–665; and Puckett, S. M., & Kindelberger, J. C. mass reduction on safety were estimated (in parentheses) means that fatalities are Relationships between Fatality Risk, Mass, and relative to estimated baseline levels of projected to decrease. Footprint in Model Year 2003–2010 Passenger Cars and LTVs—Preliminary Report, National Highway safety across vehicle classes and model Results are driven extensively by the Traffic Safety Administration (June 2016), available years. To identify baseline levels of degree to which mass is reduced in at https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/ safety, the agency examined effects of relatively light passenger cars and in 2016-prelim-relationship-fatalityrisk-mass- identifiable safety trends over lifetimes relatively heavy vehicles because their footprint-2003-10.pdf. of vehicles produced in each model coefficients in the logistic regression 311 For example, one of the most popular models year. The projected effectiveness of of small 4-door sedans increased in curb weight analysis have the most significant existing and forthcoming safety values. We assume any impact on from 1,939 pounds in MY 1985 to 2,766 pounds in technologies and expected on-road fleet fatalities will occur over the lifetime of MY 2007, a 43% increase. A high-sales mid-size penetration of safety technologies were sedan grew from 2,385 to 3,354 pounds (41%); a the vehicle, and the chance of a fatality best-selling pickup truck from 3,390 to 4,742 incorporated into observed trends in fatality rates to estimate baseline fatality occurring in any particular year is pounds (40%) in the basic model with two-door cab directly related to the weighted vehicle and rear-wheel drive; and a popular minivan from rates in future years across vehicle 2,940 to 3,862 pounds (31%). classes and model years. miles traveled in that year.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00129 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43114 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00130 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.070 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43115

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00131 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.071 43116 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00132 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.072 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43117

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00133 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.073 43118 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00134 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.074 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43119

For all light-duty vehicles, mass decrease in fatalities over the vehicles in all alternatives evaluated. changes are estimated to lead to a cumulative lifetime of MY 1977–2029 The effects of mass changes on fatalities

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00135 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.075 43120 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

range from a combined decrease that to social costs of fatalities, but in relative to the augural standards at a (relative to the augural standards, the this case NHTSA does not have a model three-percent discount rate and by baseline) of 12 fatalities for Alternative estimating the effect of vehicle mass on between $97 million and $1.6 billion at #7 to a combined decrease of 173 injuries. Blincoe et al. estimates that a seven-percent discount rate. The fatalities for Alternative #4. The fatalities account for 39.5% of total estimated decreases in social safety difference in results by alternative comprehensive costs due to injury.312 If costs are driven by estimated decreases depends upon how much weight vehicle mass impacts non-fatal injuries in costs associated with passenger cars, reduction is used in that alternative and proportionally to its impact on fatalities, ranging from $264 million (for the types and sizes of vehicles to which then total costs would be approximately Alternative #7) to $4.4 billion (for the weight reduction applies. The 2.53 (1⁄0.395) times the value of fatalities Alternative #1) relative to the Augural decreases in fatalities are driven by alone or around $25.07 million per standards at a three-percent discount impacts within passenger cars fatality. NHTSA has selected this value (decreases of between 17 and 281 as representative of the relationship rate and by between $146 million and fatalities) and are offset by impacts between fatality costs and injury costs $2.5 billion at a seven-percent discount within light trucks (increases of between because this approach is internally rate. The estimated decreases in costs 6 and 120 fatalities). consistent among NHTSA studies. associated with passenger cars are offset Additionally, social effects of Changes in vehicle mass are estimated by estimated increases in costs increasing fatalities can be monetized to decrease social safety costs over the associated with light trucks, ranging using NHTSA’s estimated lifetime of the nine model years by from $88 million (for Alternative #7) to comprehensive cost per life of between $176 million (for Alternative $2.0 billion (for Alternative #1) relative $9,900,000 in 2016 dollars. This #7) and $2.7 billion (for Alternative #4) to the Augural standards at a three- consists of a value of a statistical life of percent discount rate and by between $9.6 million in 2015 dollars plus 312 Blincoe, L. et al., The Economic and Social $49 million and $1.3 billion at a seven- external economic costs associated with Impact of Motor Vehicle Crashes, 2010 (Revised), National Highway Traffic Safety Administration percent discount rate. fatalities such as medical care, (May 2015), available at https:// Table II–53 through Table II–55 insurance administration costs and legal crashstats.nhtsa.dot.gov/Api/Public/View presents average annual estimated safety costs, updated for inflation to 2016 Publication/812013. The estimate of 39.5% (see dollars. Table 1–8) is equal to the estimated value of MAIS6 effects of vehicle mass changes, for CYs (fatal) injuries in vehicle incidents divided by the 2035–2045: Typically, NHTSA would also estimated value of MAIS0–MAIS6 (non-fatal and estimate the effect on injuries and add fatal) injuries in vehicle incidents.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00136 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43121

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00137 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.076 43122 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00138 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.077 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43123

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00139 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.078 43124 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00140 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.079 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43125

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00141 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.080 43126 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

For all light-duty vehicles, mass average annual decrease in fatalities in 2045. The effects of mass changes on changes are estimated to lead to an all alternatives evaluated for CYs 2035– fatalities range from a combined

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00142 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.081 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43127

decrease (relative to the Augural costs in CY 2035–2045 by between $2 increases in costs associated with light standards) of 1 fatality per year for million (for Alternative #7) and $271 trucks, decreasing between $11 million Alternative #7 to a combined increase of million (for Alternative #1) relative to (for Alternative #7) and $153 million 22 fatalities per year for Alternative #1. the Augural standards at a three-percent (for Alternative #1) relative to the The difference in the results by discount rate and by between $1 million Augural standards at a three-percent alternative depends upon how much and $111 million at a seven-percent discount rate and decreasing between $5 weight reduction is used in that discount rate. The estimated decreases million and $64 million at a seven- alternative and the types and sizes of in social safety costs are generally percent discount rate. vehicles to which the weight reduction driven by estimated decreases in costs applies. The decreases in fatalities are associated with passenger cars, To help illuminate effects at the generally driven by impacts within decreasing between $13 million (for model year level, Table II–59 presents passenger cars (decreases of between 1 Alternative #7) and $424 million (for the lifetime fatality impacts associated and 33 fatalities per year relative to the Alternative #1) relative to the Augural with vehicle mass changes for passenger Augural standards) and are generally standards at a three-percent discount cars, light trucks, and all light-duty offset by impacts within light trucks rate and decreasing between $5 million vehicles by model year under (increases of between 1 and 12 fatalities and $175 million at a seven-percent Alternative #1, relative to the Augural per year). discount rate. The estimated decreases standards for the CAFE Program. Table Changes in vehicle mass are estimated in costs associated with passenger cars II–59 presents an analogous table for the to decrease average annual social safety are generally offset by estimated GHG Program.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00143 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43128 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Under Alternative #1, passenger car fatalities), peaking in MY 2025 (37 in MYs 2017 and 2018 to an increase of fatalities associated with mass changes fatalities). Corresponding estimates of 14 fatalities in MYs 2026 through 2029. are estimated to decrease generally from light truck fatalities associated with Altogether, light-duty vehicle fatality MY 2017 (decrease of three fatalities) mass changes are generally positive, reductions associated with mass through MY 2029 (decrease of 36 ranging from a decrease of one fatality changes under Alternative #1 are

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00144 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.082 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43129

estimated to be concentrated among MY Table II–61 and Table II–62 present respectively for the CAFE Program. 2023 through MY 2029 vehicles (146 out estimates of monetized lifetime social Table II–63 and Table II–64 show of 165, or 91% of net fatalities safety costs associated with mass comparable tables from the perspective mitigated). changes by model year at three-percent of the GHG Program. and seven-percent discount rates,

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00145 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43130 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00146 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.083 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43131

Lifetime social safety costs are model year, with decreases associated partially by increases associated with estimated to decrease generally by with passenger cars generally offset light trucks. At a three-percent discount

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00147 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.084 43132 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

rate, decreases in lifetime social safety a seven-percent discount rate, 97% of Green (UMTRI) in their peer reviews; costs related to passenger cars are the reduction in total lifetime costs Van Auken (DRI) in his public estimated to range from $13 million for ($486 million out of $501 million) is comments; or Wenzel in his parallel existing (MY 1977 through MY 2016) attributed to MY 2023 through MY 2029 research for DOE. The 2012 Kahane and cars, to $230 million for MY 2025 cars. light-duty vehicles. 2016 Puckett and Kindelberger reports The corresponding estimates at a seven- (e) Sensitivity Analyses provide further discussion of the models percent discount rate range from $7 and the rationales behind them. Table II–65 shows the principal million to $136 million. At a three- Alternative models use NHTSA’s percent discount rate, impacts on findings and includes sampling-error confidence bounds for the five databases and regression-analysis lifetime social safety costs related to approach but differ from the baseline light trucks are estimated to range from parameters used in the CAFE model. The confidence bounds represent the model in one or more explanatory a decrease of $5 million for MY 2017 variables, assumptions, or data light trucks to an increase of $96 million statistical uncertainty that is a restrictions. NHTSA applied the 11 for MY 2022 light trucks. The consequence of having less than a techniques to the latest databases to corresponding estimates at a seven- census of data. NHTSA’s 2011, 2012, generate alternative CAFE model percent discount rate range from $3 and 2016 reports acknowledged another coefficients. The range of estimates million to $65 million. source of uncertainty: The baseline Consistent with the analysis of fatality statistical model can be varied by produced by the sensitivity tests offers impacts by model year in Table II–61, choosing different control variables or insight to the uncertainty inherent in decreases in lifetime social safety costs redefining the vehicle classes or crash the formulation of the models, subject to associated with mass changes are types, which for example, could the caveat these 11 tests are, of course, generally concentrated in MY 2023 produce different point estimates. not an exhaustive list of conceivable through MY 2029 light-duty vehicles Beginning with the 2012 Kahane alternatives. under Alternative #1. At a three-percent report, NHTSA has provided results of The baseline and alternative results discount rate, 93% of the reduction in 11 plausible alternative models that follow, ordered from the lowest to the total lifetime costs ($872 million out of serve as sensitivity tests of the baseline highest estimated increase in societal $937 million) is attributed to MY 2023 model. Each alternative model was risk per 100-pound reduction for cars through MY 2029 light-duty vehicles; at tested or proposed by: Farmer (IIHS) or weighing less than 3,201 pounds:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00148 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.085 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43133

The sensitivity tests illustrate both the unbelted driver occupants were • 2009/2010 model year Toyota fragility and the robustness of baseline considered in this fleet simulation. The Venza finite element baseline vehicle estimates. On the one hand, the occupant injury risk from each and two design variants included a 20% variation among NHTSA’s coefficients is simulation was calculated and summed light-weight vehicle model (2010 Venza) quite large relative to the baseline to obtain combined occupant injury (Low option mass reduction vehicle estimate: In the preceding example of risk. The combined occupant injury risk funded by EPA and International cars < 3,201 pounds, the estimated was weighted according to the Council on Clean Transportation (ICCT)) coefficients range from almost zero to frequency of real world occurrences to and a 35% light-weight vehicle (2009 almost double the baseline estimate. develop overall societal risk for baseline Venza) (High option mass reduction This result underscores the key and light-weighted vehicles. Note: The vehicle funded by California Air relationship that the societal effect of generic restraint system developed and Resources Board).317 mass reduction is small and, as Wenzel used in the baseline occupant Light weight vehicles were designed has said, it ‘‘is overwhelmed by other simulations was also used in the light- to have similar vehicle crash pulses as known vehicle, driver, and crash weighted vehicle occupant simulations baseline vehicles. More than 440 vehicle factors.’’ 313 In other words, varying how as the purpose of this fleet simulation crash simulations were conducted for to model some of these other vehicle, was to understand changes in societal the range of crash speeds and crash driver, and crash factors, which is injury risks because of mass reduction configurations to generate crash pulse exactly what sensitivity tests do, can for different classes of vehicles in and intrusion data points. The crash appreciably change the estimate of the frontal crashes. No modifications to the pulse data and intrusion data points societal effect of mass reduction. restraint systems were made for light- will be used as inputs in the occupant On the other hand, variations are not weighted vehicle occupant simulations. simulation models. particularly large in absolute terms. The Any modifications to restraint systems For vehicle to vehicle impact ranges of alternative estimates are to improve occupant injury risks or simulations, four finite element models generally in line with the sampling-error societal injury risks in the light- were chosen to represent the fleet. The confidence bounds for the baseline weighted vehicle would have conflated partner vehicle models were selected to estimates. Generally, in alternative results without identifying effects of represent a range of vehicle types and models as in the baseline models, mass mass reduction only. The following weights. It was assumed vehicle models reduction tends to be relatively more sections provide an overview of the fleet would reflect the crash response for all harmful in the lighter vehicles and more simulation study: vehicles of the same type, e.g. mid-size beneficial in the heavier vehicles, just as NHTSA contracted with George car. Only the safety or injury risk for the they are in the central analysis. In all Washington University to develop a driver in the target vehicle and in the models, the point estimate of NHTSA’s fleet simulation model 314 to study the partner vehicle were evaluated in this coefficient is positive for the lightest impact and relationship of light- study. vehicle class, cars < 3,201 pounds. In weighted vehicle design with injuries As noted, vehicle simulations nine out of 11 models, the point and fatalities. In this study, there were generated vehicle deformations and estimate is negative for CUVs and eight vehicles as follows: acceleration responses utilized to drive minivans, and in eight out of 11 models • 2001 model year Ford Taurus finite occupant restraint simulations and the point estimate is negative for LTVs element model baseline and two simple ≥ predict the risk of injury to the head, 5,014 pounds. design variants included a 25% lighter neck, chest, and lower extremities. In (f) Fleet Simulation Model vehicle while maintaining the same all, more than 1,520 occupant restraint vehicle front end stiffness and 25% simulations were conducted to evaluate NHTSA has traditionally used real overall stiffer vehicle while maintaining the risk of injury for mid-size male and world crash data as the basis for the same overall vehicle mass.315 small female drivers. projecting the future safety implications • 2011 model year Honda Accord The computed societal injury risk for regulatory changes. However, finite element baseline vehicle and its (SIR) for a target vehicle v in frontal because lightweight vehicle designs are 20% light-weight vehicle designed by crashes is an aggregate of individual introducing fundamental changes to the Electricore. (This mass reduction study serious crash injury risks weighted by structure of the vehicle, there is some was sponsored by NHTSA).316 concern that historical safety trends may real-world frequency of occurrence (v) not apply. To address this concern, of a frontal crash incident. A crash 314 Samaha, R. R. et al., Methodology for incident corresponds to a crash with NHTSA developed an approach to Evaluating Fleet Protection of New Vehicle Designs: utilize lightweight vehicle designs to Application to Lightweight Vehicle Designs, different partners (Npartner) at a given evaluate safety in a subset of real-world National Highway Traffic Safety Administration impact speed (Pspeed), for a given representative crashes. The (Aug. 2014), available at https://www.nhtsa.gov/ driver occupant size (Loccsize), in the crashworthiness/vehicle-aggressivity-and-fleet- target or partner vehicle (T/P), in a given methodology focused on frontal crashes compatibility-research (accessed by clicking on the because of the availability of existing .zip file for DOT HS 812 051). crash configuration (Mconfig), and in a vehicle and occupant restraint models. 315 Samaha, R. R. et al., Methodology for single- or two-vehicle crash (Kevent). Representative crashes were simulated Evaluating Fleet Protection of New Vehicle Designs: CIR (v) represents the combined injury Application to Lightweight Vehicle Designs, between baseline and lightweight risk (by body region) in a single crash appendices, National Highway Traffic Safety incident. (v) designates the weighting vehicles against a range of vehicles and Administration (Aug. 2014), available at https:// roadside objects using two different size www.nhtsa.gov/crashworthiness/vehicle- factor, i.e., percent of occurrence, belted driver occupants (adult male and aggressivity-and-fleet-compatibility-research derived from National Automotive (accessed by clicking on the .zip file for DOT HS small female) only. No passenger(s) or Sampling System Crashworthiness Data 812 051 [appendices are Part 2]). System (NASS CDS) for the crash 316 Singh, H. et al., Update to future midsize 313 Wenzel, T. Assessment of NHTSA’s Report lightweight vehicle findings in response to incident. A driver age group of 16 to 50 ‘‘Relationships Between Fatality Risk, Mass, and manufacturer review and IIHS small-overlap Footprint in Model Year 2000–2007 Passenger Cars testing, National Highway Traffic Safety 317 Light-Duty Vehicle Mass Reduction and Cost and LTVs,’’ Lawrence Berkeley National Laboratory Administration (Feb. 2016), available at https:// Analysis — Midsize Crossover Utility Vehicle, U.S. at iv (Nov. 2011), available at Docket ID NHTSA– www.nhtsa.gov/sites/nhtsa.dot.gov/files/812237_ EPA (Aug. 2012), https://cfpub.epa.gov/si/si_ 2010–0152–0026. lightweightvehiclereport.pdf. public_record_report.cfm?dirEntryID=230748.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00149 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43134 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

years old was chosen to provide a will provide insight into the estimate of In general, the societal injury risk in population with a similar, i.e., more modification needed in the restraint and the frontal crash simulation associated consistent, injury tolerance. airbag systems of lightweight vehicles. If with the small size driver is elevated The fleet simulation was performed the difference extends beyond the when compared to that of the mid-size using the best available engineering expected baseline vehicle restraint and driver. However, both occupant sizes models, with base vehicle restraint and airbag capability, then adjustments to had reasonable injury risk in the airbag settings, to estimate societal risks the structural designs would be needed. simulated impact configurations of future lightweight vehicles. The range Results from the fleet simulation study representative of the regulatory and of the predicted risks for the baseline show the trend of increased societal consumer information testing. NHTSA vehicles is from 1.25% to 1.56%, with injury risk for light-weighted vehicle examined three methods for combining an average of 1.39%, for the NASS designs, as compared to their baselines, injuries with different body regions. frontal crashes that were simulated. The occurs for both single vehicle and two- One observation was the baseline mid- change in driver injury risk between the vehicle crashes. Results are listed in size CUV model was more sensitive to baseline and light-weighted vehicles Table II–66. leg injuries.

This study only looked at lightweight the fleet simulation societal risk to the vehicle designs. As explained earlier, designs for a midsize sedan and a mid- 2016 update of the NHTSA 2012 the fleet simulation study assumed size CUV and did not examine safety regression analysis and the updated restraint equipment to be as in the implications for heavier vehicles. The analysis used in this NPRM, the risk baseline model, in which restraints/ study was also limited to only frontal assessment from the fleet simulation is airbags are not redesigned to be optimal crash configurations and considered just similarly directionally consistent with with light-weighting. mid-size CUVs whereas the statistical the passenger car risk assessment from 2. Impact of Vehicle Scrappage and regression model considered all CUVs the regression analysis. As noted, fleet Sales Response on Fatalities and all crash modes. simulations were performed only in Previous versions of the CAFE model, The change in the safety risk from the frontal crash mode and did not consider and the accompanying regulatory MY 2010 fleet simulation study was other crash modes including rollover 319 analyses relying on it, did not carry a directionally consistent with results for crashes. This fleet simulation study does not representation of the full on-road passenger cars from NHTSA 2012 provide information that can be used to vehicle population, only those vehicles regression analysis study,318 which modify coefficients derived for the from model years regulated under covered data for MY 2000–MY 2007. NPRM regression analysis because of proposed (or final) standards. The The NHTSA 2012 regression analysis the restricted types of crashes 320 and omission of an on-road fleet implicitly study was updated in 2016 to reflect assumed the population of vehicles newer MY 2003 to MY 2010. Comparing 319 The risk assessment for CUV in the regression registered at the time a set of CAFE model combined CUVs and minivans in all crash standards is promulgated is not affected 318 The 2012 Kahane study considered only modes and included belted and unbelted fatalities, whereas, the fleet simulation study occupants. by those standards. However, there are considered severe (AIS 3+) injuries and fatalities 320 The fleet simulation considered only frontal several mechanisms by which CAFE (DOT HS 811 665). crashes. standards can affect the existing vehicle

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00150 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.086 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43135

population. The most significant of meet travel demand through the use of 3. Safety Model these is deferred retirement of older older, less safe vehicles. As an The analysis supporting the CAFE vehicles. CAFE standards force illustration, if we simplify by ignoring rule for MYs 2017 and beyond did not manufacturers to apply fuel saving that the share of new vehicles that are account for differences in exposure or technologies to offered vehicles and passenger cars changes with the inherent safety risk as vehicles aged then pass along the cost of those stringency of the alternatives, simply throughout their useful lives. However, technologies (to the extent possible) to changing the number of new vehicles the relationship between vehicle age buyers of new vehicles. These price between scenarios affects the mileage and fatality risk is an important one. In increases affect the length of loan terms accumulation of the fleet and therefore a 2013 Research Note,321 NHTSA’s and the desired length of ownership for all fleet level effects. Reducing the National Center for Statistics and new vehicle buyers and can discourage number of new vehicles sold, relative to Analysis concluded a driver of a vehicle some buyers on the margin from buying a baseline forecasted value, reduces the that is four to seven years old is 10% a new vehicle in a given year. To the size of the registered vehicle fleet that more likely to be killed in a crash than extent new vehicle purchases offset is able to service the underlying demand the driver of a vehicle zero to three pending vehicle retirements, delaying for travel. years old, accounting for the other new purchases in favor of continuing to Consider a simple example where we factors related to the crash. This trend use an aging vehicle affects the overall show sales effects operating on a micro- safety of the on-road fleet even if the continued for older vehicles more scale for a single household whose generally, with a driver of a vehicle 18 vehicle whose retirement was delayed choices of whether to purchase a new was not directly subject to a binding years or older being 71% more likely to vehicle is affected by vehicle price. A be killed in a crash than a driver in a CAFE standard in the model year during household starts with three vehicles, its production. new vehicle. While there are more aged three, five, and eight years old. In registered vehicles that are zero to three The sales response in the CAFE model a scenario with no CAFE standards and acts to modify new vehicle sales in two years old than there are 20 years or therefore no related changes in vehicle older (nearly three times as many) ways: sales prices, the household buys a new 1. Changes in new vehicle prices because most of the vehicles in earlier car and scraps the eight-year old car; the either increase or decrease total sales vintages are retired sooner, the average other two cars in the fleet each get a year (passenger cars and light trucks age of vehicles in the United States is older. In a scenario where CAFE combined) each year in the context of 11.6 years old and has risen standards become more stringent 322 forecasted macroeconomic conditions. significantly in the past decade. This 2. Changes in new vehicle attributes causing vehicle sales prices to increase, relationship reflects a general trend and fuel prices influence the share of this household chooses to delay buying visible in the Fatality Analysis new vehicles sold that are light trucks, a new car and each of their three Reporting System (FARS) when looking and therefore also passenger cars. existing cars gets a year older. In both at a series of calendar years: Newer These two responses change the total cases, all three vehicles (including the vintages are safer than older vintages, number of new vehicles sold in each new car in the first scenario, and the over time, at each age. This is likely model year across regulatory year-year-old car in the second scenario) because of advancements in safety alternatives and the relative proportion have to serve the family’s travel technology, like side-impact airbags, of new vehicles that are passenger cars demand. electronic stability control, and (more and light trucks. This response has two The scrappage effect is visible in the recently) sophisticated crash avoidance effects on safety. The first response household’s vehicle fleet as it moves systems starting to work their way into slows the rate at which new vehicles, from the first scenario to the second the vehicle population. In fact, the 2013 and their associated safety scenario with changes in CAFE Research Note indicated that the improvements, enter the on-road standards. In the second scenario, the percentage of occupants fatally injured population. The second response nine-year-old car remains in the in fatal crashes increased with vehicle influences the mix of vehicles on the household’s fleet to service demand for age: From 27% for vehicles three or road—with more stringent CAFE travel, when it would otherwise have fewer years old, to 41% for vehicles 12– standards leading to a higher share of been retired. While the scrappage effect 14 years old, to 50% for vehicles 18 or light trucks sold in the new vehicle can be symmetrical to the sales effect, it more years old. market, assuming all else is equal. Light need not be. The ‘‘new car’’ in the With an integrated fleet model now trucks have higher rates of fatal crashes scenario without CAFE standards could part of the analytical framework for when interacting with passenger cars be a new vehicle from the current model CAFE analysis, any effects on fleet and, as earlier sections discussed, year or a used car that is of a newer turnover (either from delayed vehicle different directional responses to mass vintage than the 8-year-old vehicle it retirement or deferred sales of new reduction technology based on the replaces. The latter instance is an effect vehicles) will affect the distribution of existing mass and body style of the of scrappage decisions that do not both ages and model years present in vehicle. directly affect new vehicle sales. the on-road fleet. Because each of these The sales response and scrappage Eventually, new vehicles transition to vintages carries with it inherent rates of response influence safety outcomes the used car market, but that on average fatal crashes, and newer vintages are through the same basic mechanism, fleet take several years, and the shift is slow. generally safer than older ones, turnover. In the case of the scrappage At the household level, the scrappage changing that distribution will change response, delaying fleet turnover keeps decision occurs in a single year, each drivers in older vehicles likely to be less year, for every vehicle in the fleet. To 321 National Center for Statistics and Analysis, safe than newer model year vehicles the extent CAFE standards affect new How Vehicle Age and Model Year Relate to Driver that could replace them. Similarly, vehicle prices and fuel economies, Injury Severity in Fatal Crashes, National Highway delaying the sale of new vehicles can relative to vehicles already owned, Traffic Safety Administration (Aug. 2013), available at https://crashstats.nhtsa.dot.gov/Api/Public/View force households to keep older vehicles scrappage could accelerate or decelerate Publication/811825. in use longer, reallocate VMT within depending upon the direction (and 322 Based on data acquired from Ward’s their household fleet, and generally magnitude) of the changes. Automotive.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00151 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43136 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

the total number of on-road fatalities this level; the safety analysis in the explicitly accounts for: The distribution under each regulatory alternative. CAFE model is statistical, using of vehicle ages in the fleet and the To estimate the empirical relationship aggregate values to represent the totality number of miles driven by those between vehicle age, model year of fleet interactions over time. The vehicles at each age. To accomplish this, vintage, and fatalities, DOT conducted a second challenge is perhaps the more DOT used data from the FARS database statistical analysis linking data from the significant of the two: The CAFE at a lower level of resolution; rather FARS database, a time series of Polk analysis is inherently forward-looking. than looking at each crash and the registration data to represent the on- To implement a statistical model specific factors that contributed to its road vehicle population, and assumed analogous to the one developed by occurrence, staff looked at the total per-vehicle mileage accumulation rates NCSA, the CAFE model would require number of fatal crashes involving light- (the derivation of which is discussed in forecasts of all factors considered in the duty vehicles over time with a focus on detail in PRIA Chapter 11). These data NCSA model—about vehicle speeds in the influence of vehicle age and vehicle were used to construct per-mile fatality crashes, driver behavior, driver and vintage. When considering the number rates that varied by vehicle vintage, passenger ages, vehicle vintages, and so of fatalities relative to the number of accounting for the influence of vehicle on. In particular, the model would registered vehicles for a given model age. However, unlike the NCSA study require distributions (joint distributions, year (without regard to the passenger referenced above, any attempt to in most cases) of these factors over a car/light-truck distinction, which has account for this relationship in the period of time spanning decades. Any evolved over time and can create CAFE analysis faces two challenges. The such forecasts would be highly inconsistent comparisons), a somewhat first challenge is the CAFE model lacks uncertain and would be likely to assume noisy pattern develops. Using data from the internal structure to account for a continuation of current conditions. calendar year 1996 through 2015, some other factors related to observed fatal Instead of trying to replicate the consistent stories develop. The points in crashes—for example, vehicle speed, NCSA work at a similar level of detail, Figure II–4 represent the number of seat belt use, drug use, or age of DOT conducted a simpler statistical fatalities per registered vehicle with involved drivers or passengers. Vehicle analysis to separate the safety impact of darker circles associated with interactions are simply not modeled at the two factors the CAFE model increasingly current calendar years.

As shown in Figure II–4, fatalities per vehicle) start out at a low point, rise schedule on the registered population registered vehicle have generally through age 15 or so, then decline and examining fatalities per billion declined over time across all vehicle through age 30 (at which point little of miles of VMT. ages (the darker points representing the initial model year cohort is still The mileage accumulation schedule newer vintages being closer to the x- registered). While this pattern is evident used in this analysis was developed axis) and, across most recent calendar in the registration data, it is magnified using odometer readings of vehicles years, fatality rates (per registered by imposing a mileage accumulation aged 0–15 years in calendar year 2015.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00152 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.087 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43137

The years spanned by the FARS economies, which would be associated look different than the oldest vehicles in database cover all model years from with higher per-mile driving costs, they the FARS dataset. calendar year 1996 through 2015. Given also (mostly) faced lower fuel prices. When the per-vehicle fatality rates are that there is a significant number of This adjustment increased the mileage converted into per-mile fatality rates, years between the older vehicles in the accumulation for older vehicles, but not the pattern observed in the registration 1996 CY data and the most recent model by large amounts. Because the CAFE comparison becomes clearer. As Figure years in the odometer data the informed model uses the mileage accumulation the mileage accumulation schedules, schedule and applies it to all vehicles in II–5 shows, the trend present in the staff applied an elasticity of ¥0.20 to the fleet, it is necessary to use the same fatality data on a per-registration basis is the change in the average cost per mile schedule to estimate per-mile fatality even clearer on a per-mile basis: Newer of vehicles over their lives. While the rates in the statistical analysis—even if vintages are safer than older vintages, at older vehicles had lower fuel the schedule is based on vehicles that each age, over time.

The shape of the curve in Figure II– DOT’s statistical model is based on that registered vehicles) on vehicle age with 5 suggests a polynomial relationship structure. fixed effects for the model years present between fatality rate and vehicle age, so The final model is a weighted quartic in the dataset: 323 polynomial regression (by number of

The coefficient estimates and model summary are in Table II–67.

323 Note: The dataset included MY 1975, but that term acts as the fixed effect for 1975 and all others fixed effect is excluded from the set. The constant are relative to that one.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00153 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.088 EP24AU18.089 43138 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00154 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.090 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43139

This function is now embedded in the improvement is almost certain to occur. capability provided by sensing CAFE model, so the combination of The difficulty is for most of these technologies such as RADAR, LIDAR, VMT per vehicle and the distribution of technologies, their effectiveness against video camera, etc. CIB systems mitigate ages and model years present in the on- fatalities and the pace of their adoption crash severity by automatically applying road fleet determine the number of are highly uncertain. Moreover, the vehicle’s brakes shortly before the fatalities in a given calendar year. The autonomous vehicles offer the expected impact (i.e., without requiring model reproduces the observed fatalities possibility of significantly reducing or the driver to apply force to the brake of a given model year, at each age, eventually even eliminating the effect of pedal). reasonably well with more recent model human error in crash causation, a Dynamic Brake Support (DBS) is a years (to which the VMT schedule is a contributing factor in roughly 94% of all technology that actively increases the better match) estimated with smaller crashes. This conservative assumption amount of braking provided to the errors. may cause the NPRM to understate the driver during a rear-end crash avoidance While the final specification was not beneficial effect of proposed standards maneuver. If the driver has applied the only one considered, the fact this on improving (reducing) the number of force to the brake pedal, DBS uses model was intended to live inside the fatalities. forward-looking sensor data provided by CAFE model to dynamically estimate Advanced technologies that are technologies such as RADAR, LIDAR, fatalities for a dynamically changing on- currently deployed or in development video cameras, etc. to assess the road vehicle population was a include: potential for a rear-end crash. Should Forward Collision Warning (FCW) constraining factor. DBS ascertain a crash is likely (i.e., the systems are intended to passively assist sensor data indicate the driver has not (a) Predicting Future Safety Trends the driver in avoiding or mitigating the applied enough braking to avoid the The base model predicts a net impact of rear-end collisions (i.e., a increase in fatalities due primarily to vehicle striking the rear portion of a crash), DBS automatically intervenes. slower adoption of safer vehicles and vehicle traveling in the same direction Although the manner in which DBS has added driving because of less costly directly in front of it). FCW uses been implemented differs among vehicle operating costs. In earlier forward-looking vehicle detection vehicle manufacturers, the objective of calendar years, the improvement in capability, such as RADAR, LIDAR the interventions is largely the same: To safety of the on-road fleet produces a net (laser), camera, etc., to detect other supplement the driver’s commanded reduction in fatalities, but from the mid- vehicles ahead and use the information brake input by increasing the output of 2020s forward, the baseline model from these sensors to warn the driver the foundation brake system. In some predicts no further increase in safety, and to prevent crashes. FCW systems situations, the increased braking and the added risk from more VMT and provide an audible, visual, or haptic provided by DBS may allow the driver older vehicles produces a net increase warning, or any combination thereof, to to avoid a crash. In other cases, DBS in fatalities. This model thus reflects a alert the driver of an FCW-equipped interventions mitigate crash severity. conservative limitation; it implicitly vehicle of a potential collision with Pedestrian AEB (PAEB) systems assumes the trend toward increasingly another vehicle or vehicles in the provide automatic braking for vehicles safe vehicles that has been apparent for anticipated forward pathway of the when pedestrians are in the forward the past 3 decades will flatten in mid- vehicle. path of travel and the driver has taken 2020s. The agency does not assert this Crash Imminent Braking (CIB) insufficient action to avoid an imminent is the most likely case. In fact, the systems are intended to actively assist crash. Like CIB, PAEB safety systems development of advanced crash the driver by mitigating the impact of use information from forward-looking avoidance technologies in recent years rear-end collisions. These safety systems sensors to automatically apply or indicates some level of safety have forward-looking vehicle detection supplement the brakes in certain driving

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00155 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.091 43140 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

situations in which the system FMVSS No. 108 permits a rear turn to existing fatality target populations determines a pedestrian is in imminent signal lamp color of amber or red. and adjusted for current technology danger of being hit by the vehicle. Many Lane Departure Warning (LDW) penetration in the on-road fleet, taking PAEB systems use the same sensors and system is a driver assistance system that into account the retirement of existing technologies used by CIB and DBS. monitors lane markings on the road and vehicles and the pace of future Rear Automatic Braking feature alerts the driver when their vehicle is penetration required to meet statutory means installed vehicle equipment that about to drift beyond a delineated edge compliance requirements, as well as has the ability to sense the presence of line of their current travel lane. adjustments for overlapping target objects behind a reversing vehicle, alert Blind Spot Detection (BSD) systems populations. Based on these factors, as the driver of the presence of the uses digital camera imaging technology well as assumptions regarding future object(s) via auditory and visual alerts, or radar sensor technology to detect one VMT, the study predicted future fatality or more vehicles in either of the and automatically engage the available levels and rates. Because the safety braking system(s) to stop the vehicle. adjacent lanes that may not be apparent impact in the CAFE model Semi-automatic Headlamp Beam to the driver. The system warns the independently predicts future VMT, we Switching device provides either driver of an approaching vehicle’s removed the VMT growth rate from the automatic or manual control of presence to help facilitate safe lane NCSA study and developed a prediction headlamp beam switching at the option changes. of the driver. When the control is These technologies are either under of vehicle fatality trends based only on automatic, headlamps switch from the development or are currently being the penetration pace of new safety upper beam to the lower beam when offered, typically in luxury vehicles, as technologies into the on-road fleet. illuminated by headlamps on an either optional or standard equipment. These data were then normalized into approaching vehicle and switch back to To estimate baseline fatality rates in relative safety factors with CY 2015 as the upper beam when the road ahead is future years, NHTSA examined the baseline (to match the baseline dark. When the control is manual, the predicted results from a previous NCSA fatality year used in this CAFE analysis). driver may obtain either beam manually study 324 that measured the effect of These factors were then converted into regardless of the conditions ahead of the known safety regulations on fatality equivalent fatality rates/100 million vehicle. rates. This study relied on statistical VMT by anchoring them to the 2015 Rear Turn Signal Lamp Color Turn evaluations of the effectiveness of motor fatality rate/100 million VMT published signal lamps are the signaling element vehicle safety technologies based on real by NHTSA. Figure II–6 below illustrates of a turn signal system, which indicates world performance in the on-road the modelling output and projected the intention to turn or change direction vehicle fleet to determine the fatality trend from the analysis of the by giving a flashing light on the side effectiveness of each safety technology. NCSA study, prior to adjustment to toward which the turn will be made. These effectiveness rates were applied fatality rates/100 million VMT.

324 Blincoe, L. & Shankar, U. The Impact of Safety Fatality Rates, National Highway Traffic Safety www.nhtsa.gov/sites/nhtsa.dot.gov/files/ Standards and Behavioral Trends on Motor Vehicle Administration (Jan. 2007), available at https:// documents/810777v3.pdf.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00156 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.092 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43141

This model was based on inputs heavy truck stopping distance 2015 timeframe. In addition, the NCSA representing the impact of technology improvements (FMVSS 121). These study did not attempt to adjust for safety improvement through CY 2020. technologies were estimated to be impacts that may have resulted from Projecting this trend beyond 2020 can installed in only 40–70% of the on-road changes in the vehicle sales mix be justified based on the continued fleet as of CY 2020, implying further (vehicle types and sizes creating transformation of the on-road fleet to safety improvement well beyond the different interactions in crashes), in 100% inclusion of the known safety 2020 calendar year. commuting patterns, or in shopping or technologies. Based on projections in The NCSA study focused on socializing habits associated with the NCSA study, significant further projections to reflect known technology internet access and use. To address this, technology penetration can be expected adaptation requirements, but it was NHTSA also examined the actual in the on-road fleet for side impact conducted prior to the 2008 recession, change in the fatality rate as measured improvements (FMVSSS 214), which disrupted the economy and by fatality counts and VMT estimates. electronic stability control (FMVSS changed travel patterns throughout the 126), upper interior head impact country. Thus, while the relative trends Figure II–7 below illustrates the actual protection (FMVSS 301), tire pressure it predicts seem reasonable, they cannot fatality rates measured from 2000 monitoring systems (FMVSS 138), account for the real-world disruption through 2016 and the modeled fatality ejection mitigation (FMVSS 226), and and recovery that occurred in the 2008– rate trend based on these historical data.

The effect of the recession and vehicle safety rates. Because of the continued improvement well beyond subsequent recovery can be seen in upward shift over the 2014–2015 the 2020 timeframe, which is when the chaotic shift in the fatality rate trend period, this model, which does not historical fatality rate based model starting in 2008. The generally gradual reflect technology trend savings after breaks down. By contrast, the historical decline that had been occurring over the 2015, will predict an upward shift of fatality rate model reflects shifts in previous decade was interrupted by a fatality rates after 2020. safety not captured by the NCSA model, slowdown in the rate of change Predicting future safety trends has but gives arguably implausible results followed by subsequent upward and significant uncertainty. Although after 2020. It essentially represents a downward shifts. More recently, the rate further safety improvements are scenario in which economic, market, or has begun to increase. These shifts expected because of advanced safety behavioral factors minimize or offset reflect some combination of factors not technologies such as automatic braking much of the potential impact of future captured in the NCSA analysis and eventually, fully automated safety technology. mentioned above. The significance of vehicles, the pace of development and For the NPRM, the analysis examines this is that although there was a steady extent of consumer acceptance of these a scenario projecting safety increase in the penetration of safety improvements is uncertain. Thus, two improvements beyond 2015 using a technologies into the on-road fleet imperfect models exist for predicting simple average of the NCSA and between 2008 and 2015, other unknown future safety trends. The NCSA model historical fatality rate models, accepting factors offset their positive influence reflects the expected trend from each as an illustration of different and and eventually reversed the trend in required technologies and indicates conflicting possible future scenarios. As

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00157 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.093 43142 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

both models eventually curve up The results indicate roughly a 19% use, which was below 10% in 1960 and because of their quadratic form, each reduction in fatality rates between 2015 had risen to roughly 70% by 2000, and models’ results are flattened at the point and 2050. This is a slower pace than is now more than 90%. Because belt use where they begin to trend upward. This what has historically occurred over the is now above 90%, further such occurs in 2045 for the NCSA model and past several decades, but the biggest improvements are unlikely unless they in 2021 for the historical model. The influence on historical rates was come from new technologies. results are shown in Figure II–8 below. significant improvement in safety belt

A difficulty with these trend models VMT and the mix of drivers and type of for technologies enabling this is they are based on calendar year driving to an extent we do not believe improvement. In a standard elasticity predictions, which are derived from the the recession era gives an accurate model, sales impacts are a function of full on-road vehicle fleet rather than the picture of the safety trends inherent in the percent change in vehicle price. model year fleet, which is the basis for the vehicles themselves. Therefore, Hypothetically, increasing the base calculations in the CAFE model. As beginning in 2008, NHTSA price for added safety technologies such they are useful primarily as approximated a trend for safety would decrease the impact of higher indicators that vehicle safety has improvement through about MY 2035 to prices due to impacts of CAFE standards steadily improved over the past several reflect the continued effect of improved on vehicle sales. The percentage change decades, and given the advanced safety safety technologies such as advanced in baseline price would decrease, which technologies under current automatic braking, which manufacturers would mean a lower elasticity effect, development, we would expect some have announced will be in all new which would mean a lower impact on continuation of improvement in MY vehicles by MY 2022. The agency sales. NHTSA will consider possible vehicle safety over the near and mid- recognize this is only an estimate, and term future. To account for this, NHTSA actual MY trends could be above or ways to address this issue before the approximated a model year safety trend below this line. NHTSA examined final rule, and we request comments on continuing through about 2035 (Figure alternate trends in a sensitivity analysis the need and/or practicability for such II–9). For this trend the agency used and request comments on the best way an adjustment, as well as any data and actual data from FARS to calculate the to address future safety trends. other relevant information that could change in fatality rates through 2007. NHTSA also notes although we support such an analysis of these costs, The recession, which struck our project vehicles will continue to become as well as the future pace of economy in 2008, distorted normal safer going forward to about 2035, we do technological adoption within the behavioral patterns and affected both not have corresponding cost information vehicle fleet.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00158 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.094 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43143

(b) Adjusting for Behavioral Impacts newer vehicle will not in itself cause vehicles for model years 2001 through The influence of delayed purchases of those drivers to suddenly stop wearing 2008. She developed several statistical new vehicles is estimated to have the seat belts, speed, drive under the regression models that specifically most significant effect on safety influence, or shift driving to different controlled for most behavioral factors to imposed by CAFE standards. Because of land use areas. The goal of this analysis isolate model year vehicle a combination of safety regulations and is to measure the effect of different characteristics. However, her study did voluntary safety improvements, vehicle designs that change by model not specifically report the change in MY passenger vehicles have become safer year. The modelling process for fatality rates—rather, she reported total over time. Compared to prior decades, estimating safety essentially involves fatalities that could have been saved in fatality rates have declined significantly substituting fatality rates of older MY a baseline year (2008) had all vehicles because of technological improvements, vehicles for improved rates that would in the on-road fleet had the same safety as well as behavioral shifts, such as have been experienced with a newer features as the MY 2001 through MY increased seat belt use. As these safer vehicle. Therefore, it is important to 2008 vehicles. This study potentially vehicles replace older less safe vehicles control for behavioral aspects associated provides a basis for comparison with in the fleet, the on-road fleet is replaced with vehicle age so only vehicle design results of the CAFE safety estimates. To with vehicles reflecting the improved differences are reflected in the estimate make this comparison, the CY 2008 fatality rates of newer, safer vehicles. of safety impacts. To address this, the passenger car and light truck fatalities However, fatality rates associated with CAFE safety model was run to control total from FARS were modified by different model year vehicles are for vehicle age. That is, it does not subtracting the values found in Figure influenced by the vehicle itself and by reflect a decision to replace an older II–9 of her study. This gives a stream of driver behavior. Over time, used model year vehicle that is, for example, comparable hypothetical CY 2008 vehicles are purchased by drivers in 10 years old with a new vehicle. Rather, fatality totals under progressively less different demographic circumstances it reflects the difference in the average safe model year designs. Results who also tend to have different fatality rate of each model year across its indicated that had the 2008 on-road behavioral characteristics. Drivers of entire lifespan. This will account for fleet been equipped with MY 2008 older vehicles, on average, tend to have most of the difference because of vehicle safety equipment and vehicle lower belt use rates, are more likely to age, but it may still reflect a bias caused characteristics, total fatalities would drive inebriated, and are more likely to by the upward trend in societal seat belt have been reduced by 25% compared to drive over the speed limit. Additionally, use over time. Because of this secular vehicles that were actually on the road older vehicles are more likely to be trend, each subsequent model year’s in 2008. Similar results were calculated driven on rural roadways, which useful life will occur under increasingly for each model years’ vehicle typically have higher speeds and higher average seat belt use rates. This characteristics back to 2001. produce more serious crashes. These could cause some level of behavioral For comparison, predicted MY fatality relationships are illustrated graphically safety improvement to be ascribed to the rates were derived from the CAFE safety in Chapter 11 of the PRIA model year instead of the driver cohort. model and applied to the CY 2008 VMT accompanying this proposed rule. However, it is difficult to separate this calculated by that model. This gives an The behavior being modelled and effect from the belt use impacts of estimate of CY 2008 fatalities under ascribed to CAFE involves decisions by changing driver cohorts as vehicles age. each model years’ fatality rate, which, drivers who are contemplating buying a Glassbrenner (2012) analyzed the when compared to the predicted CY new vehicle, and the purchase of a effect of improved safety in newer fatality total, gives a trendline

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00159 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.095 43144 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

comparable to the Glassbrenner (instead of 12.15) and produces the through 2006, but in the last few years trendline illustrating the change in MY result shown in Figure II–10. The CAFE change decreases. This might reflect the fatality rates. Both models are sensitive model trendline shifts up, which trend in societal belt use, which rose to the initial 2008 baseline fatality total, narrows the difference in early years but steadily through 2005 and levelled off. and because the predicted CAFE total is expands it in later years. However, VMT Later model years’ fatality rates would somewhat lower than the actual total, and fatalities are linked in the CAFE benefit from this trend while earlier the agency ran a third trendline to model, so the actual level of the MY model years would suffer. This seems examine the influence of this difference. safety predicted by the CAFE curve has consistent with our using lifetime MY Results are shown in Figure II–10. uncertainty. Perhaps the most fatality rates to reflect MY change rather Using the corrected fatality count, but meaningful result from this comparison than first year MY fatality rates retaining the predicted VMT changes is the difference in slopes; the CAFE (although even first year rates would the initial 2018 CY fatality rate to 12.62 model predicts more rapid change reflect this bias, but not as much).

To provide another perspective on that time span. Kahane is currently which isolates vehicle design from the safety impacts, NHTSA accessed data working under contract for NHTSA to primary behavioral impact—seat belt from a comprehensive study of the update this study through 2016. At usage. We normalized Kahane’s index to effects of safety technologies on motor NHTSA’s request, Kahane accessed his MY 1975 and did the same to the ‘‘fixed vehicle fatalities. Kahane (2015) 325 database to provide a measure of effects’’ we are currently using from our examined all safety effects of vehicle relative MY vehicle design safety by safety model to compare the trends in safety technologies from 1960 through controlling for seat belt use. The result MY safety from the two methods. 2012 and found these technologies was a MY safety index illustrating the Results are shown in Figure II–11. saved more than 600,000 lives during progress in vehicle safety by model year

325 Kahane, C.J. Lives Saved by Safety Standards 26 FMVSS and the Effectiveness of their Associated Administration (Jan. 2015), available at https:// and Associated Vehicle Safety Technologies, 1960– Safety Technologies in Reducing Fatalities, Injuries, crashstats.nhtsa.dot.gov/Api/Public/View 2012—Passenger Cars and LTVs—with Reviews of and Crashes, National Highway Traffic Safety Publication/812069.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00160 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.096 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43145

From Figure II–11 both approaches demographics driving older vehicles, contributing to uncertainty, and given show similar long-term downward there is a secular trend toward more belt this, their differences are not surprising. trends, but this model shows a steeper use reflecting the increase in societal It is encouraging they show similar slope than Kahane’s model. The two awareness of belt use importance over directional trends, reinforcing the basic models involve completely different time. This trend is illustrated in Figure concept we are measuring. NHTSA approaches, so some difference is to be II–12 below.326 NHTSA’s current recognizes predicting future fatality expected. However, it is also possible approach removes the age trend in belt impacts, as well as sales impacts that this reflects different methods used to use, but it’s not clear whether it cause them, is a difficult and imprecise isolate vehicle design safety from accounts for the full impacts of the task. NHTSA will continue to behavioral impacts. As discussed secular trend as well. If not, some investigate this issue, and we seek previously, NHTSA addressed this issue portion of the gap between the two comment on these estimates as well as by removing vehicle age impacts from trendlines could reflect behavioral alternate methods for predicting the its model, whereas Kahane’s model does impacts rather than vehicle design. it by controlling for belt use. As noted These models (NHTSA, Glassbrenner, safety effects associated with delayed previously, aside from the age impact on and Kahane) involve differing new vehicle purchases. belt use associated with the different approaches and assumptions

326 Note: The drop occurring in 1994 reflects a National Occupant Protection Survey (NOPUS). surveys produced inflated results, especially in the shift in the basis for determining belt use rates. Prior to this, a conglomeration of state studies 1991–1993 period. Effective in 1994, data were reported from the provided the basis. It is likely the pre-NOPUS

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00161 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.097 43146 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

4. Impact of Rebound Effect on Fatalities in 2016. This gives a societal value of and alternative methods to estimate Based on historical data, it is possible $9.9 million for each fatality. The CAFE injury impacts. to calculate a baseline fatality rate for safety model estimates effects on traffic The adjustment factor is derived from vehicles of any model year vintage. By fatalities but does not address Tables 1–8 and I–3 in Blincoe et al. simply taking the total number of corresponding effects on non-fatal (2015). Incidence in Table I–3 reflects vehicles involved in fatal accidents over injuries and property damage that the Abbreviated Injury Scale (AIS), all ages for a model year and dividing would result from the same factors which ranks nonfatal injury severity by the cumulative VMT over the useful influencing fatalities. To address this, based on an ascending 5 level scale with life of every vehicle produced in that we developed an adjustment factor that the most severe injuries ranked as level would account for these crashes. model year, one arrives at a baseline 5. More information on the basis for Development of this factor is based on hazard rate denominated in fatalities per these classifications is available from billion miles. The fatalities associated the assumption nonfatal crashes will be affected by CAFE standards in the Association for the Advancement of with vehicles produced in that model Automotive Medicine at https:// year are then proportional to the proportion to their nationwide incidence and severity. That is, NHTSA www.aaam.org/abbreviated-injury-scale- cumulative lifetime VMT, where total ais/. fatalities equal the product of the assumes the same injury profile, the baseline hazard rate and VMT. A more relative number of cases of each injury Table 1–3 in Blincoe lists injured comprehensive discussion of the severity level, that occur nationwide, persons with their highest (maximum) rebound effect and the basis for will be increased or decreased because injury determining the AIS level calculating its impact on mileage and of CAFE. The agency recognizes this (MAIS). This scale is represented in risk is in Chapter 8 of the PRIA may not be the case, but the agency does terms of MAIS level, or maximum accompanying this proposed rule. not have data to support individual abbreviated injury scale. MAIS0 refers estimates across injury severities. There to uninjured occupants in injury 5. Adjustment for Non-Fatal Crashes are reasons why this may not be true. vehicles, MAIS1 are generally Fatalities estimated to be caused by For example, because older model year considered minor injuries, MAIS2 various alternative CAFE standards are vehicles are generally less safe than moderate injuries, MAIS3 serious valued as a societal cost within the newer vehicles, fatalities may make up injuries, MAIS4 severe injuries, and CAFE models’ cost/benefit accounting. a larger portion of the total injury MAIS5 critical injuries. PDO refers to Their value is based on the picture than they do for newer vehicles. property damage only crashes, and comprehensive value of a fatality This would imply lower ratios across counts for PDOs refer to vehicles in derived from data in Blincoe et al. the non-fatal injury and PDO profile and which no one was injured. From Table (2015), adjusted to 2016 economics and would imply our adjustment may II–68, ratios of injury incidence/fatality updated to reflect the official DOT overstate total societal impacts. NHTSA are derived for each injury severity level guidance on the value of a statistical life requests comments on this assumption as follows:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00162 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.098 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43147

For each fatality that occurs vehicles are worth less and will cost less 2016. There is a minor timing nationwide in traffic crashes, there are to repair, if they are repaired at all. The discrepancy in these data because the 561 vehicles involved in PDOs, 139 consumer’s property damage loss is thus new vehicle data represent January uninjured occupants in injury vehicles, reduced by longer retention of these 2017, and the used vehicle price is for 105 minor injuries, 10 moderate vehicles. To estimate this loss, average the average over 2016. NHTSA was injuries, 3 serious injuries, and new and used vehicle prices were unable to locate exact matching data at fractional numbers of the most serious compared. New vehicle transaction this time, but the agency believes the categories which include severe and prices were estimated from a study difference will be minor. critical nonfatal injuries. For each published by Kelley Blue Book.327 Based on these data, new vehicles are fatality ascribed to CAFE it is assumed Based on these data, the average new on average worth 82% more than used there will be nonfatal crashes in these vehicle transaction price in January vehicles. To estimate the effect of higher same ratios. 2017 was $34,968. Used vehicle property damage costs for newer Property damage costs associated with transaction prices were obtained from vehicles on crashes, the per unit delayed new vehicle purchases must be Edmonds Used Vehicle Market Report property damage costs from Table I–9 in treated differently because crashes that published in February of 2017.328 Blincoe et al. (2015) were multiplied by subsequently occur damage older used Edmonds data indicate the average used this factor. Results are illustrated in vehicles instead of newer vehicles. Used vehicle transaction price was $19,189 in Table II–69.

The total property damage cost F = change in fatalities estimated for CAFE reduction was then calculated as a due to retaining used vehicles function of the number of fatalities Where: r = ratio of nonfatal injuries or PDO vehicles reduced or increased by CAFE as S = total property damage savings from to fatalities (F) retaining used vehicles longer p = value of property damage prevented by follows: retaining older vehicle

327 Press Release, Kelley Blue Book, New-Car mediaroom.kbb.com/2017-02-01-New-Car- 328 Edmunds Used Vehicle Market Report, Transaction Prices Remain High, Up More Than 3 Transaction-Prices-Remain-High-Up-More-Than-3- Edmunds (Feb. 2017), https:// Percent Year-Over-Year in January 2017, According Percent-Year-Over-Year-In-January-2017- dealers.edmunds.com/static/assets/articles/2017_ to Kelley Blue Book (Feb. 1, 2017), https:// According-To-Kelley-Blue-Book. Feb_Used_Market_Report.pdf.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00163 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.099 EP24AU18.100 EP24AU18.145 43148 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

n = the 8 injury severity categories calculation was more direct, a simple contribution to total fatalities of The number of fatalities ascribed to application of the ratio of the portion of different factors. As noted, although a CAFE because of older vehicle retention costs produced by fatalities. In this case, safety impact from the rebound effect is was multiplied by the unit cost per there is no need to adjust for property calculated, these impacts are considered fatality from Table I–9 in Blincoe et al. damage because all impacts were to be freely chosen rather than imposed (2015) to determine the societal impact derived from the mix of vehicles in the by CAFE and imply personal benefits at accounted for by these fatalities.329 on-road fleet. Again, from Table I–8 in least equal to the sum of their added From Table I–8 in Blincoe et al. (2015), Blincoe et al (2015), we derive this ratio costs and safety consequences. The NHTSA subtracted property damage based on all cost factors including impacts of this nonfatal crash costs from all injury severity levels and property damage to be .36. These adjustment affect costs and benefits recalculated the total comprehensive calculations are summarized as follows: equally. When considering safety value of societal losses from crashes. impacts actually imposed by CAFE The agency then divided the portion of standards, only those from mass these crashes because of fatalities by the changes and vehicle purchase delays are resulting total to estimate the portion of Where: considered. NHTSA has two different crashes excluding property damage that SV = Value of societal Impacts of all crashes factors depending on which metric is are accounted for by fatalities. Results F = change in fatalities estimated for CAFE indicate fatalities accounted for due to retaining used vehicles considered. The agency created these v = Comprehensive societal value of factors by weighting components by the approximately 40% of all societal costs preventing 1 fatality exclusive of property damage. NHTSA relative contribution to changes in x = Percent of total societal loss from crashes fatalities associated with each then divided the total cost of the added excluding property damage accounted fatalities by 0.4 to estimate the total cost for by fatalities component. This process and results are of all crashes prevented exclusive of the S = total property damage savings from shown in Table II–70. Note: For the savings in property damage. After retaining used vehicles longer NPRM, NHTSA applied the average subtracting the total savings in property M = change in fatalities due to changes in weighted factor to all fatalities. This will damage from this value, we divided the vehicle mass to meet CAFE standards tend to slightly overstate costs because fatality cost by it to estimate that c = Percent of total societal loss from all cost of sales and scrappage and understate overall, fatalities account for 43% of the factors in all crashes accounted for by fatalities costs associated with mass and rebound. total costs that would result from older The agency will consider ways to adjust vehicle retention. For purposes of application in the this minor discrepancy for the final rule. For the fatalities that occur because of CAFE model, these two factors were mass effects or to the rebound effect, the combined based on the relative

Table II–71, Table II–72, Table II–73, 7% discount rates. As noted in Section undiscounted; only the monetized and Table II–74 summarize the safety II.F.5, societal impacts are valued using societal impact is discounted. effects of CAFE standards across the a $9.9 million value per statistical life various alternatives under the 3% and (VSL). Fatalities in these tables are

329 Note: These calculations used the original adjusting for economics. These calculations produce ratios and are thus not sensitive to values in the Blincoe et all (2015) tables without adjustments for inflation.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00164 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.101 EP24AU18.102 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43149

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00165 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.103 43150 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00166 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.104 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43151

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00167 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.105 43152 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00168 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.106 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43153

Table II–75 through Table II–78 noted in Section II.F.5, societal impacts monetized societal impact is summarize the safety effects of GHG are valued using a $9.9 million value discounted. standards across the various alternatives per statistical life (VSL). Fatalities in under the 3% and 7% discount rates. As these tables are undiscounted; only the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00169 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.107 43154 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00170 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.108 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43155

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00171 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.109 43156 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00172 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.110 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43157

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00173 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.111 43158 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

While NHTSA notes the value of These costs implicitly involve a cost G. How the Model Analyzes Different rebound effect fatalities, as well as total and a benefit that are offsetting. The Potential CAFE and CO2 Standards fatalities from all causes, the agency relevant safety impacts attributable to 1. Specification of No-Action and Other does not add rebound effects to the CAFE are highlighted in bold in the Regulatory Alternatives other CAFE-related impacts because above tables. rebound-related fatalities and injuries (a) Mathematical Functions Defining result from risk that is freely chosen and Passenger Car and Light Trucks offset by societal valuations that at a Standards for Each Model Year During minimum exceed the aggregate value of 2016–2032 safety consequences plus added vehicle In the U.S. market, the stringency of operating and maintenance costs.330 CAFE and CO2 standards can influence the design of new vehicles offered for 330 It would also include some level of consumer surplus, which we have estimated using the sale by requiring manufacturers to standard triangular function. This is discussed in produce increasingly fuel efficient Chapter 8.5.1 of the PRIA. vehicles in order to meet program

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00174 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.112 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43159

requirements. This is also true in the (b) Off-Cycle and A/C Efficiency these forward to future years with a few CAFE model simulation, where the Adjustments Anticipated for Each exceptions. Several of the technologies standards can be defined with a great Model Year described in Section II.D are associated with A/C efficiency and off-cycle FCIVs. deal of flexibility to examine the impact Another aspect of credit accounting is In particular, stop-start systems, of different program specifications on partially implemented in the CAFE integrated starter generators, and full the auto industry. Standards are defined model at this point—those related to the hybrids are assumed to generate off- for each model year and can represent application of off-cycle and A/C cycle adjustments when applied to different slopes that relate fuel economy efficiency adjustments, which vehicles to improve their fuel economy. to footprint, different regions of flat manufacturers earn by taking actions Similarly, higher levels of aerodynamic slopes, and different rates of increase for such as special window glazing or using improvements are assumed to include each of three regulatory classes covered reflective paints that provide fuel active grille shutters on the vehicle, by the CAFE program (domestic economy improvements in real-world which also qualify for off-cycle FCIVs. passenger cars, imported passenger cars, operation but do not produce The analysis assumes that any off- and light trucks). measurable improvements in fuel cycle FCIVs that are associated with consumption on the 2-cycle test. The CAFE model takes, as inputs, the actions outside of the technologies NHTSA’s inclusion of off-cycle and coefficients of the mathematical discussed in Section II.D (either chosen A/C efficiency adjustments began in MY functions described in Sections III and from the pre-approved ‘‘pick list,’’ or 2017, while EPA has collected several IV. It uses these coefficients and the granted in response to individual years’ worth of submissions from function to which they belong to define manufacturer petitions) remain at the manufacturers about off-cycle and A/C the target for each vehicle in the fleet, levels claimed by manufacturers in MY efficiency technology deployment. then computes the standard using the 2017. Any additional A/C efficiency and Currently, the level of deployment can off-cycle adjustments that accrue as the harmonic average of the targets for each vary considerably by manufacturer with result of explicit technology application manufacturer and fleet. The model also several claiming extensive Fuel are calculated dynamically in each allows the user to define the extent and Consumption Improvement Values model year for each alternative. The off- duration of various compliance (FCIV) for off-cycle and A/C efficiency cycle FCIVs for each manufacturer and flexibilities (e.g., limits on the amount technologies and others almost none. fleet, denominated in grams CO2 per of credit that a manufacturer may claim The analysis of alternatives presented mile,331 are provided in Table II–79. related to air conditioning efficiency here does not attempt to project how improvements or off-cycle fuel economy future off-cycle and A/C efficiency 331 For the purpose of estimating their adjustments) as well as limits on the technology use will evolve or speculate contribution to CAFE compliance, the grams CO2/ number of years that CAFE credits may about the potential proliferation of FCIV mile values in Table II–79 are converted to gallons/ be carried forward or the amount that proposals submitted to the agencies. mile and applied to a manufacturer’s 2-cycle CAFE performance. When calculating compliance with may be transferred between a Rather, this analysis uses the off-cycle EPA’s GHG program, there is no conversion manufacturer’s fleets. credits submitted by each manufacturer necessary (as standards are also denominated in for MY 2017 compliance and carries grams/mile).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00175 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43160 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The model currently accounts for any consistently achieved fuel economy penalties may be a transfer from one off-cycle adjustments associated with levels below their standard. As in OEM to another. technologies that are included in the set previous versions of the CAFE model, While the Energy Policy and of fuel-saving technologies explicitly the current version allows the user to Conservation Act (EPCA), as amended simulated as part of this proposal (for specify inputs identifying such in 2007 by the Energy Independence example, start-stop systems that reduce manufacturers and to consider their and Security Act, prescribes these fuel consumption during idle or active compliance decisions as if they are specific civil penalty provisions for grille shutters that improve willing to pay civil penalties for non- CAFE standards, the Clean Air Act aerodynamic drag at highway speeds) compliance with the CAFE program. (CAA) does not contain similar and accumulates these adjustments up The assumed civil penalty rate in the provisions. Rather, the CAA’s to the 10 g/mi cap. As a practical matter, current analysis is $5.50 per 1/10 of a provisions regarding noncompliance most of the adjustments for which mile per gallon, per vehicle sold. constitute a de facto prohibition against manufacturers are claiming off-cycle selling vehicles failing to comply with FCIV exist outside of the technology It is worth noting that treating a emissions standards. Therefore, inputs tree, so the cap is rarely reached during manufacturer as if they are willing to regarding civil penalties—including compliance simulation. If those FCIVs pay civil penalties does not necessarily inputs regarding manufacturers’ become a more important compliance mean that it is expected to pay penalties potential willingness to treat civil mechanism, it may be necessary to in reality. It merely implies that the penalty payment as an economic model their application explicitly. manufacturer will only apply fuel choice—apply only to simulation of However, doing so will require data on economy technology up to a point, and CAFE standards. which vehicle models already possess then stop, regardless of whether or not these improvements as well as the cost its corporate average fuel economy is (d) Treatment of Credit Provisions for and expected value of applying them to above its standard. In practice, we ‘‘Standard Setting’’ and other models in the future. Comment is expect that many of these manufacturers ‘‘Unconstrained’’ Analyses sought on both the data requirements will continue to be active in the credit NHTSA may not consider the and strategic decisions associated with market, using trades with other application of CAFE credits toward manufacturers’ use of A/C efficiency manufacturers to transfer credits into compliance with new standards when and off-cycle technologies to improve specific fleets that are challenged in any establishing the standards CAFE and CO2 compliance. given year, rather than paying penalties themselves.332 As such, this analysis (c) Civil Penalty Rate and OEMs’ to resolve CAFE deficits. The CAFE considers 2020 to be the last model year Anticipated Willingness To Treat Civil model calculates the amount of in which carried-forward or transferred Penalties as a Program Flexibility penalties paid by each manufacturer, credits can be applied for the CAFE but it does not simulate trades between program. Beginning in model year 2021, Throughout the history of the CAFE manufacturers. In practice, some program, some manufacturers have (possibly most) of the total estimated 332 49 U.S.C. 32902(h) (2007).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00176 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.113 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43161

today’s ‘‘standard setting’’ analysis is model now carries a full representation (3) For manufacturers assumed to be conducted assuming each fleet must of the production multipliers related to willing to pay civil penalties (in the CAFE comply with the CAFE standard electric vehicles, fuel cell vehicles, program), the manufacturer reaches the point separately in every model year. at which doing so would be more cost- plug-in hybrids, and CNG vehicles, all effective (from the manufacturer’s The ‘‘unconstrained’’ perspective of which vary by year through MY 2021. perspective) than adding further technology. acknowledges that these flexibilities exist as part of the program and, while 2. Simulation of Manufacturers’ [and The model accounts explicitly for not considered in NHTSA’s decision of Buyers’] Potential Responses to Each each model year, applying technologies the preferred alternative, are important Alternative when vehicles are scheduled to be to consider when attempting to estimate The CAFE model provides a way of redesigned or freshened and carrying the real impact of any alternative. Under estimating how manufacturers could forward technologies between model the ‘‘unconstrained’’ perspective, credits attempt to comply with a given CAFE years once they are applied (until, if may be earned, transferred, and applied standard by adding technology to fleets applicable, they are superseded by other to deficits in the CAFE program that the agencies anticipate they will technologies). The model then uses throughout the full range of model years produce in future model years. This these simulated manufacturer fleets to in the analysis. The Draft Environmental exercise constitutes a simulation of generate both a representation of the Impact Analysis (DEIS) accompanying manufacturers’ decisions regarding U.S. auto industry and to modify a representation of the entire light-duty today’s NPRM presents results of compliance with CAFE or CO2 ‘‘unconstrained’’ modeling. Also, standards. registered vehicle population. From because the CAA provides no direction This compliance simulation begins these fleets, the model estimates regarding consideration of any CO2 with the following inputs: (a) The changes in physical quantities (gallons credit provisions, today’s analysis analysis fleet of vehicles from model of fuel, pollutant emissions, traffic includes simulation of carried-forward year 2016 discussed above in Section fatalities, etc.) and calculates the relative costs and benefits of regulatory and transferred CO2 credits in all model II.B, (b) fuel economy improving years. technology estimates discussed above in alternatives under consideration. The CAFE model accounts explicitly (e) Treatment of AFVs for ‘‘Standard Section II.D, (c) economic inputs discussed above in Section II.E, and (d) for each model year, in turn, because Setting’’ and ‘‘Unconstrained’’ Analyses manufacturers actually ‘‘carry forward’’ inputs defining baseline and potential most technologies between model years, NHTSA is also prohibited from new CAFE standards. For each tending to concentrate the application of considering the possibility that a manufacturer, the model applies new technology to vehicle redesigns or manufacturer might produce technologies in both a logical sequence mid-cycle ‘‘freshenings,’’ and design alternatively fueled vehicles as a and a cost-minimizing strategy in order 333 cycles vary widely among compliance mechanism, taking to identify a set of technologies the manufacturers and specific products. advantage of credit provisions related to manufacturer could apply in response to Comments by manufacturers and model AFVs that significantly increase their new CAFE or CO standards. The model fuel economy for CAFE compliance 2 peer reviewers strongly support explicit applies technologies to each of the year-by-year simulation. Year-by-year purposes. Under the ‘‘standard setting’’ projected individual vehicles in a perspective, these technologies (pure accounting also enables accounting for manufacturer’s fleet, considering the credit banking (i.e., carry-forward), as battery electric vehicles and fuel cell combined effect of regulatory and 334 discussed above, and at least four vehicles ) are not available in the market incentives while attempting to compliance simulation to improve fuel environmental organizations recently account for manufacturers’ production submitted comments urging the economy. Under the ‘‘unconstrained’’ constraints. Depending on how the perspective, such as is documented in agencies to consider such credits, citing model is exercised, it will apply the DEIS, the CAFE model considers NHTSA’s 2016 results showing impacts technology until one of the following these technologies in the context of all of carried-forward credits.337 Moreover, occurs: other available technologies and may EPCA/EISA requires that NHTSA make apply them if they represent cost- (1) The manufacturer’s fleet achieves a year-by-year determination of the effective compliance pathways. compliance 335 with the applicable standard appropriate level of stringency and then However, under both perspectives, the and continuing to add technology in the set the standard at that level, while analysis continues to include dedicated current model year would be attractive ensuring ratable increases in average neither in terms of stand-alone (i.e., absent fuel economy through MY 2020. The AFVs that already exist in the MY 2016 regulatory need) cost-effectiveness nor in fleet (and their projected future terms of facilitating compliance in future multi-year planning capability, volumes) in CAFE calculations. Also, model years; (optional) simulation of ‘‘market-driven because the CAA provides no direction (2) The manufacturer ‘‘exhausts’’ available overcompliance,’’ and EPCA credit regarding consideration of alternative technologies; 336 or mechanisms (again, for purposes of fuels, today’s analysis includes modeling the CAFE program) increase simulation of the potential that some 335 When determining whether compliance has the model’s ability to simulate manufacturers might introduce new been achieved in the CAFE program, existing CAFE manufacturers’ real-world behavior, credits that may be carried over from prior model AFVs in response to CO2 standards. To years or transferred between fleets are also used to accounting for the fact that fully represent the compliance benefit determine compliance status. For purposes of from such a response, NHTSA modified determining the effect of maximum feasible CAFE to be redesigned, (b) vehicles have moved to the the CAFE model to include the specific standards, NHTSA cannot consider these ends of each (relevant) technology pathway, after mechanisms for years being considered (though which no additional options exist, or (c) provisions related to AFVs under the does so for model years that are already final) and engineering aspects of available vehicles make CO2 standards. In particular, the CAFE exercises the CAFE model without enabling these available technology inapplicable (e.g., secondary options. axle disconnect cannot be applied to two-wheel 333 Id. 336 In a given model year, it is possible that drive vehicles). 334 Dedicated compressed natural gas (CNG) production constraints cause a manufacturer to 337 Comment by Environmental Law & Policy vehicles should also be excluded in this perspective ‘‘run out’’ of available technology before achieving Center, Natural Resources Defense Council (NRDC), but are not considered as a compliance strategy compliance with standards. This can occur when: Public Citizen, and Sierra Club, Docket ID EPA– under any perspective in this analysis. (a) An insufficient volume of vehicles are expected HQ–OAR–2015–0827–9826, at 28–29.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00177 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43162 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

manufacturers will seek out compliance Camry) allows the CAFE model to another compliance option in years with paths for several model years at a time, explicitly account for the fact that few planned redesigns. Finally, it while accommodating the year-by-year individual vehicle models undergo should be noted that today’s analysis requirement. This same multi-year significant redesigns relatively does not account for future new planning structure is used to simulate infrequently. Many popular vehicle products (or discontinued products)— responses to standards defined in grams models are only redesigned every six past trends suggest that some years in CO2/mile, and utilizing the set of years or so, with some larger/legacy which an OEM had few redesigns may specific credit provisions defined under platforms (the old Ford Econoline Vans, have been years when that OEM EPA’s program. for example) stretching more than a introduced significant new products. decade between significant redesigns. Such changes in product offerings can (a) Representation of Manufacturers’ Engines, which are often shared among obviously be important to Production Constraints many different models and platforms for manufacturers’ compliance positions After the light-duty rulemaking a single manufacturer, can last even but cannot be systematically and analysis accompanying the 2012 final longer—eight to ten years in most cases. transparently accounted for with a fleet rule that finalized NHTSA’s standards While these characterizations of forecast extrapolated forward 10 or more through MY 2021, NHTSA began work product cadence are important to any years from a largely-known fleet. While on changes to the CAFE model with the evaluation of the impacts of CAFE or manufacturers’ actual plans reflect intention of better reflecting constraints CO2 standards, they are not known with intentions to discontinue some products of product planning and cadence for certainty—even by the manufacturers and introduce others, those plans are which previous analyses did not themselves over time horizons as long considered CBI. Further research would account. as those covered by this analysis. be required in order to determine (b) Product Cadence However, lack of certainty about whether and, if so, how it would be redesign schedules is not license to practicable to simulate such decisions, Past comments on the CAFE model ignore them. Indeed, when especially without relying on CBI. have stressed the importance of product manufacturers meet with the agencies to Additionally, each technology cadence—i.e., the development and discuss manufacturers’ plans vis-a`-vis considered for application by the CAFE periodic redesign and freshening of CAFE and CO2 requirements, model is assigned to either a ‘‘refresh’’ vehicles—in terms of involving manufacturers typically present specific or ‘‘redesign’’ cadence that dictates technical, financial, and other practical and detailed year-by-year information when it can be applied to a vehicle. constraints on applying new that explicitly accounts for anticipated Technologies that are assigned to technologies, and DOT has steadily redesigns. Such year-by-year analysis is ‘‘refresh/redesign’’ can be applied at made changes to both the CAFE model also essential to manufacturers’ plans to either a refresh or redesign, while and its inputs with a view toward make use of provisions (for CAFE, technologies that are assigned to accounting for these considerations. For statutory and specific) allowing credits ‘‘redesign’’ can only be applied during example, early versions of the model to be carried forward to future model a significant vehicle redesign. Table II– added explicit ‘‘carrying forward’’ of years, carried back from future model 80 and Table II–81 show the applied technologies between model years, transferred between regulated technologies available to manufacturers years, subsequent versions applied fleets, and traded with other in the compliance simulation, the level assumptions that most technologies will manufacturers. Manufacturers are never at which they are applied (described in be applied when vehicles are freshened certain about future plans, but they greater detail in the CAFE model or redesigned, and more recent versions spend considerable effort developing, documentation), whether they are applied assumptions that manufacturers continually adjusting, and available outside of a vehicle redesign, would sometimes apply technology implementing them. and a short description of each. A brief earlier than ‘‘necessary’’ in order to For every model that appears in the examination of the tables shows that facilitate compliance with standards in MY 2016 analysis fleet, the model years most technologies are only assumed to ensuing model years. Thus, for example, have been estimated in which future be available during a vehicle redesign— if a manufacturer is expected to redesign redesigns (and less significant and nearly all engine improvements are many of its products in model years ‘‘freshenings,’’ which offer assumed to be available only during 2018 and 2023, and the standard’s manufacturers the opportunity to make redesign. In a departure from past CAFE stringency increases significantly in less significant changes to models) will analyses, all transmission improvements model year 2021, the CAFE model will occur. These appear in the market data are assumed to be available during estimate the potential that the file for each model variant. Mid-cycle refresh as well as redesign. While there manufacturer will add more technology freshenings provide additional are past and recent examples of mid- than necessary for compliance in MY opportunities to add some technologies cycle product changes, it seems 2018, in order to carry those product in years where smaller shares of a reasonable to expect that manufacturers changes forward through the next manufacturer’s portfolio is scheduled to will tend to attempt to keep engineering redesign and contribute to compliance be redesigned. In addition, the analysis and other costs down by applying most with the MY 2021 standard. This accounts for multiyear planning—that major changes mainly during vehicle explicit simulation of multiyear is, the potential that manufacturers may redesigns and some mostly modest planning plays an important role in apply ‘‘extra’’ technology in an early changes during product freshenings. As determining year-by-year analytical model year with many planned mentioned below, comment is sought on results. redesigns in order to carry technology the approach to account for product As in previous iterations of CAFE forward to facilitate compliance in a cadence. rulemaking analysis, the simulation of later model year with fewer planned compliance actions that manufacturers redesigns. Further, the analysis accounts (c) Component Sharing and Inheritance might take is constrained by the pace at for the potential that manufacturers (Engines, Transmissions, and Platforms) which new technologies can be applied could earn CAFE and/or CO2 credits in In practice, manufacturers are limited in the new vehicle market. Operating at some model years and use those credits in the number of engines and the Make/Model level (e.g., Toyota in later model years, thereby providing transmissions that they produce.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00178 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43163

Typically, a manufacturer produces a modifies the engine on the ‘‘engine possible data sources and approaches number of engines—perhaps six or eight leader’’ (or ‘‘transmission leader’’), the that could be used to represent any engines for a large manufacturer—and changes to that engine propagate additional costs associated with phased tunes them for slight variants in output through to the other vehicles that share introduction of new engines or for a variety of car and truck that engine (or transmission) in transmissions. applications. Manufacturers limit subsequent years as those vehicles are There are some logical consequences complexity in their engine portfolio for redesigned. The CAFE model has been of this approach, the first of which is much the same reason as they limit modified to provide additional that forcing engine and transmission complexity in vehicle variants: They flexibility vis-a`-vis product cadence. In changes to propagate through to other face engineering manpower limitations, a recent public comment, NRDC noted: vehicles in this way effectively dictates and supplier, production, and service EPA and NHTSA currently constrain their the pace at which new technology can costs that scale with the number of parts model to apply significant fuel-efficient be applied and limits the total number produced. technologies mainly during a product- of unique engines that the model In previous analyses that used the redesign as opposed to product-refresh (or simulates. In the past, NHTSA used CAFE model (with the exception of the mid-cycle). This was identified as one of the ‘‘phase-in caps’’ (see discussion below) 2016 Draft TAR), engines and most sensitive assumptions affecting overall to limit the amount of technology that transmissions in individual vehicle program costs by NHTSA in the TAR. By can be applied to any vehicle in a given models were allowed relative freedom constraining the model, the agencies have year. However, by explicitly tying the likely under-estimated the ability of auto in technology application, potentially manufacturers to incorporate some engine changes to a specific vehicle’s leading to solutions that would, if technologies during their product refreshes. product cadence, rather than letting the followed, create many more unique This is particularly true regarding the critical timing of changes vary across all the engines and transmissions than exist in powertrain technologies which are vehicles that share an engine, the model the analysis fleet (or in the market) for undergoing continuous improvement. The ensures that an engine is only changed a given model year. This multiplicity agency should account for these trends and when its leader is redesigned (at most). likely failed to sufficiently account for incorporate greater flexibility for Given that most vehicle redesign cycles costs associated with such increased automakers—within their models—to are five to eight years, this approach still incorporate more mid-cycle complexity in the product portfolio and enhancements.339 represents shorter average lives than may have represented an unrealistic most engines in the market, which tend diffusion of products for manufacturers While engine redesigns are only to be in production for eight to ten years that are consolidating global production applied to the engine leader when it is or more. It is also the case that vehicles to increasingly smaller numbers of redesigned in the model, followers may which share an engine in the analysis shared engines and platforms.338 The now inherit upgraded engines (that they fleet (MY 2016, for this analysis) are lack of a constraint in this area allowed share with the leader) at either refresh assumed to share that same engine the model to apply different levels of or redesign. All transmission changes, throughout the analysis—unless one or technology to the engine in each vehicle whether upgrades to the ‘‘leader’’ or both of them are converted to power- in which it was present at the time that inheritance to ‘‘followers’’ can occur at split hybrids (or farther) on the vehicle was redesigned or refreshed, refresh as well as redesign. This electrification path. In the market, this independent of what was done to other provides additional opportunities for is not true—since a manufacturer will vehicles using a previously identical technology diffusion within choose an engine from among the engine. manufacturers’ product portfolios. engines it produces to fulfill the One peer reviewer of the CAFE model While ‘‘follower’’ vehicles are efficiency and power demands of a recently commented, ‘‘The integration awaiting redesign (or, for transmissions, vehicle model upon redesign. That of inheritance and sharing of engines, refreshing as applicable), they carry a engine need not be from the same family transmissions, and platforms across a legacy version of the shared engine or of engines as the prior version of that manufacturer’s light duty fleet and transmission. As one peer reviewer vehicle. This is a simplifying separately across its light duty truck recently stated, ‘‘Most of the time a assumption in the model. While the fleet is standard practice within the manufacturer will convert only a single model already accommodates detailed industry.’’ In the current version of the plant within a model year. Thus both inputs regarding redesign schedules for CAFE model, engines and transmissions the ‘old’ and ‘new’ variant of the engine specific vehicles and commercial that are shared between vehicles must (or transmission) will produced for a information sources are available to 340 apply the same levels of technology, in finite number of years.’’ The CAFE inform these inputs, further research all technologies, dictated by engine or model currently carries no additional would be needed to determine whether transmission inheritance. This forced cost associated with producing both design schedules for specific engines adoption is referred to as ‘‘engine earlier revisions of an engine and the and transmissions can practicably be inheritance’’ in the model updated version simultaneously. simulated. documentation. In practice, the model Further research would be needed to The CAFE model has implemented a first chooses an ‘‘engine leader’’ among determine whether sufficient data is similar structure to address shared vehicles sharing the same engine—the likely to be available to explicitly vehicle platforms. The term ‘‘platform’’ vehicle with the highest sales in MY specify and apply additional costs is used loosely in industry but generally 2016. If there is a tie, the vehicle with involved with continuing to produce an refers to a common structure shared by the highest average MSRP is chosen, existing engine or transmission for some a group of vehicle variants. The degree representing the idea that manufacturers vehicles that have not yet progressed to of commonality varies with some will choose to pilot the newest a newer version of that engine or platform variants exhibiting traditional technology on premium vehicles if transmission. Comment is sought on ‘‘badge engineering’’ where two possible. The model applies the same products are differentiated by little more logic with respect to the application of 339 Comment by Environmental Law & Policy than insignias, while other platforms Center, Natural Resources Defense Council (NRDC), transmission changes. After the model Public Citizen, and Sierra Club, Docket ID EPA– may be used to produce a broad suite of HQ–OAR–2015–0827–9826, at 32. vehicles that bear little outer 338 2015 NAS Report, at pg. 258–259. 340 CAFE Model Peer Review, p. 19. resemblance to one another.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00179 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43164 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Given the degree of commonality practical restrictions on technology on phase-in caps and, if greater reliance between variants of a single platform, application, including the on phase-in caps is recommended, what manufacturers do not have complete improvements described above. Unlike approach and information can be used freedom to apply technology to a vehicle-specific restrictions related to to define and apply these caps. vehicle: While some technologies (e.g. redesign, refreshes or platforms/engines, (e) Interactions Between Regulatory low rolling resistance tires) are very phase-in caps constrain technology Classes nearly ‘‘bolt-on’’ technologies, others application at the vehicle manufacturer involve substantial changes to the level for a given model year. Introduced Like earlier versions, the current structure and design of the vehicle, and in the 2006 version of the CAFE model, CAFE model provides the capability for therefore necessarily are constant among they were intended to reflect a integrated analysis spanning different vehicles that share a common platform. manufacturer’s overall resource capacity regulatory classes, accounting both for NHTSA has, therefore, modified the available for implementing new standards that apply separately to CAFE model such that all mass technologies (such as engineering different classes and for interactions reduction technologies are forced to be research and development personnel between regulatory classes. Light constant among variants of a platform. and financial resources), thereby vehicle CAFE and CO2 standards are Within the analysis fleet, each vehicle ensuring that resource capacity is specified separately for passenger cars is associated with a specific platform. accounted for in the modeling process. and light trucks. However, there is Similar to the application of engine and Compared to prior analyses of light- considerable sharing between these two transmission technologies, the CAFE duty standards, these model changes regulatory classes—where a single model defines a platform ‘‘leader’’ as the result in some changes in the broad engine, transmission, or platform can vehicle variant of a given platform that characteristics of the model’s appear in both the passenger car and has the highest level of observed mass application of technology to light truck regulatory class. For reduction present in the analysis fleet. manufacturers’ fleets. Since the use of example, some sport-utility vehicles are If there is a tie, the CAFE model begins phase-in caps has been de-emphasized offered in 2WD versions classified as mass reduction technology on the and manufacturer technology passenger cars and 4WD versions vehicle with the highest sales in model deployment remains tied strongly to classified as light trucks. Integrated year 2016. If there remains a tie, the estimated product redesign and analysis of manufacturers’ passenger car model begins by choosing the vehicle freshening schedules, technology and light truck fleets provides the with the highest MSRP in MY 2016. As penetration rates may jump more ability to account for such sharing and the model applies technologies, it quickly as manufacturers apply reduces the likelihood of finding effectively levels up all variants on a technology to high-volume products in solutions that could involve introducing platform to the highest level of mass their portfolio. As a result, the model impractical levels of complexity in reduction technology on the platform. will ignore a phase-in cap to apply manufacturers’ product lines. So, if the platform leader is already at inherited technology to vehicles on Additionally, integrated fleet analysis MR3 in MY 2016, and a ‘‘follower’’ shared engines, transmissions, and provides the ability to simulate the starts at MR0 in MY 2016, the follower platforms. potential that manufacturers could earn will get MR3 at its next redesign (unless In previous CAFE rulemakings, CAFE and CO2 credits by over the leader is redesigned again before redesign/refresh schedules and phase-in complying with the standard in one that time, and further increases the MR caps were the primary mechanisms to fleet and use those credits toward level associated with that platform, then reflect an OEM’s limited pool of compliance with the standard in the follower would receive the new MR available resources during the another fleet (i.e., to simulate credit level). rulemaking time frame and the years transfers between regulatory classes). In the 2015 NPRM proposing new fuel preceding it, especially in years where While previous versions of the CAFE consumption and GHG standards for many models may be scheduled for model have represented manufacturers’ heavy-duty pickups and vans, NHTSA refresh or redesign. The newly- fleets by drawing a distinction between specifically requested comment on the introduced representation of platform-, passenger cars and light trucks, the general use of shared engines, engine-, and transmission-related current version of the CAFE model adds transmissions, and platforms within considerations discussed above augment a further distinction, capturing the CAFE rulemakings. While no the model’s preexisting representation difference between passenger cars commenter responded to this specific of redesign cycles and eliminate the classified as domestic passenger cars request, comments from some need to rely on phase-in caps. By and those classified as imports. The environmental organizations cited design, restrictions that enforce CAFE program regulates those passenger examples of technology sharing between commonality of mass reduction on cars separately, and the current version light- and heavy-duty products. NHTSA variants of a platform, and those that of the CAFE model simulates all three has continued to refine its enforce engine and transmission CAFE regulatory classes separately: implementation of an approach inheritance, will result in fewer vehicle- Domestic Passenger Cars (DC), Imported accounting for shared engines, technology combinations in a Passenger Cars (IC), and Light Trucks transmissions, and platforms, and again manufacturer’s future modeled fleet. (LT). CAFE regulations state that seeks comment on the approach, The integration of shared components standards, fuel economy levels, and recommendations regarding any other and product cadence as a mechanism to compliance are all calculated separately approaches, and any information that control the pace of technology for each class. These requirements are would facilitate implementation of the application also more accurately specified explicitly by the Energy Policy agency’s current approach or any represents each manufacturer’s unique and Conservation Act (EPCA), with the alternative approaches. position in the market and its existing 2007 Energy Independence and Security technology footprint, rather than a Act (EISA) having added the (d) Phase-In Caps technology-specific phase-in cap that is requirement to enforce minimum The CAFE model retains the ability to uniformly applied to all manufacturers standards for domestic passenger cars. use phase-in caps (specified in model in a given year. Comment is sought This update to the accounting imposes inputs) as proxies for a variety of regarding this shift away from relying two additional constraints on

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00180 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43165

manufacturers that sell vehicles in the reflected integrated analysis of the light- The ‘‘application level’’ describes the U.S.: (1) The domestic minimum floor, duty and HD PUV fleets, thereby system of the vehicle to which the and (2) Limited transfers between cars accounting for sharing between the technology is applied, which in turn classified as ‘‘domestic’’ versus those fleets. However, for most OEMs, that determines the extent to which that classified as ‘‘imported.’’ The domestic analysis showed considerably less decision affects other vehicles in a minimum floor creates a threshold that sharing between light-duty and HD PUV manufacturer’s fleet. For example, if a every manufacturer’s domestic car fleet fleets than initially expected. Today’s technology is applied at the ‘‘engine’’ must exceed without the application of analysis includes only vehicles subject level, it naturally affects all other CAFE credits. If a manufacturer’s to CAFE and light-duty CO2 standards, vehicles that share that same engine calculated standard is below the and the agencies invite comment on (though not until they themselves are domestic minimum floor, then the whether integrated analysis of the two redesigned, if it happens to be in a domestic floor is the binding constraint fleets should be pursued further. future model year). Technologies (even for manufacturers that are applied at the ‘‘vehicle’’ level can be 3. Technology Application Algorithm assumed to be willing to pay fines for applied to a vehicle model without non-compliance). The second constraint (a) Technology Representation and impacting the other models with which poses challenges for manufacturers that Pathways it shares components. Platform-level sell cars from both the domestic and technologies affect all of the vehicles on imported passenger car categories. While some properties of the a given platform, which can easily span While previous versions of the CAFE technologies included in the analysis technology classes, regulatory classes, model considered those fleets as a single are specified by the user (e.g., cost of the and redesign cycles. fleet (i.e., passenger cars), the model technology), the set of included now forces them to comply separately technologies is part of the model itself, The ‘‘application schedule’’ identifies and limits the volume of credits that can which contains the information about when manufacturers are assumed to be be shifted between them for compliance. the relationships between able to apply a given technology—with However, the CAA provides no technologies.341 In particular, the CAFE many available only during vehicle direction regarding compliance by model contains the information about redesigns. The application schedule also domestic and imported vehicles; EPA the sequence of technologies, the paths accounts for which technologies the has not adopted provisions similar to on which they reside, any prerequisites CAFE model tracks but does not apply. the aforementioned EPCA/EISA associated with a technology’s These enter as part of the analysis fleet requirements and is not doing so today. application, and any exclusions that (‘‘Baseline Only’’), and while they are Therefore, consistent with current and naturally follow once it is applied. necessary for accounting related to cost proposed CO2 regulations, the CAFE and incremental fuel economy model determines compliance for 341 Unlike the 2012 Final Rule, where each improvement, they do not represent a manufacturers’ overall passenger car technology had a single effectiveness value for the choice that manufacturers make in the fleets for EPA’s program. CAFE analysis, technology effectiveness in the model. As discussed in Section II.B, the current version of the CAFE model is based on the analysis fleet contains the information During 2015–2016, a single version of ANL simulation project and defined for each the CAFE model was applied to produce combination of technologies, resulting in more than about each vehicle model, engine, and analyses supporting both a rulemaking 100,000 technology effectiveness values for each of transmission selected for simulation and regarding heavy-duty pickups and vans ten technology classes. This large database is defines the initial technology state of extracted locally the first time the model is run and the fleet relative to the sets of (HD PUV) and the 2016 draft TAR can be modified by the user in that location to regarding CAFE standards for passenger reflect alternative assumptions about technology technologies in Table II–80 and Table cars and light trucks. Both analyses effectiveness. II–81.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00181 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43166 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00182 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.114 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43167

As Table II–80 and Table II–81 show, updates to shared components during example) throughout a manufacturer’s all of the engine technologies may only refresh. For example, assume Vehicle A portfolio and reduces the number of be applied (for the first time) during and Vehicle B share Engine 1, and years during which a manufacturer redesign. New transmissions can be engine 1 is redesigned as part of Vehicle would build both new and legacy applied during either refresh or A’s redesign in MY 2020. Vehicle B is versions of the same engine. Despite redesign, except for manual not redesigned until 2025 but is increasing the rate of technology transmissions, which can only be refreshed in MY 2022. In the current diffusion, this change still restricts the upgraded during redesign. Unlike version of the CAFE model, Vehicle B pace at which new engines (for previous versions of the model, which would inherit the updated version of example) can be designed and built (i.e., only allowed significant changes to Engine 1 when it is freshened in MY no faster than the redesign schedule of vehicle powertrains at redesign, this 2022. This change allows more rapid the ‘‘leader’’ vehicle to which they are version allows vehicles to inherit diffusion of powertrain updates (for tied). The only technology for which

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00183 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.115 43168 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

this does not hold is mass reduction system or sub-system (e.g., engine TAR analysis, the model uses separate improvements; these occur at the technologies). technology classes for compact cars, platform level, and each model on that Technology classes, shown in Table– midsize cars, small SUVs, large SUVs, platform must be redesigned (not merely II–82, are a means for specifying and pickup trucks. However, in this refreshed) in order to receive the newest common technology input assumptions analysis, each of those distinctions also version of the platform that contains the for vehicles that share similar has a ‘‘performance’’ version, that most current mass reduction characteristics. Predominantly, these represents another class with similar technology. classes signify the degree of body style but higher levels of The CAFE model defines several applicability of each of the available performance attributes (for a total of 10 ‘‘technology classes’’ and ‘‘technology technologies to a specific class of technology classes). As the model pathways’’ for logically grouping all vehicles and represent a specific set of available technologies for application on Autonomie simulations (conducted as simulates compliance, identifying a vehicle. Technology classes provide part of the Argonne National Lab large- technologies that can be applied to a costs and improvement factors shared scale simulation study) that determine given manufacturer’s product portfolio by all vehicles with similar body styles, the effectiveness of each technology to to improve fleet fuel economy, it relies curb weights, footprints, and engine improve fuel economy. The vehicle on the vehicle class to provide relevant types, while technology pathways technology classes also define, for each cost and effectiveness information for establish a logical progression of technology, the additional cost each vehicle model. technologies on a vehicle within a associated with application.342 Like the

The model defines technology evaluating all paths, the model selects technology’s effectiveness, cost, and pathways for grouping and establishing the most cost-effective solution among interaction with all other technologies a logical progression of technologies on all pathways. This parallel path both present and available. a vehicle. Each pathway (or path) is approach allows the modeling system to (b) Technology Paths evaluated independently and in progress through technologies in any parallel, with technologies on these given pathway without being The modeling system incorporates 16 unnecessarily prevented from paths being considered in sequential technology pathways for evaluation as order. As the model traverses each path, considering technologies in other paths. Rather than rely on a specific set of shown in Table–II—83. Similar to the costs and fuel economy technology combinations or packages, individual technologies, each path improvements are accumulated on an the model considers the universe of carries an intrinsic application level that incremental basis with relation to the applicable technologies, dynamically denotes the scope of applicability of all preceding technology. The system stops identifying the most cost-effective technologies present within that path examining a given path once a combination of technologies for each and whether the pathway is evaluated combination of one or more manufacturer’s vehicle fleet based on on one vehicle at a time, or on a technologies results in a ‘‘best’’ each vehicle’s initial technology content collection of vehicles that share the technology solution for that path. After and the assumptions about each same platform, engine, or transmission.

342 Inputs are specified to assign each vehicle in the analysis fleet to one of these technology classes, as discussed in Section II.B.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00184 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.116 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43169

The technologies that comprise the compression ratio engines) was not used engine paths (Turbo, HCR, Diesel, and five Engine-Level paths available within in this analysis but is present within the ADEAC) simultaneously. They are the model are presented in Figure-II-13. model. Unlike earlier versions of the assumed to be mutually exclusive. Once Note: The baseline-level technologies CAFE model, that enforced strictly one path is taken, it locks out the others (SOHC, DOHC, OHV, and CNG) appear sequential application of technologies to avoid situations where the model in gray boxes. These technologies are like VVL and SGDI, this version of the could be perceived to force used to inform the modeling system of CAFE model allows basic engine manufacturers to radically change the initial engine’s configuration and are technologies to be applied in any order engine architecture with each redesign, not otherwise applicable during the once an engine has VVT (the base state incurring stranded capital costs and lost analysis. Additionally, the VCR path of all ANL simulations). Once the model opportunities for learning. (intended to house fuel economy progresses past the basic engine path, it improvements from variable considers all of the more advanced

For all pathways, the technologies are technology is deemed ineffective, the later in the pathway, it will ‘‘backfill’’ evaluated and applied to a vehicle in system will bypass it and skip ahead to anything that was previously skipped in sequential order, as shown from top to the next technology. If the modeling order to fully account for costs and fuel bottom. In some cases, however, if a system applies a technology that resides economy improvements of the full

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00185 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.117 EP24AU18.118 43170 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

technology combination.343 For any The Basic Engine path begins with analysis fleet. Similarly, all automatic technology that is already present on a SOHC, DOHC, and OHV technologies transmissions with five forward gears or vehicle (either from the MY 2016 fleet defining the initial configuration of the fewer have been assigned the AT5 or previously applied by the model), the vehicle’s engine. Since these technology. The model preserves the system skips over those technologies as technologies are not available during initial configuration for as long as well and proceeds to the next. These modeling, the system evaluates this possible, and prohibits manual skipped technologies, however, will not pathway starting with VVT. Whenever a transmissions from becoming automatic be applied again during backfill. technology pathway forks into two or transmissions at any point. Automatic While costs are still purely more branch points, as the engine path transmissions may become CVT level 2 incremental, technology effectiveness is does at the end of the basic engine path, after progressing though the 6-speed no longer constructed that way. The all of the branches are treated as automatic. While the structure of the non-sequential nature of the basic mutually exclusive. The model model still allows automatic engine technologies have no obvious evaluates all technologies forming the transmissions to consider the move to preceding technology except for VVT, branch simultaneously and selects the DCT, in practice they are restricted from the root of our engine path. It was a most cost-effective for the application, doing so in the market data file. This natural extension to carry this approach while disabling the unchosen remaining allows vehicles that enter with a DCT to to the other branches as well. The paths. improve it (if opportunities to do so technology effectiveness estimates are The technologies that make up the exist) but does not allow automatic now an integrated part of the CAFE four Transmission-Level paths defined transmissions to become DCTs, in model and represent a translation of the by the modeling system are shown in recognition of low consumer Argonne simulation database that Figure-II-14. The baseline-level enthusiasm for the earlier versions the compares the fuel consumption of any technologies (AT5, MT5 and CVT) transmission that have been introduced combination of technologies (across all appear in gray boxes and are only used over the last decade. The model does paths) to the base vehicle (that has only to represent the initial configuration of not attempt to simulate ‘‘reversion’’ to VVT, 5-speed automatic transmission, a vehicle’s transmission. For simplicity, less advanced transmission no electrification, and no body-level all manual transmissions with five technologies, such as replacing a 6- improvements).344 forward gears or fewer have been speed AT with a DCT and then assigned the MT5 technology in the replacing that DCT with a 10-speed AT. 343 More detail about how the Argonne simulation The agencies invite comment on database was integrated into the CAFE model can manual transmissions, nor the inverse, technology whether or not the model should be be found in PRIA Chapter 6. combinations containing manual transmissions use 344 This is true for all combinations other than a reference point identical to the base vehicle modified to simulate such ‘‘reversion’’ those containing manual transmissions. Because the description, but containing a 5-speed manual rather and, if so, how this possible behavior model does not convert automatic transmissions to than automatic transmission. might be practicably simulated.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00186 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43171

The root of the Electrification path, larger architectural change of the two. In electrification technologies, there are shown in Figure-II-15, is a conventional general, the electrification technologies two electrical system improvements, powertrain (CONV) with no are applied as vehicle-level electric power steering (EPS) and electrification. The two strong hybrid technologies, meaning that the model accessory improvements (IACC), which technologies (SHEVP2 and SHEVPS) on applies them without affecting were not part of the ANL simulation the Hybrid/Electric path, are defined as components that might be shared with project and are applied by the model as stand-alone and mutually exclusive. other vehicles. In the case of the more fixed percentage improvements to all These technologies are not incremental advanced electrification technologies, technology combinations in a particular over each other for cost or effectiveness where engines and transmissions are technology class. Their improvements and do not follow a traditional removed or replaced, the model will are superseded by technologies in the progression logic present on other paths. choose a new vehicle to be the leader on other electrification paths, BISG or While the SHEVP2 represents a hybrid that component (if necessary) and will CISG, in the case of EPS, and strong system paired with the existing engine not force other vehicles sharing that hybrids (and above) in the case of IACC, on a given vehicle, the SHEVPS removes engine or transmission to become which are assumed to include those and replaces that engine, making it the hybrids (or EVs). In addition to the improvements already.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00187 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.119 43172 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The technology paths related to load vehicle level, and technologies within technology path. In addition to choosing reduction of the vehicle are shown in each path are considered purely among possible engine, transmission, Figure-II-16. Of these, only the Mass sequential. For mass reduction, and electrification improvements to Reduction (MR) path is applied at the aerodynamic improvements, and improve a vehicle’s fuel economy, the platform level, thus affecting all reductions in rolling resistance, the base CAFE model will consider technologies vehicles (across classes and body styles) level of each path is the ‘‘zero state,’’ in each of the possible load improvement on a given platform. The remaining which a vehicle has exhibited none of paths simultaneously. technology paths are all applied at the the improvements associated with the

Even though the model evaluates each the pathways are interconnected to and incremental accounting of technology path independently, some of allow for additional logical progression technologies. For example, the cost of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00188 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.120 EP24AU18.121 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43173

SHEVPS (power-split strong hybrid/ on that vehicle. However, if it becomes of evaluation. For example, a vehicle electric) on the Hybrid/Electric path is necessary for a vehicle to progress to the with a 6-speed automatic transmission defined as incremental over the power-split hybrid, the model will will not be able to get improvements complete basic engine path (an engine virtually apply the technologies from a Manual Transmission path. For that contains VVT, VVL, SGDI, and associated with the reference point in this reason, the model implements logic DEAC), the AT5 (5-speed automatic) order to evaluate the attractiveness of to explicitly disable certain paths technology on the Automatic transitioning to the strong hybrid. whenever a constraining technology Transmission path, and the CISG (crank Of the 17 technology pathways from another path is applied on a present in the model, all Engine paths, mounted integrated starter/generator) vehicle. On occasion, not all of the the Automatic Transmission path, the technology on the Electrification path. technologies present within a pathway Electrification path, and both Hybrid/ For that reason, whenever the model Electric paths are logically linked for may produce compatibility constraints evaluates the SHEVPS technology for incremental technology progression. with another path. In such a case, the application on a vehicle, it ensures that, Some of the technology pathways, as model will selectively disable a at a minimum, all the aforementioned defined in the model and shown in conflicting pathway (or part of the technologies (as well as their Figure-II-17, may not be compatible pathway) as required by the predecessors) have already been applied with a vehicle given its state at the time incompatible technology.

For any interlinked technology to always move in the direction of MY 2016 analysis fleet contains pathways shown in Figure-II-17, the increasing technological sophistication information about each manufacturer’s: model also disables all preceding each time they are reevaluated by the • Vehicle models offered for sale—their technology paths whenever a vehicle model. current (i.e., MY 2016) production volumes, transitions to a succeeding pathway. For 4. Simulating Manufacturer Compliance manufacturer suggested retail prices (MSRPs), fuel saving technology content example, if the model applies SHEVPS With Standards technology on a vehicle, the model (relative to the set of technologies described disables the Turbo, HCR, ADEAC, and As a starting point, the model needs in Table II–80 and Table II–81), and other Diesel Engine paths, as well as the Basic enough information to represent each attributes (curb weight, drive type, assignment to technology class and Engine, the Automatic Transmission, manufacturer covered by the program. As discussed above in Section II.B, the regulatory class), and the Electrification paths (all of • Production constraints—product which precede the Hybrid/Electric cadence of vehicle models (i.e., schedule of 345 vehicle. This technology may be present in path). This implicitly forces vehicles conjunction with any engine-level technology, and model redesigns and ‘‘freshenings’’), vehicle as such, the Basic Engine path is not disabled upon platform membership, degree of engine and/ 345 The only notable exception to this rule occurs application of SHEVP2 technology, even though or transmission sharing (for each model whenever SHEVP2 technology is applied on a this pathway precedes the Hybrid/Electric path. variant) with other vehicles in the fleet,

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00189 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.122 43174 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

• Compliance constraints and this aspect of the model even more technology for which the effective cost flexibilities—historical preference for full realistic. is negative). Then the model applies compliance or penalty payment/credit This construction allows the model to expiring overcompliance credits (if application, willingness to apply additional choose technologies that both improve a allowed to under the perspective of cost-effective fuel saving technology in manufacturer’s regulatory compliance either the ‘‘unconstrained’’ or ‘‘standard excess of regulatory requirements, projected applicable flexible fuel credits, and current position and are most likely to be setting’’ analysis, for CAFE purposes). credit balance (by model year and regulatory attractive to its consumers. This also At this point, the model checks the class) in first model year of simulation. means that different assumptions about manufacturer’s compliance status again. Each manufacturer’s regulatory future fuel prices will produce different If the manufacturer is still not compliant requirement represents the production- rankings of technologies when the (and is unwilling to pay civil penalties, weighted harmonic mean of their model evaluates available technologies again for CAFE), the model will add vehicle’s targets in each regulated fleet. for application. For example, in a high technologies that are not cost-effective This means that no individual vehicle fuel price regime, an expensive but very until the manufacturer reaches has a ‘‘standard,’’ merely a target, and efficient technology may look attractive compliance. If the manufacturer each manufacturer is free to identify a to manufacturers because the value of exhausts opportunities to comply with compliance strategy that makes the most the fuel savings is sufficiently high to the standard by improving fuel sense given its unique combination of both counteract the higher cost of the economy/reducing emissions (typically vehicle models, consumers, and technology and, implicitly, satisfy due to a limited percentage of its fleet competitive position in the various consumer demand to balance price being redesigned in that year), the market segments. As the CAFE model increases with reductions in operating model will apply banked CAFE or CO2 provides flexibility when defining a set cost. Similarly, technologies for which credits to offset the remaining deficit. If of regulatory standards, each there exist consumer welfare losses no credits exist to offset the remaining manufacturer’s requirement is (discussed in Section II.E) will be seen deficit, the model will reach back in dynamically defined based on the as less attractive to manufacturers who time to alter technology solutions in specification of the standards for any may be concerned about their ability to earlier model years. simulation and the distribution of recover the full amount of the The CAFE model implements multi- footprints within each fleet. technology cost during the sale of the year planning by looking back, rather Given this information, the model vehicle. The model continues to add than forward. When a manufacturer is technology until a manufacturer either: attempts to apply technology to each unable to comply through cost-effective (a) Reaches compliance with regulatory (i.e., producing effective cost values less manufacturer’s fleet in a manner than standards (possibly through the than zero) technology improvements or minimizes ‘‘effective costs.’’ The accumulation and application of credit application in a given year, the effective cost captures more than the overcompliance credits), (b) reaches a model will ‘‘reach back’’ to earlier years incremental cost of a given technology; point at which it is more cost effective and apply the most cost-effective it represents the difference between to pay penalties than to add more technologies that were not applied at their incremental cost and the value of technology (for CAFE), or (c) reaches a that time and then carry those fuel savings to a potential buyer over the point beyond compliance where the technologies forward into the future and first 30 months of ownership.346 In manufacturer assumes its consumers re-evaluate the manufacturer’s addition to the technology cost and fuel will be unwilling to pay for additional compliance position. The model repeats savings, the effective cost also includes fuel saving/emissions reducing this process until compliance in the the change in fines from applying a technologies. current year is achieved, dynamically given technology and any estimated In general, the model adds technology rebuilding previous model year fleets welfare losses associated with the for several reasons but checks these and carrying them forward into the technology (e.g., earlier versions of the sequentially. The model then applies future, accumulating CAFE or CO2 CAFE model simulated low-range any ‘‘forced’’ technologies. Currently, credits from over-compliance with the electric vehicles that produced a welfare only VVT is forced to be applied to standard wherever appropriate. loss to buyers who valued standard vehicles at redesign since it is the root In a given model year, the model operating ranges between re-fueling of the engine path and the reference determines applicability of each events). The effective cost metric point for all future engine technology technology to each vehicle model, applied by the model does not attempt applications.347 The model next applies platform, engine, and transmission. The to reflect all costs of vehicle ownership. any inherited technologies that were compliance simulation algorithm begins Further research would be required in applied to a leader vehicle and carried the process of applying technologies order to support simulation that forward into future model years where based on the CAFE or CO2 standards assumes buyers behave as if they follower vehicles (on the shared system) specified during the current model year. actually consider all ownership costs, are freshened or redesigned (and thus This involves repeatedly evaluating the and that assumes manufacturers eligible to receive the updated version degree of noncompliance, identifying respond accordingly. The agencies will of the shared component). In practice, the next ‘‘best’’ technology (ranked by continue to consider the metric applied very few vehicle models enter without the effective cost discussed earlier) to represent manufactuers’ approach to VVT, so inheritance is typically the first available on each of the parallel making decisions regarding the step in the compliance loop. Then the technology paths described above and application of fuel-saving technologies model evaluates the manufacturer’s applying the best of these. The and invite comment regarding any compliance status, applying all cost- algorithm combines some of the practicable changes that might make effective technologies regardless of pathways, evaluating them sequentially compliance status (essentially any instead of in parallel, in order to ensure 346 The length of time over which to value fuel appropriate incremental progression of savings in the effective cost calculation is a model 347 As a practical matter, this affects very few technologies. input that can be modified by the user. This vehicles. More than 95% of vehicles in the market The algorithm first finds the best next analysis uses 30 months’ worth of fuel savings in file either already have VVT present or have the effective cost calculation, using the price of fuel surpassed the basic engine path through the applicable technology in each of the at the time of vehicle purchase. application of hybrids or electric vehicles. technology pathways then selects the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00190 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43175

best among these. For CAFE purposes, The Chevrolet Equinox is represented and XTS,349 the Chevrolet Colorado, the model applies the technology to the in the model inputs as a single Impala and Traverse (which is the affected vehicles if a manufacturer is nameplate, with five model variants designated ‘‘leader’’), and the GMC either unwilling to pay penalties or if distinguished by the presence of all- Acadia, Canyon, and Terrain. This is an applying the technology is more cost- wheel drive and four distinct example of how shared and inherited effective than paying penalties. powertrain configurations (two engines components interact with product Afterwards, the algorithm reevaluates paired with two different cadence: when the Equinox nameplate the manufacturer’s degree of transmissions). Across all five model is redesigned, the CAFE model has more noncompliance and continues variants, GM produced above 220,000 leverage over some variants than others application of technology. Once a units of the Equinox nameplate. About and cannot make changes to the engines manufacturer reaches compliance (i.e., 150,000 units of that production volume of the variants of the Equinox with V– the manufacturer would no longer need is regulated as Domestic Passenger Car, 6 unless that change is consistent with to pay penalties), the algorithm with the remainder regulated as Light all of the other nameplates just listed. proceeds to apply any additional Trucks. The easiest way to describe the The transmissions on the other variants technology determined to be cost- actions taken by the CAFE model is to of the Equinox are similarly widely effective (as discussed above). focus on a single model variant of the shared and represent the same kind of Conversely, if a manufacturer is Equinox (one row in the market data production constraint just described assumed to prefer to pay penalties, the file). The model variant of the Equinox with respect to the engine. When algorithm only applies technology up to with the highest production volume, accounting for the full set of engines, the point where doing so is less costly about 130,000 units in MY 2016, is transmissions, and platforms than paying penalties. The algorithm vehicle code 110111.348 This unique represented across the Equinox stops applying additional technology to model variant is the basis for the nameplate’s five variants, components this manufacturer’s products once no example. However, because it is only are shared across all three regulatory more cost-effective solutions are one of five variants on the Equinox classes. encountered. This process is repeated nameplate, the modifications made to This example uses a ‘‘standard for each manufacturer present in the that model in the simulation will affect setting’’ perspective to minimize the input fleet. It is then repeated again for the rest of the Equinox variants and amount of credit generation and each model year. Once all model years other vehicles across all fleets. application, in order to focus on the have been processed, the compliance The example Equinox variant is mechanics of technology application simulation algorithm concludes. The designated as an engine and platform and component sharing. The actions leader. As discussed earlier, this implies taken by the CAFE model when process for CO2 standard compliance simulation is similar, but without the that modifications to its engine (11031, operating on the example Equinox option of penalty payment. a 2.4L I–4) are tied to the redesign during GM’s compliance simulation are cadence of this Equinox, as are shown in Table–II–84. In general, the (a) Compliance Example modifications to its platform (Theta/TE). example Equinox begins the compliance The engine is shared by the Buick The following example will illustrate simulation with the technology LaCrosse, Regal, and Verano, and by the the features discussed above for the observed in its MY 2016 incarnation— GMC Terrain (as well as appearing in CAFE program. While the example a 2.6L I–4 with VVT and SGDI, a 6- two other variants of the Equinox). So describes the actions that General speed automatic transmission, low those vehicles, if redesigned after this Motors takes to modify the Chevrolet rolling resistance tires (ROLL20) and a Equinox, will inherit changes to engine Equinox in order to comply with the 10% realized improvement in 11031 when they are redesigned, augural standards (the baseline in this aerodynamic drag (AERO10). In MY carrying the legacy version of the engine analysis), and the logical consequences 2018, the Equinox is redesigned, at until then. Similarly, this Equinox of these actions, a similar example which time the engine adds VVL and shares its platform with the Cadillac level-1 turbocharging. The transmission would develop if instead simulating SRX and GMC Terrain, which will compliance with the EPA standards for on the Malibu is upgraded to an 8-speed inherit changes made to this platform automatic in 2018, which the Equinox those years. The structure of GM’s fleet when they are redesigned (if later than and the mechanisms at work in the also gets. The platform, for which this the Equinox, as is the case with the Equinox is the designated leader, gets CAFE model are identical in both cases, SRX). but different features of each program level-4 mass reduction. The CAFE This specific Equinox is a model also applies a few vehicle-level (unlimited credit transfers between transmission ‘‘follower,’’ getting updates fleets, for example) would likely cause technologies: low-drag brakes, made to its transmission leader (the electronic accessory improvements, and the model to choose different Chevrolet Malibu) when it is freshened technology solutions. additional aerodynamic improvements or redesigned. Additionally, two other (AERO20). Upon refresh in MY 2021, it At the start of the simulation in MY variants of the Equinox nameplate (the acquires an upgraded 10-speed 2016, GM has 30 unique engines shared more powerful versions, containing a transmission (AT10) from the Malibu. across over 33 unique nameplates, 260 3.6L V–6 engine) are not ‘‘leaders’’ on model variants, and three regulatory any of the primary components. Those 349 The agencies recognized that GM last classes. As discussed earlier, the CAFE variants are built on the same platform produced the Cadillac SRX for MY 2016, and note model will attempt to preserve that level as the example Equinox variant but this as one example of the limitations of using an of sharing across GM’s fleets to avoid share their engine with the Buick analysis fleet defined in terms of even a recent actual model year. Section II.B discusses these introducing additional production Enclave and LaCrosse, the Cadillac SRX tradeoffs, and the tentative judgment that, as a complexity for which the agencies do foundation for analysis presented here, it was better not estimate additional costs. An even 348 This numeric designation is not important to to develop the analysis fleet using the best smaller number of transmissions (16) understand the example but will allow an information available for MY 2016 than to have interested reader to identify the vehicle in model used manufacturers’ CBI to construct an analysis and platforms (12) are shared across the outputs to either recreate the example or use it as fleet that, though more current, would have limited same set of nameplates, model variants, a template to create similar examples for other the agencies’ ability to make public all analytical and regulatory classes. manufacturers and vehicles. inputs and outputs.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00191 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43176 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Then in MY 2025 it is redesigned again mass of the platform (MR5). Using an persistent CAFE deficits in one or more and upgrades the engine to level-2 ‘‘unconstrained’’ perspective would fleets. In the ‘‘standard setting’’ turbocharging, replaces the 10-speed possibly lead to additional actions taken perspective, that forces compliance automatic transmission with a 8-speed after MY 2025, where GM may have without the use of CAFE credits, this is automatic transmission, adds a P2 been simulated to use credits earned in not an issue. strong hybrid, and further reduces the earlier model years to offset small,

The technology applications engine it inherits is the same one that upgraded again until MY 2026, so the described in Table–II–84 have was redesigned in MY 2018. This means performance versions of the Equinox consequences beyond the single variant that the Verano will improve its fuel continue with the 2018 version of the of the Equinox shown in the table. In economy in MY 2019 when the new engine throughout the remainder of the particular, two other variants of the engine is inherited but only to the simulation. While these inheritances Equinox (both of which are regulated as extent that the new version of the and sharing dynamics are not a perfect Light Trucks) get the upgraded engine, engine is an improvement over the representation of each manufacturer’s which they share with the example, in legacy version in the context of the specific constraints, nor the flexibilities MY 2018. Thus, this application of Verano’s other technology (which it is— available to shift strategies in real-time engine technology to a single variant of the Verano moves from 32 MPG to 44 as a response to changing market or the Equinox in the Domestic Car fleet, MPG when accounting for the other regulatory conditions, they are a ‘‘spills over’’ into the Light Truck fleet, technologies added during the MY 2019 reasonable way to consider the resource generating improvements in fuel redesign). constraints that prohibit fleet-wide economy and additional costs. This same story continues with the technology diffusion over shorter Furthermore, the Buick LaCrosse and diffusion of platform improvements windows than have been observed Regal, and the GMC Terrain also get the simulated by the CAFE model in MY historically and for which the agencies same engine, which they share with the 2018. The GMC Terrain is simulated to have no way to impose additional costs. example, in MY 2018. Those vehicles be redesigned in MY 2018, in Aside from the technology application also span the Domestic Car and Light conjunction with the Equinox. The and its consequences throughout the Truck fleets. However, the Buick performance variants of the Equinox, GM product portfolio, discussed above, Verano, which is not redesigned until with a 3.5L V–6, also upgrade their there are other important conclusions to MY 2019, continues with the legacy engines in MY 2018 (in conjunction draw from the technology application (i.e., MY 2016) version of the shared with the estimated Chevrolet Traverse example. The first of these is that engine until it is redesigned. When it redesign). However, when the Equinox product cadence matters, and only by inherits the new engine in MY 2019, it is next redesigned in MY 2025, the taking a year-by-year perspective can does so without modification; the engine shared with the Traverse is not this be seen. When the example Equinox

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00192 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.123 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43177

is redesigned in MY 2018, the CAFE years and the standards present in those credits to offset deficits in the DC and model takes actions that cause the years. This is one reason why the CAFE LT fleets. In MY 2028, when GM is redesigned Equinox to significantly model, rather than OMEGA, was chosen simulated to aggressively apply exceed its fuel economy target. While no to examine the impacts of the proposed technology to the example Equinox, the single vehicle has a ‘‘standard,’’ having standards in this analysis. DC fleet exceeds its standard while the high volume vehicles significantly Another feature of note in Table–II–84 LT fleet still generates deficits. The below their individual targets can is the cost of applying these CAFE model offset that deficit with present compliance challenges for technologies. The costs are all expiring (and possibly transferred) manufacturers who must compensate by denominated in dollars and represent credits. However, by MY 2020 the exceeding targets on other vehicles. incremental cost increases relative to ‘‘standard setting’’ perspective removes While the example Equinox exceeds its the MY 2016 version of the Equinox. the option of using CAFE credits to MY 2018 target by almost 9 mpg, this Aside from the cost increase of over offset deficits and GM exceeds the version of the Equinox is not eligible to $5,000 in MY 2025 when the vehicle is standard in all three fleets, though by converted to a strong hybrid, the see significant technology changes again almost 2 mpg in DC and LT. As the incremental technology costs display a before MY 2025 (except for the Equinox example showed, many of the consistent trend between application transmission upgrade that occurs in MY vehicles redesigned in MY 2020 will events—decreasing steadily over time as 2021). Thus, the CAFE model is the cost associated with each given still be produced at the MY 2020 redesigning the Equinox in MY 2018 combination of technologies ‘‘learns technology level in MY 2025 where GM with respect to future targets and down.’’ By MY 2032, even the most is simulated to comply exactly across all standards—this Equinox is nearly 2 mpg expensive version of the example three fleets. Under an ‘‘unconstrained’’ below its target in MY 2024 before being Equinox costs nearly $800 less to perspective, the CAFE model would use redesigned in MY 2025. This reflects a produce than it did in MY 2025. the CAFE credits earned through over- real challenge that manufacturers face in The technology application in the compliance with the standards in MYs the context of continually increasing example occurs in the context of GM’s 2020–2023 to offset deficits created by CAFE standards, and represents a clear attempt to comply with the augural under-compliance as the standards example of why considering two model standards. As some of the components continued to increase, pushing some year snapshots where all vehicles are on the Equinox nameplate are shared technology application until later years assumed to be redesigned is across all three regulated fleets, Table– when the standards stabilized and those unrealistically simplistic. The MY 2018 II–85 shows the compliance status of credits expired. The CAFE model version of the example Equinox persists each fleet in MYs 2016–2025. In MY simulates compliance through MY 2032 (with little change) through six model 2017, the CAFE model applies expiring to account for this behavior.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00193 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43178 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

(b) Representation of OEMs’ Potential consequences of not satisfying those savings that occurred in the simulation Responsiveness to Buyers’ Willingness requirements), but also on what they after MY 2016 to the proposed standards To Pay for Fuel Economy Improvements think they can sell, technology-wise, to because none of the fuel economy consumers. In the 2012 final rule, the improvements were considered likely to The CAFE model simulates manufacturer responses to both agencies analyzed alternatives under the occur in the absence of increasing regulatory standards and technology assumption that manufacturers would standards. However, this assumption availability. In order to do so, it requires not improve the fuel economy of new contradicted much of the literature on assumptions about how the industry vehicles at all unless compelled to do so this topic and the industry’s recent views consumer demand for additional by the existence of increasingly experience with CAFE compliance, and fuel economy because manufacturer stringent CAFE and GHG standards.350 for CAFE standards, the analysis responses to potential standards depend This ‘‘flat baseline’’ assumption led the published in 2016 applied a reference not just on what they think they are best agencies to attribute all of the fuel case estimate that manufacturers will off producing to satisfy regulatory treat all technologies that pay for requirements (considering the 350 See, e.g., 75 FR 62844, 75 FR 63105. themselves within the first three years

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00194 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.124 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43179

of ownership (through reduced though, have thus far declined to manufacturers different compliance expenditures on fuel) as if the cost of disclose the actual terms of CAFE or positions relative to the standards, and 351 that technology were negative. CO2 credit trades, so this calculation in light of the likelihood that some The industry has exceeded the currently uses the CAFE civil penalty OEMs will continue to use civil required CAFE level for both passenger rate as the basis to estimate this value. penalties as a means to resolve CAFE cars and light trucks in the past; It seems reasonable to assume that the deficits (at least for some fleets). notably, by almost 5 mpg during the fuel CAFE civil penalty rate likely sets an While described in greater detail in price spikes of the 2000s when CAFE effective ceiling on the price of any the CAFE model documentation, the standards for passenger cars were still traded CAFE credits, and considering effective cost reflects an assumption not frozen at levels established for the 1990 that each manufacturer can only about consumers’ actual willingness to model year.352 In fact, a number of produce one fleet of vehicles for sale in pay for additional fuel economy but manufacturers that traditionally paid the U.S., prices of CO2 credits might about what manufacturers believe CAFE civil penalties even reached reasonably be expected to be equivalent consumers are willing to pay. The compliance during years with to prices of CAFE credits. However, the reference case estimate for today’s sufficiently high fuel prices.353 The current CAFE model does not explicitly analysis is that manufacturers will treat model attempts to account for this simulate credit trading; therefore, the all technologies that pay for themselves observed consumer preference for fuel change in the value of CO2 credits within the first 21⁄2 years of ownership economy, above and beyond that should only capture the change in (through reduced expenditures on fuel) required by the regulatory standards, by manufacturer’s own cost of compliance, as if the cost of that technology were allowing fuel price to influence the so the compliance simulation algorithm negative. Manufacturers have repeatedly ranking of technologies that the model applies a ceiling at 0 (zero) to each indicated to the agencies that new 355 considers when modifying a calculated value of the CO2 credits. vehicle buyers are only willing to pay manufacturer’s fleet in order to achieve Just as manufacturers’ actual for fuel economy-improving technology compliance. In particular, the model approaches to vehicle pricing are if it pays back within the first two to ranks available technology not by cost, closely held, manufacturers’ actual three years of vehicle ownership.356 but by ‘‘effective cost.’’ future approaches to making decisions NHTSA has therefore incorporated this When the model chooses which about technology are not perfectly assumption (of willingness to pay for technology to apply next, it calculates knowable. The CAFE model is intended technology that pays back within 30 the effective cost of available to illustrate ways manufacturers could months) into today’s analysis. technologies and chooses the respond to standards, given a set of Alternatives to this 30-month estimate technology with the lowest effective production constraints, not to predict are considered in the sensitivity cost. The ‘‘effective cost’’ itself is a how they will respond. Alternatives to analysis included in today’s notice. In combination of the technology cost, the these ‘‘effective cost’’ metrics have been the current version of the model, this fuel savings that would occur if that considered and will continue to be assumption holds whether or not a technology were applied to a given considered. For example, instead of manufacturer has already achieved vehicle, the resulting change in CAFE using a dollar value, the model could compliance. This means that the most penalties (as appropriate), and the use a ratio, such as the net cost cost-effective technologies (those that affected volumes. User inputs determine (technology cost minus fuel savings) of pay back within the first 21⁄2 years) are how much fuel savings manufacturers an application of technology divided by applied to new vehicles even in the believe new car buyers will pay for corresponding quantity of avoided fuel absence of regulatory pressure. (denominated in the number of years consumption or CO2 emissions. Any However, because the value of fuel before a technology ‘‘pays back’’ its alternative metric has the potential to savings depends upon the price of fuel, cost). shift simulated choices among the model will add more technology Because the civil penalty provisions technology application options, and even without regulatory pressure when specified for CAFE in EPCA do not some metrics would be less suited to the fuel prices are high compared to apply to CO2 standards, the effective CAFE model’s consideration of simulations where fuel prices are cost calculation applied when multiyear product planning, or less assumed to be low. This assumption is simulating compliance with CO2 adaptable than others to any future consistent with observed historical standards uses an estimate of the simulation of credit trading. Comment is compliance behavior (and consumer potential value of CO2 credits. Including sought regarding the definition and demand for fuel economy in the new a valuation of CO2 credits in the application of criteria to select among vehicle market), as discussed above. effective cost metric provides a potential technology options and determine when One implication of this assumption is basis for future explicit modeling of to stop applying technology (consider that futures with higher, or lower, fuel credit trading.354 Manufacturers, not only standards, but also factors such prices produce different sets of as fuel prices, civil penalties for CAFE, attractive technologies (and at different 351 Draft TAR, p. 13–10, available at https:// and the potential value of credits for times). For example, if fuel prices were www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/ both programs), and this aspect of the above $7/gallon, many of the Draft-TAR-Final.pdf (last accessed June 15, 2018). model may be further revised. Any technologies in this analysis could pay 352 NHTSA, Summary of Fuel Economy future revision to the effective cost Performance, 2014, available at https:// for themselves within the first year or www.nhtsa.gov/sites/nhtsa.dot.gov/files/ would be considered in light of two and would be applied at high rates performance-summary-report-12152014-v2.pdf (last in all of the alternatives. Similarly, at accessed June 27, 2018). and future efforts will focus consideration on more the other extreme (significantly reduced 353 Ibid. Additional data available at https:// plausible imperfect trading. fuel prices), almost no additional fuel one.nhtsa.gov/cafe_pic/CAFE_PIC_Mfr_LIVE.html 355 Having the model continue to add technology (last accessed June 27, 2018). in order to build a surplus of credits as warranted economy would be observed. 354 By treating all passenger cars and light trucks by the estimated (whether specified as a model as being manufactured by a single ‘‘OEM,’’ inputs input or calculated dynamically as a clearing price) 356 This is supported by the 2015 NAS study, to the CAFE model can be structured to simulate market value of credits would provide part of the which found that consumers seek to recoup added perfect trading. However, competitive and other basis for having the model build the supply side of upfront purchasing costs within two or three years. factors make perfect trading exceedingly unlikely, an explicitly-simulated credit trading market. See 2015 NAS Report, at pg. 317.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00195 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43180 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

While these assumptions about effective cost of the technology (which model years, or negotiated settlements desired payback period and consumer includes technology cost, consumer fuel with NHTSA to resolve deficits). preferences for fuel economy may not savings, consumer welfare changes, and In the current analysis, NHTSA has affect the eventual level of achieved the cost of penalties for non-compliance relied on past compliance behavior and CAFE and CO2 emissions in the later with the standard) is less costly than certified transactions in the credit years of the program, they will affect the paying civil penalties or purchasing market to designate some manufacturers amount of additional technology cost credits. Unlike previous versions of the as being willing to pay CAFE penalties and fuel savings that are attributable to model, the current implementation in some model years. The full set of the standard. The approach currently further acknowledges that some assumptions regarding manufacturer only addresses the inherent trade-off manufacturers experience transitions behavior with respect to civil penalties between additional technology cost and between product lines where they rely is presented in Table–II–86, which the value of fuel savings, but other costs heavily on credits (either carried shows all manufacturers are assumed to could be relevant as well. Further forward from earlier model years or be willing to pay civil penalties prior to research would be required to support acquired from other manufacturers) or simulations that assume buyers behave MY 2020. This is largely a reflection of simply pay penalties in one or more either existing credit balances (which as if they consider all ownership costs fleets for some number of years. The (e.g., additional excise taxes and manufacturers will use to offset CAFE model now allows the user to specify, insurance costs) at the time of purchase deficits until the credits reach their when appropriate for the regulatory and that manufacturers respond expiration dates) or assumed trades program being simulated, on a year-by- accordingly. Comment is sought on the between manufacturers that are likely to approach described above, the current year basis, whether each manufacturer happen in the near-future based on values ascribed to manufacturers’ belief should be considered as willing to pay previous behavior. The manufacturers about consumer willingness-to-pay for penalties for non-compliance. This in the table whose names appear in bold fuel economy, and practicable provides additional flexibility, all had at least one regulated fleet (of suggestions for future improvements particularly in the early years of the three) whose CAFE was below its and refinements, considering the simulation. As discussed above, this standard in MY 2016. Because the model’s purpose and structure. assumption is best considered as a analysis began with the MY 2016 fleet, method to allow a manufacturer to and no technology can be added to (c) Representation of Some OEMs’ under-comply with its standard in some vehicles that are already designed and Willingness To Treat Civil Penalties as model years—treating the civil penalty built, all manufacturers can generate a Program Flexibility rate and payment option as a proxy for civil penalties in MY 2016. However, When considering technology other actions it may take that are not once a manufacturer is designated as applications to improve fleet fuel represented in the CAFE model (e.g., unwilling to pay penalties, the CAFE economy, the model will add purchasing credits from another model will attempt to add technology to technology up to the point at which the manufacturer, carry-back from future the respective fleets to avoid shortfalls.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00196 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.125 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43181

Several of the manufacturers in compliance deficits. Thus far, NHTSA challenging to simulate. This will be an Table–II–86 that are assumed to be has not attempted to include simulation area of focus moving forward. willing to pay civil penalties in the early of credit carry-back or trading in the NHTSA introduced the CAFE Public years of the program have no history of CAFE model. Unlike past versions, the Information Center 357 to provide public paying civil penalties. However, several current CAFE model provides a basis to access to a range of information of those manufacturers have either specify (in model inputs) CAFE credits regarding the CAFE program, including bought or sold credits—or transferred available from model years earlier than manufacturers’ credit balances. credits from one fleet to another to offset those being simulated explicitly. For However, there is a data lag in the a shortfall in the underperforming fleet. example, with this analysis representing information presented on the CAFE PIC As the CAFE model does not simulate model years 2016–2032 explicitly, that may not capture credit actions credit trades between manufacturers, credits earned in model year 2012 are across the industry for as much as providing this additional flexibility in made available for use through model several months. Additionally, CAFE the modeling avoids the outcome where year 2017 (given the current five-year credits that are traded between limit on carry-forward of credits). The the CAFE model applies more manufacturers are adjusted to preserve banked credits are specific to both technology than would be needed in the the gallons saved that each credit model year and fleet in which they were context of the full set of compliance represents.358 The adjustment occurs at earned. Comment and supporting flexibilities at the industry level. By the time of application rather than at the information are invited regarding statute, NHTSA cannot consider credit time the credits are traded. This means whether and, if so, how the CAFE model flexibilities when setting standards, so that a manufacturer who has acquired and inputs might practicably be most manufacturers (those without a credits through trade, but has not yet modified to account for trading of history of civil penalty payment) are applied them, may show a credit credits between manufacturers and/or assumed to comply with their standard balance that is either considerably through fuel economy improvements for carry-back of credits from later to earlier model years. higher or lower than the real value of the model years being considered in this the credits when they are applied. For analysis. The notable exception to this As discussed in the CAFE model documentation, the model’s default example, a manufacturer that buys 40 is FCA, who we expect will still satisfy million credits from Tesla, may show a the requirements of the program through logic attempts to maximize credit carry- forward—that is, to ‘‘hold on’’ to credits credit balance in excess of 40 million. a combination of credit application and However, when those credits are civil penalties through MY 2025 before for as long as possible. If a manufacturer needs to cover a shortfall that occurs applied, they may be worth only 1/10 as eventually complying exclusively much—making that manufacturer’s true through fuel economy improvements in when insufficient opportunities exist to add technology in order to achieve credit balance closer to 4 million than MY 2026. 40 million. As mentioned above, the CAA does compliance with a standard, the model not provide civil penalty provisions will apply credits. Otherwise it carries Having reviewed credit balances (as of similar to those specified in EPCA/ forward credits until they are about to October 23, 2017) and estimated the EISA, and the above-mentioned expire, at which point it will use them potential that some manufacturers could corresponding inputs apply only to before adding technology that is not trade credits, NHTSA developed inputs simulation of compliance with CAFE considered cost-effective. The model that make carried-forward credits standards. attempts to use credits that will expire available as summarized in Table–II–87, within the next three years as a means Table–II–88, and Table–II–89, after (d) Representation of CAFE and CO2 to smooth out technology application subtracting credits assumed to be traded Credit Provisions over time to avoid both compliance to other manufacturers, adding credits The model’s approach to simulating shortfalls and high levels of over- assumed to be acquired from other compliance decisions accounts for the compliance that can result in a surplus manufacturers through such trades, and potential to earn and use CAFE credits of credits. As further discussed in the adjusting any traded credits (up or as provided by EPCA/EISA. The model CAFE model documentation, model down) to reflect their true value for the similarly accumulates and applies CO2 inputs can be used to adjust this logic fleet and model year into which they credits when simulating compliance to shift the use of credits ahead by one were traded.359 While the CAFE model with EPA’s standards. Like past or more model years. In general, the will transfer expiring credits into versions, the current CAFE model can logic used to generate credits and apply another fleet (e.g., moving expiring be used to simulate credit carry-forward them to compensate for compliance credits from the domestic car credit (a.k.a. banking) between model years shortfalls, both in a given fleet and bank into the light truck fleet), some of and transfers between the passenger car across regulatory fleets, is an area that these credits were moved in the initial and light truck fleets but not credit requires more attention in the next banks to improve the efficiency of carry-back (a.k.a. borrowing) from future phase of model development. While the application and to better reflect both the model years or trading between current model correctly accounts for projected shortfalls of each manufacturers. Some manufacturers credits earned when a manufacturer manufacturer’s regulated fleets, and to have made occasional use of credit exceeds its standard in a given year, the represent observed behavior. For carry-back provisions, although the strategic decision of whether to earn context, a manufacturer that produces analysis does not assume use of carry- additional credits to bank for future one million vehicles in a given fleet, back as a compliance strategy because of years (in the current fleet or to transfer and experiences a shortfall of 2 mpg, the risk in relying on future into another regulatory fleet) and when would need 20 million credits to improvements to offset earlier to optimally apply them to deficits is completely offset the shortfall.

357 CAFE Public Information Center, http:// adjustment when traded between manufacturers or would be applied to the model year in which they www.nhtsa.gov/CAFE_PIC/CAFE_PIC_Home.htm fleets. were set to expire. For example, credits traded into (last visited June 22, 2018). 359 The adjustments, which are based upon the a domestic passenger car fleet for MY 2014 were 358 GHG credits for EPA’s program are standard, CAFE and year of both the party adjusted assuming they would be applied in the denominated in metric tons of CO2 rather than originally earning the credits and the party applying domestic passenger car fleet for MY 2019. gram/mile compliance credits and require no them, were implemented assuming the credits

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00197 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43182 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00198 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.126 EP24AU18.127 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43183

In addition to the inclusion of these but to examine the impact of proposed was not modified to allow exceptions to existing credit banks, the CAFE model standards based solely on fuel economy the life-span of compliance credits also updated its treatment of credits in improvements in all years for which treating them all as if they may be the rulemaking analysis. Congress has new standards are being considered. carried forward for no more than five declared that NHTSA set CAFE Comment is sought regarding the years, so the initial credit banks were standards at maximum feasible levels model’s representation of the CAFE and modified to anticipate the years in for each model year under consideration CO2 credit provisions, recommendations which those credits might be needed. without consideration of the program’s regarding any other options, and any The fact that MY 2016 is simulated credit mechanisms. However, as CAFE information that could help to refine the explicitly prohibited the inclusion of current approach or develop and rulemakings have evaluated longer time these banked credits in MY 2016 (which implement an alternative approach. periods in recent years, the early actions The CAFE model has also been could be carried forward from MY 2016 taken by manufacturers required more modified to include a similar to MY 2021), and thus underestimates nuanced representation. Therefore, the representation of existing credit banks the extent to which individual CAFE model now allows a ‘‘last year to in EPA’s CO2 program. While the life of manufacturers, and the industry as a consider credits,’’ set at the last year for a CO2 credit, denominated in metric whole, may rely on these early credits which new standards are not being tons CO2, has a five-year life, matching to comply with EPA standards between considered (MY 2019 in this analysis). the lifespan of CAFE credits, credits MY 2016 and MY 2021. The credit This allows the model to replicate the earned in the early years of the EPA banks with which the simulations in practical application of existing credits program, MY 2009–2011, may be used this analysis were conducted are toward CAFE compliance in early years through MY 2021.360 The CAFE model presented in the following tables:

360 In response to comments, EPA placed limits expire prior to this rule. However, credits generated traded, and applied to deficits generated through on credits earned in MY 2009, causing them to in MYs 2010–2011 may be carried forward, or MY 2021.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00199 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.128 43184 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

While the CAFE model does not the strategic accumulation and between fleets to improve the simulate the ability to trade credits application of compliance credits, as compliance position of a less efficient between manufacturers, it does simulate well as the ability to transfer credits fleet by leveraging credits earned by a

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00200 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.129 EP24AU18.130 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43185

more efficient fleet. The model prefers phase out in MY 2019), they may be targets (by volume) within a fleet, just to hold on to earned compliance credits included under either a ‘‘standard as the standards prescribe. Unlike within a given fleet, carrying them setting’’ or ‘‘unconstrained’’ analysis earlier versions of the CAFE model, the forward into the future to offset perspective. current version further disaggregates potential future deficits. This When the CAFE model simulates passenger cars into domestic and assumption is consistent with observed EPA’s program, the treatment of A/C imported classes (which manufacturers strategic behavior dating back to 2009. efficiency and off-cycle credits is report to NHTSA and EPA as part of From 2009 to present, no similar, but the model also accounts for their CAFE compliance submissions). manufacturer has transferred CAFE A/C leakage (which is not part of This allows the CAFE model to more credits into a fleet to offset a deficit in NHTSA’s program). When determining accurately estimate the requirement on the same year in which they were the compliance status of a the two passenger car fleets, represent earned. This has occurred with credits manufacturer’s fleet (in the case of the domestic passenger car floor (which acquired from other manufacturers via EPA’s program, PC and LT are the only must be exceeded by every trade but not with a manufacturer’s own fleet distinctions), the CAFE model manufacturer’s domestic fleet, without credits. Therefore, the current weighs future compliance actions the use of credits, but with the representation of credit transfers against the presence of existing (and possibility of civil penalty payment), between fleets—where the model expiring) CO2 credits resulting from and allows it to enforce the transfer cap prefers to transfer expiring, or soon-to- over-compliance with earlier years’ limit that exists between domestic and be-expiring credits rather than newly standards, A/C efficiency credits, A/C imported passenger cars, all for earned credits—is both appropriate and leakage credits, and off-cycle credits. purposes of the CAFE program. consistent with observed industry In calculating the achieved CAFE behavior. 5. Impacts on Each OEM and Overall Industry level, the model uses the prescribed This may not be the case for GHG harmonic average of fuel economy standards, though it is difficult to be (a) Technology Application and ratings within a vehicle fleet. Under an certain at this point. The GHG program Penetration Rates ‘‘unconstrained’’ analysis, or in a model seeded the industry with a large The CAFE model tracks and reports year for which standards are already quantity of early compliance credits final, it is possible for a manufacturer’s 361 technology application and penetration (earned in MYs 2009–2011 ) prior to rates for each manufacturer, regulatory CAFE to fall below its required level the existence formal standards of the class, and model year, calculated as the without generating penalties because EPA program. These early credits do not volume of vehicles with a given the model will apply expiring or expire until 2021. So, for manufacturers technology divided by the total volume. transferred credits to deficits if it is looking to offset deficits, it is more The ‘‘application rate’’ accounts only for strategically appropriate to do so. sensible to use current-year credits that those technologies applied by the model Consistent with current EPA expire in the next five years, rather than during the compliance simulation, regulations, the model applies simple draw down the bank of credits that can while the ‘‘penetration rate’’ accounts (not harmonic) production-weighted be used until MY 2021. The first model for the total percentage of a technology averaging to calculate average CO2 year for which earned credits outlive the present in a given fleet, whether applied levels. initial bank is MY 2017, for which final by the CAFE model or already present (c) Costs compliance actions and deficit at the start of the simulation. resolutions are still pending. Regardless, In addition to the aggregate For each technology that the model in order to accurately represent some of representation of technology adds to a given vehicle, it accumulates the observed behavior in the GHG credit penetration, the model also tracks each cost. The technology costs are defined system, the CAFE model allows (and individual vehicle model on which it incrementally and vary both over time encourages) within-year transfers has operated. Each row in the market and by technology class, where the same between regulated fleets for the purpose data file (the representation of vehicles technology may cost more to apply to of simulating compliance with the GHG offered for sale in MY 2016 in the U.S., larger vehicles as it involves more raw standards. discussed in detail in Section II.B.a and materials or requires different In addition to more rigorous PRIA Chapter 6) contains a record for specifications to preserve some accounting of CAFE and CO2 credits, the every model year and every alternative, performance attributes. While learning- model now also accounts for air that identifies with which technologies by-doing can bring down cost, and conditioning efficiency and off-cycle the vehicle started the simulation, should reasonably be implemented in adjustments. NHTSA’s program which technologies were applied, and the CAFE model as a rate of cost considers those adjustments in a whether those technologies were reduction that is applied to the manufacturer’s compliance calculation applied directly or through inheritance cumulative volume of a given starting in MY 2017, and the current (discussed above). Interested parties technology produced by either a single model uses the adjustments claimed by may use these outputs to assess how the manufacturer or the industry as a whole, each manufacturer in MY 2016 as the compliance simulation modified any in practice this notion is implemented starting point for all future years. vehicle that was offered for sale in MY as a function of time, rather than Because the air conditioning and off- 2016 in response to a given regulatory production volume. Thus, depending cycle adjustments are not credits in alternative. upon where a given technology starts NHTSA’s program, but rather along its learning curve, it may appear adjustments to compliance fuel (b) Required and Achieved CAFE and to be cost-effective in later years where economy (much like the Flexible Fuel Average CO2 Levels it was not in earlier years. As the model Vehicle adjustments that are due to The model fully represents the carries forward technologies that it has required CAFE (and now, CO2) levels for already applied to future model years, it 361 In response to public comment, EPA eliminated the use of credits earned in MY 2009 for every manufacturer and every fleet. The similarly adjusts the costs of those future model years. However, credits earned in MY standard for each manufacturer is based technologies based on their individual 2010 and MY 2011 remain. on the harmonic average of footprint learning rates.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00201 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43186 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The other costs that manufacturers (regardless of regulatory class or body compliance value in each model year incur as a result of CAFE standards are style), in response to certain (and class). Once the simulation begins, civil penalties resulting from non- macroeconomic inputs and changes in new vehicles are added to the compliance with CAFE standards. The the average new vehicle price. The price population from the market data file and CAFE model accumulates costs of $5.50 impact is modest relative to the age throughout their useful lives during per 1/10–MPG under the standard, influence of the macroeconomic factors the simulation, with some fraction of multiplied by the number of vehicles in the model. The combination of these them being retired (or scrapped) along produced in that fleet, in that model two models modify the total number of the way. For example, in calendar year year. The model reports as the full new vehicles, the share of passenger 2017, the new vehicles (age zero) are ‘‘regulatory cost,’’ the sum of total cars and light trucks, and, as a MY 2017 vehicles (added by the CAFE technology cost and total fines by the consequence, the number of each given model simulation and represented at the manufacturer, fleet, and model year. As model sold by a given manufacturer. same level of detail used to simulate mentioned above, the relevant EPCA/ However, these two factors are compliance), the age one vehicles are EISA provisions do not also appear in insufficient to cause large changes to the MY 2016 vehicles (added by the CAFE the CAA, so this option and these costs composition of any of a manufacturer’s model simulation), and the age two apply only to simulated compliance fleets. In order to significantly change vehicles are MY 2015 vehicles with CAFE standards. the mix of models produced within a (inherited from the registered vehicle population and carried through the (d) Sales given fleet, the CAFE model would require a way to trade off the production analysis with less granularity). This In all previous versions of the CAFE of one vehicle versus another both national registered fleet is used to model, the total number of vehicles sold within a manufacturer’s fleet and across calculate annual fuel consumption, in any model year, in fact the number the industry. While NHTSA has vehicle miles traveled (VMT), pollutant of each individual vehicle model sold in experimented with fully-integrated emissions, and safety impacts under each year, has been a static input that consumer choice models, their each regulatory alternative. did not vary in response to price performance has yet to satisfy the In addition to dynamically modifying increases induced by CAFE standards, requirements of a rulemaking analysis. the total number of new vehicles sold, nor changes in fuel prices, or any other There are multiple levels of sales a dynamic model of vehicle retirement, input to the model. The only way to impacts that could result from or scrappage, has also been alter sales, was to update the entire increasing the prices of new vehicles implemented. The model implements forecast in the market input file. across the industry. Any estimate of the scrappage response by defining the However, in the 2012 final rule, NHTSA impacts at the manufacturer, or model, instantaneous scrappage rate at any age included a dynamic fleet share model level would be subject to an assumed using two functions. For ages less than that was based on a module in the pricing strategy that spreads technology 20, instantaneous scrappage is defined Energy Information Administration’s cost increases across available models in as a function of vehicle age, new vehicle NEMS model. This fleet share model a way that may cross-subsidize specific price, cost per mile of driving (the ratio did not change the size of the new models or segments at the expense of of fuel price and fuel economy), and a vehicle fleet in any year, but it did others. However, at the industry level, it small number of macroeconomic factors. change the share of new vehicles that is reasonable to assume that all For ages greater than 20, the were classified as passenger cars (or incremental technology costs can be instantaneous scrappage rate is a simple light trucks). That capability was not captured by the average price of a new exponential function of age. While the included in the central analysis but was vehicle. To the extent that this factor scrappage response does not affect included in the uncertainty analysis, influences the total number of new manufacturer compliance calculations, which looked at the baseline and vehicles sold in a given model year, it it impacts the lifetime mileage preferred alternative in the context of accumulation (and thus fuel savings) of can be included in an empirical model thousands of possible future states of all vehicles. Previous CAFE analyses of annual sales. However, there is the world. As some of those futures have focused exclusively on new limited historical evidence that the contained extreme cases of fuel prices, vehicles, tracing the fuel consumption average price of a new vehicle is a it was important to ensure consistent and social costs of these vehicles strong determining factor in the total modeling responses within that context. throughout their useful lives; the number of annual new vehicle sales. For example, at a gasoline price of $7/ scrappage effect also impacts the gallon, it would be unrealistic to expect 6. National Impacts registered vehicle fleet that exists when the new vehicle market’s light truck (a) Vehicle Stock and Fleet Turnover a set of standards is implemented. share to be the same as the future where As new vehicles enter the registered gasoline cost $2/gallon. The current The CAFE model carries a complete population their retirement rates are model has slightly modified, and fully representation of the registered vehicle governed by the scrappage model, so are integrated, the dynamic fleet share population in each calendar year, the vehicles already registered at the model. Every regulatory alternative and starting with an aggregated version of start of model year 2016. To the extent sensitivity case considered in this the most recent available data about the that a given set of CAFE or CO2 analysis reflects a dynamically registered population for the first year of standards accelerates or decelerates the responsive fleet mix in the new vehicle the simulation. In this analysis, the first retirement of those vehicles, additional market. model year considered is MY 2016, and fuel consumption and social costs may While the dynamic fleet share model the registered vehicle population enters accrue to those vehicles under that adjusts unit sales across body styles the model as it appeared at the end of standard. The CAFE model accounts for (cars, SUVs, and trucks), it does not calendar year 2015. The initial vehicle those costs and benefits, as well as modify the total number of new vehicles population is stratified by age (or model tracking all of the standard benefits and sold in a given year. The CAFE model year cohort) and regulatory class—to costs associated with the lifetimes of now includes a separate function to which the CAFE model assigns average new vehicles produced under the rule. account for changes in the total number fuel economies based on the reported For more detail about the derivation of of new vehicles sold in a given year regulatory class industry average the scrappage functions, see Section

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00202 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43187

II.E, and PRIA Chapter 8. Comment is (c) Fuel Consumption and GHG Federal fuel excise tax is levied on every sought on the specification and Emissions gallon of gasoline and diesel sold in the inclusion of these factors in the current For every vehicle model in the market U.S., with diesel facing a higher per- model. file, the model estimates the VMT per gallon tax rate. The model uses a national perspective, where the state (b) Highway Travel vehicle (using the assumed VMT schedule, the vehicle fuel economy, fuel taxes present in the input files represent an estimated average fuel tax across all In support of prior CAFE rulemakings, price, and the rebound assumption). U.S. states. Accordingly, while the the CAFE model accounted for new Those miles are multiplied by the CAFE model cannot reasonably estimate travel that results from fuel economy volume for each vehicle. Fuel consumption is the product of miles potential losses to state fuel tax revenue improvements that reduce the cost of from increasingly the fuel economy of driven and fuel economy, which can be driving. The magnitude of the increase new vehicles, it can do so for the HTF, tracked by model year cohort in the in travel demand is determined by the and the agencies invite comment on the model. Carbon dioxide emissions from rebound effect. In both previous proposed standards’ implications for the vehicle tailpipes are the simple product versions and the current version of the HTF. CAFE model, the amount of travel of gallons consumed and the carbon In addition to the tailpipe emissions demanded by the existing fleet of content of each gallon. of carbon dioxide, each gallon of In order to calculate calendar year vehicles is also responsive to the gasoline produced for consumption by rebound effect (representing the price fuel consumption, the model needs to the on-road fleet has associated account for the inherited on-road fleet elasticity of demand for travel)— ‘‘upstream’’ emissions that occur in the in addition to the model year cohorts increasing when fuel prices decrease extraction, transportation, refining, and affected by this proposed rule. Using the relative to the fuel price when the VMT distribution of the fuel. The model VMT of the average passenger car and accounts for these emissions as well (on on which our mileage accumulation light truck from each cohort, the model schedules were built was observed. a per-gallon basis) and reports them computes the fuel consumption of each accordingly. Since the fuel economy of those model year class of vehicles for its age vehicles is already fixed, only the fuel in a given CY. The sum across all ages (d) Criteria Pollutant Emissions price influences their travel demand (and thus, model year cohorts) in a The CAFE model uses the entire on- relative to the mileage accumulation given CY provides estimated CY fuel road fleet, calculated VMT (discussed schedule and so is identical for all consumption. above), and emissions factors (which are regulatory alternatives. Rather than rely on the compliance an input to the CAFE model, specified While the average mileage values of fuel economy for either by model year and age) to calculate accumulation per vehicle by age is not historical vehicles or vehicles that go tailpipe emissions associated with a influenced by the rebound effect in a through the full compliance simulation, given alternative. Just as it does for way that differs by regulatory the model applies an ‘‘on-road gap’’ to additional GHG emissions associated alternative, three other factors influence represent the expected difference with upstream emissions from fuel total VMT in the model in a way that between fuel economy on the laboratory production, the model captures criteria produces different total mileage test cycle and fuel economy under real- pollutants that occur during other parts accumulation by regulatory alternative. world operation. This was a topic of of the fuel life cycle. While this is The first factor is the total industry sales interest in the recent peer review of the typically a function of the number of response: New vehicles are both driven CAFE model. While the model currently gallons of gasoline consumed (and miles more than older vehicles and are more allows the user to specify an on-road driven, for tailpipe criteria pollutant gap that varies by fuel type (gasoline, fuel efficient (thus producing more emissions), the CAFE model also E85, diesel, electricity, hydrogen, and rebound miles). To the extent that more estimates electricity consumption and CNG), it does not vary over time, by (or fewer) of these new models enter the the associated upstream emissions vehicle age, or by technology vehicle fleet in each model year, total (resource extraction and generation, combination. It is possible that the VMT will increase (or decrease) as a based on U.S. grid mix). ‘‘gap’’ between laboratory fuel economy result. The second factor is the dynamic and real-world fuel economy has (e) Highway Fatalities fleet share model. The fleet share changed over time, that fuel economy Earlier versions of the CAFE model influences not only the fuel economy degrades over time as a vehicle ages, or accounted for the safety impacts distribution of the fleet, as light trucks that specific combinations of fuel-saving associated with reducing vehicle mass are less efficient than passenger cars on technologies have a larger discrepancy in order to improve fuel economy. In average, but the total miles are between laboratory and real-world fuel particular, NHTSA’s safety analysis influenced by fact that light trucks are economy than others. Further research estimated the additional fatalities that driven more than passenger cars as well. would be required to determine whether would occur as a result of new vehicles Both of the first two factors can magnify the model should include a functional getting lighter, then interacting with the the influence of the rebound effect on representation of the on-road gap to on-road vehicle population. In general, vehicles that go through the compliance address these various factors, and taking mass out of the heaviest new simulation (MY 2016–2032) in the comment is sought on the data sources vehicles improved safety outcomes, manner discussed above and in Section and implementation strategies available while taking mass from the lightest new II.E. The third factor influencing total to do so. vehicles resulted in a greater number of annual VMT is the scrappage model. By Because the model produces an expected highway fatalities. However, modifying the retirement rates of on- estimate of the aggregate number of the change in fatalities did not road vehicles under each regulatory gallons sold in each CY, it is possible to adequately account for changes in alternative, the scrappage model either calculate both the total expenditures on exposure that occur as a result of increases or decreases the lifetime miles motor fuel and the total contribution to increased demand for travel as vehicles that accrue to vehicles in a given model the Highway Trust Fund (HTF) that become cheaper to operate. The current year cohort. result from that fuel consumption. The version of the model resolves that

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00203 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43188 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

limitation and addresses additional accelerates) the retirement of registered independent of any changes to new sources of fatalities that can result from vehicles, impacts the number of vehicle mass, price, or longevity. the implementation of CAFE or CO2 fatalities through this mechanism— The CAFE model now estimates standards. These are discussed in importantly affecting not just new fatalities by combining the effects greater detail in Section 0 and PRIA vehicles sold from model years 2016– discussed above. In particular, the Chapter 11. 2032 but existing vehicles that are model estimates the fatality rate per NHTSA has observed that older already part of the on-road fleet. vehicles in the population are Similarly, to the extent that a CAFE or billion miles VMT for each model year vehicle in the population (the newest of responsible for a disproportionate CO2 alternative reduces new vehicle number of fatalities, both by number of sales, it can slow the transition from which are the new vehicles produced registrations and by number of miles older vehicles to newer vehicles, that model year). This estimate is driven. Accordingly, any factor that reducing the share of total vehicle miles independent of regulatory class and causes the population of vehicles to turn that are driven by newer, more varies only by year (and not vehicle over more slowly will induce additional technologically advanced vehicles. age). The estimated fatality rate is then fatalities—as those older vehicles Accounting for the change in vehicle multiplied by the estimated VMT for continue to be driven, rather than being miles traveled that occurs when each vehicle in the population and the retired and replaced with newer (even if vehicles become cheaper to operate has product of the change in curb weight not brand new) vehicle models. The led to a number of fatalities that can be and the relevant safety coefficient, as in scrappage effect, which delays (or attributed to the rebound effect, the equation below.

For the vehicles in the historical fleet, generate social costs. The most obvious incremental fuel expenditures are meaning all those vehicles that are cost associated with the program is the greater for the alternatives than for the already part of the registered vehicle cost of additional fuel economy baseline. population in CY 2016, only the model improving/CO2 emissions reducing Other social costs and benefits emerge year effect that determines the technology that is added to new as the result of physical phenomena, ‘‘FatalityEstimate’’ is relevant. However, vehicles as a result of the rule. However, like tailpipe emissions or highway each vehicle that is simulated explicitly the model does not inherently draw a fatalities, which are the result of by the CAFE model, and is eligible to distinction between costs and benefits. changes in the composition and use of receive mass reduction technologies, For example, the model tracks fuel the on-road fleet. The social costs must also consider the change between consumption and the dollar value of associated with those quantities its curb weight and the threshold fuel consumed. This is the cost of travel represent an economic estimate of the weights that are used to define safety under a given alternative (including the social damages associated with the classes. For vehicles above the baseline). The ‘‘cost’’ or ‘‘benefit’’ changes in each quantity. The model threshold, reducing vehicle mass can associated with the value of fuel tracks and reports each of these have a smaller negative impact on consumed is determined by the quantities by: Model year and vehicle fatalities (or even reduce fatalities, in reference point against which each age (the combination of which can be the case of the heaviest light trucks). alternative is considered. The CAFE used to produce calendar year totals), The ‘‘ChangePer100Lbs’’ depends upon regulatory class, fuel type, and social this difference. The sum of all estimated model reports absolute values for the discount rate. fatalities for each model year vehicle in amount of money spent on fuel in the the on-road fleet determines the baseline, then reports the amount spent The full list of potential costs and reported fatalities, which can be on fuel in the alternatives relative to the benefits is presented in Table–II–92 as summarized by either model year or baseline. If the baseline standard were well as the population of vehicles that calendar year. fixed at the current level, and an determines the size of the factor (either alternative achieves 100 mpg by 2025, new vehicles or all registered vehicles) (f) Costs and Benefits the total expenditures on fuel in the and the mechanism that determines the As the CAFE model simulates alternative would be lower, creating a size of the effect (whether driven by the manufacturer compliance with fuel savings ‘‘benefit.’’ This analysis number of miles driven, the number of regulatory alternatives, it estimates and uses a baseline that is more stringent gallons consumed, or the number of tracks a number of consequences that than each alternative considered, so the vehicles produced).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00204 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.131 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43189

III. Proposed CAFE and CO2 Standards For MYs since 2011 for CAFE and and higher CO2 grams/mile targets than for MYs 2021–2026 since 2012 for CO2, standards have smaller vehicles. This is because, taken the form of fuel economy and CO2 generally speaking, smaller vehicles are A. Form of the Standards targets expressed as functions of vehicle more capable of achieving higher levels NHTSA and EPA are proposing that footprint (the product of vehicle of fuel economy/lower levels of CO2 wheelbase and average track width). emissions, mostly because they tend not the form of the CAFE and CO2 standards for MYs 2021–2026 would follow the NHTSA and EPA continue to believe to have to work as hard to perform their that footprint is the most appropriate driving task. Although a manufacturer’s form of those standards in prior model attribute on which to base the proposed fleet average standards could be years. NHTSA has specific statutory standards, as discussed in Section II.C. estimated throughout the model year requirements for the form of CAFE Under the footprint-based standards, the based on the projected production standards: Specifically, EPCA, as function defines a CO2 or fuel economy volume of its vehicle fleet (and are amended by EISA, requires that CAFE performance target for each unique estimated as part of EPA’s certification standards be issued separately for footprint combination within a car or process), the standards to which the passenger cars and light trucks, and that truck model type. Using the functions, manufacturer must comply will be each standard be specified as a each manufacturer thus will have a determined by its final model year mathematical function expressed in CAFE and CO2 average standard for production figures. A manufacturer’s terms of one or more vehicle attributes each year that is unique to each of its calculation of its fleet average standards related to fuel economy. Although the fleets,363 depending on the footprints as well as its fleets’ average performance CAA does not have comparable specific and production volumes of the vehicle at the end of the model year will thus requirements for the form of CO2 models produced by that manufacturer. be based on the production-weighted standards for light-duty vehicles, EPA A manufacturer will have separate average target and performance of each has concluded that it is appropriate to footprint-based standards for cars and model in its fleet.364 set CO2 standards according to vehicle for trucks. The functions are mostly For passenger cars, consistent with footprint, consistent with the EPCA/ sloped, so that generally, larger vehicles prior rulemakings, NHTSA is proposing EISA requirements, which simplifies (i.e., vehicles with larger footprints) will to define fuel economy targets as compliance for the industry.362 be subject to lower CAFE mpg targets follows:

362 Such an approach is permissible under section car fleets whereas EPA combines all passenger cars comparing the fleet average standard (based on the 202(a) of the CAA and EPA has used the attribute- into one fleet. production-weighted average of the target levels for based approach in issuing standards under 364 As in prior rulemakings, a manufacturer may each model) with fleet average performance (based analogous provisions of the CAA. have some vehicle models that exceed their target on the production-weighted average of the 363 EPCA/EISA requires NHTSA to separate and some that are below their target. Compliance performance of each model). passenger cars into domestic and import passenger with a fleet average standard is determined by

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00205 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.132 43190 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Where: c is the slope (in gallons per mile per square included values. For example, foot, or gpm, per square foot) of a line TARGETFE is the fuel economy target (in MIN[40,35] = 35 and MAX(40, 25) = 40, mpg) applicable to a specific vehicle relating fuel consumption (the inverse of such that MIN[MAX(40, 25), 35] = 35. fuel economy) to footprint, and model type with a unique footprint d is an intercept (in gpm) of the same line. For light trucks, also consistent with combination, prior rulemakings, NHTSA is proposing a is a minimum fuel economy target (in mpg), Here, MIN and MAX are functions b is a maximum fuel economy target (in that take the minimum and maximum to define fuel economy targets as mpg), values, respectively, of the set of follows:

Where: functions that are similar, with a, b, c, and d are as for passenger cars, but TARGETFE is the fuel economy target (in coefficients a–h corresponding to those taking values specific to light trucks, mpg) applicable to a specific vehicle listed above.365 For passenger cars, EPA e is a second minimum CO2 target (in g/mi), model type with a unique footprint is proposing to define CO2 targets as f is a second maximum CO2 target (in g/mi), combination, follows: g is the slope (in g/mi per square foot) of a a, b, c, and d are as for passenger cars, but second line relating CO2 emissions to TARGETCO2 = MIN[b,MAX[a,c × taking values specific to light trucks, footprint, and e is a second minimum fuel economy target FOOTPRINT + d]] h is an intercept (in g/mi) of the same second (in mpg), Where: line. f is a second maximum fuel economy target TARGETCO2 is the CO2 target (in grams per (in mpg), mile, or g/mi) applicable to a specific To be clear, as has been the case since g is the slope (in gpm per square foot) of a vehicle model configuration, second line relating fuel consumption the agencies began establishing a is a minimum CO2 target (in g/mi), (the inverse of fuel economy) to attribute-based standards, no vehicle b is a maximum CO2 target (in g/mi), footprint, and c is the slope (in g/mi, per square foot) of a need meet the specific applicable fuel h is an intercept (in gpm) of the same second line relating CO emissions to footprint, economy or CO2 targets, because line. 2 and compliance with either CAFE or CO2 Although the general model of the d is an intercept (in g/mi) of the same line. standards is determined based on target function equation is the same for For light trucks, CO targets are corporate average fuel economy or fleet each vehicle category (passenger cars 2 defined as follows: average CO2 emission rates. The and light trucks) and each model year, TARGETCO2 = MIN[MIN[b, MAX[a,c × required CAFE level applicable to a the parameters of the function equation FOOTPRINT + d]], MIN[f,MAX[e, g given fleet in a given model year is differ for cars and trucks. For MYs × FOOTPRINT + H]] determined by calculating the 2020–2026, the parameters are production-weighted harmonic average unchanged, resulting in the same Where: of fuel economy targets applicable to stringency in each of those model years. TARGETCO2 is the CO2 target (in g/mi) specific vehicle model configurations in Mathematical functions defining the applicable to a specific vehicle model the fleet, as follows: proposed CO2 targets are expressed as configuration,

Where: PRODUCTIONi is the number of model Similarly, the required average CO2 CAFErequired is the CAFE level the fleet is configuration i produced for sale in the level applicable to a given fleet in a required to achieve, U.S., and given model year is determined by i refers to specific vehicle model/ TARGETFE,i the fuel economy target (as calculating the production-weighted configurations in the fleet, defined above) for model configuration i.

365 EPA regulations use a different but specify requirements separately for different ranges equivalent and more efficient to present the targets mathematically equivalent approach to specify of vehicle footprint. Because these ranges reflect the as in this Section. targets. Rather than using a function with nested combined application of the listed minima, minima and maxima functions, EPA regulations maxima, and linear functions, it is mathematically

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00206 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.133 EP24AU18.134 EP24AU18.135 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43191

average (not harmonic) of CO2 targets applicable to specific vehicle model configurations in the fleet, as follows:

Where: Today’s action would set standards one of the other regulatory alternatives CO2required is the average CO2 level the fleet that only apply to fuel economy and for the final rule. is required to achieve, CO . EPA seeks comment on this 2 B. Passenger Car Standards i refers to specific vehicle model/ approach. configurations in the fleet, Comment is sought on the proposed For passenger cars, NHTSA and EPA PRODUCTIONi is the number of model configuration i produced for sale in the standards and on the analysis presented are proposing CAFE and CO2 standards, U.S., and here; we seek any relevant data and respectively, for MYs 2021–2026 that TARGETCO2,i is the CO2 target (as defined information and will review responses. are defined by the following above) for model configuration i. That review could lead to selection of coefficients:

Section II.C above discusses in detail coefficients result in the footprint- 2017–2020 standards are also shown for how the coefficients in Table III–1 were dependent targets shown graphically comparison. developed for this proposal. The below for MYs 2021–2026. The MYs

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00207 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.136 EP24AU18.137 43192 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00208 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.138 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43193

While we do not know yet with years 2021 and beyond do not reflect ranging from three grams per mile certainty what CAFE and CO2 levels these adjustments. equivalent for Hyundai and Kia, to 17 will ultimately be required of individual EPA seeks comments on whether to grams per mile equivalent for Jaguar manufacturers, because those levels will proceed with this proposal to Land Rover.367 As related to methane depend on the mix of vehicles that they discontinue accounting for A/C leakage, (CH4) and nitrous oxide (N2O) produce for sale in future model years, methane emissions, and nitrous oxide emissions, manufacturers averaged 0.1 based on the market forecast of future emissions as part of the CO2 emissions grams per mile equivalent in deficits for sales that was used to examine today’s standards to provide for better harmony the 2016 MY, with deficits ranging from with the CAFE program, or whether to proposed standards, we currently 0.1 grams per mile equivalent for GM, continue to consider these factors estimate that the target functions shown Mazda, and Toyota, to 0.6 grams per toward compliance and retain that as a mile equivalent for Nissan.368 above would result in the following feature that differs between the average required fuel economy and CO EPA notes that since the 2010 2 programs. A/C leakage credits, which rulemaking on this subject, the agencies emissions levels for individual are accounted for in the baseline model, have accounted for the ability to apply manufacturers during MYs 2021–2026. have been extensively generated by A/C leakage credits by increasing EPA’s Prior to MY 2021, average required CO2 manufacturers, and make up a portion CO2 standard stringency by the average levels reflect underlying target functions of their compliance with EPA’s CO2 anticipated amount of credits when (specified above) that reflect the use of standards. In the 2016 MY, compared to the CAFE stringency automotive refrigerants with reduced manufacturers averaged six grams per requirements.369 For model years 2021– global warming potential (GWP) and/or mile equivalent in A/C leakage credits, 2025, the A/C leakage offset, or the use of technologies that reduce the 366 Prior to MY 2021, CO targets include refrigerant leaks. EPA is proposing to 2 367 Other manufacturers’ A/C leakage credit grams adjustments reflecting the use of automotive per mile equivalent include: BMW, Honda, exclude air conditioning refrigerants refrigerants with reduced global warming potential Mistubishi, Nissan, Toyota, and Volkswagen at 5 g/ and leakage, and nitrous oxide and (GWP) and/or the use of technologies that reduce mi; Mercedes at 6 g/mi; Ford, GM, and Volvo at 7 methane GHGs from average the refrigerant leaks and optionally nitrous oxide and methane emissions. EPA is proposing to g/mi; and FCA at 14 g/mi. performance calculations after model exclude air conditioning refrigerants and leakage, 368 Other manufacturers’ methane and nitrous year 2020; CO2 targets and resultant and nitrous oxide and methane GHGs from average oxide deficit grams per mile equivalent include fleet average requirements for model performance calculations after model year 2020; BMW at 0.2 g/mi, and Ford at 0.3 g/mi. FCA and CO2 targets (and resultant fleet average Volkswagen numbers are not reported due to an requirements) for model years 2021 and beyond do ongoing investigation and/or corrective actions. not reflect these adjustments. 369 75 FR 25330, May 7, 2010.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00209 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.139 43194 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

equivalent stringency increase basis, folding in all N2O and CH4 to regulate these programs compared to the CAFE standard, is 13.8 emissions could add up to 3–4 g/mile to independently, which could include an g/mi equivalent for passenger cars and a manufacturer’s overall CO2 emissions effective date that would result in no 17.2 g/mi equivalent for light trucks.370 level because the equivalent standard lapse in regulation of A/C leakage or For those model years, manufacturers must be used for the entire fleet, not just emissions of nitrous oxide and methane. are currently allowed to apply A/C for ‘‘problem vehicles.’’ 374 To address If the agency decides to retain the A/C leakage credits capped at 18.8 g/mi this added difficulty, EPA amended the leakage and nitrous oxide and methane equivalent for passenger cars and 24.4 g/ MY 2012–2016 standards to allow emissions provisions for CO2 371 mi equivalent for light trucks. manufacturers to use CO2 credits, on a compliance, it would likely re-insert the For methane and nitrous oxide CO -equivalent basis, to meet the light- 2 current A/C leakage offset and increase emissions, as part of the MY 2012–2016 duty N O and CH standards in those 2 4 the stringency levels for CO rulemaking, EPA finalized standards to model years. EPA subsequently 2 compliance by the offset amounts cap emissions of N2O at 0.010 g/mile extended that same credit provision to described above (i.e., 13.8 g/mi and CH4 at 0.030 g/mile for MY 2012 MY 2017 and later vehicles. EPA seeks and later vehicles.372 However, EPA comment on whether to change existing equivalent for passenger cars and 17.2 g/ also provided an optional CO2- methane and nitrous oxide standards mi equivalent for light trucks), and equivalent approach to address industry that were finalized in the 2012 rule. retain the current caps (i.e., 18.8 g/mi concerns about technological feasibility Specifically, EPA seeks information equivalent for passenger cars and 24.4 g/ and leadtime for the CH4 and N2O from the public on whether the existing mi equivalent for light trucks). The standards for MY 2012–2016 vehicles. standards are appropriate, or whether agency will publish an analysis of this The CO2 equivalent standard option they should be revised to be less alternative approach in a memo to the allowed manufacturers to fold all 2- stringent or more stringent based on any docket for this rulemaking. The agency cycle weighted N2O and CH4 emissions, updated data. seeks comment on whether the current on a CO2-equivalent basis, along with If the agency moves forward with its offsets and caps would continue to be CO2, into their CO2 emissions fleet proposal to eliminate these factors, EPA appropriate in such circumstances or average compliance level.373 EPA would consider whether it is whether changes are warranted. estimated that on a CO2 equivalent appropriate to initiate a new rulemaking

We emphasize again that the values in CAFE standard with both their another fleet, or acquired from another these tables are estimates, and not domestically-manufactured and manufacturer. On top of this necessarily the ultimate levels with imported passenger car fleets—that is, requirement, EISA expressly requires which each of these manufacturers will domestic and imported passenger car each manufacturer to meet a minimum have to comply, for the reasons fleets must comply separately with the flat fuel economy standard for described above. passenger car CAFE standard in each domestically manufactured passenger 375 376 C. Minimum Domestic Passenger Car model year. In doing so, they may use cars. According to the statute, the Standards whatever flexibilities are available to minimum standard shall be the greater them under the CAFE program, such as of (A) 27.5 miles per gallon; or (B) 92% EPCA has long required using credits ‘‘carried forward’’ from of the average fuel economy projected manufacturers to meet the passenger car prior model years, transferred from by DOT for the combined domestic and

370 77 FR 62805, Oct. 15, 2012. 374 In the final rule for MYs 2012–2016, EPA 376 Transferred or traded credits may not be used, 371 77 FR 62649, Oct. 15, 2012. acknowledged that advanced diesel or lean-burn pursuant to 49 U.S.C. 32903(g)(4) and (f)(2), to meet gasoline vehicles of the future may face greater 372 75 FR 25421–24, May 7, 2010. the domestically manufactured passenger challenges meeting the CH4 and N2O standards than automobile minimum standard specified in 49 373 77 FR 62798, Oct. 15, 2012. the rest of the fleet. [See 75 FR 25422, May 7, 2010]. U.S.C. 32902(b)(4) and in 49 CFR 531.5(d). 375 49 U.S.C. 32904(b) (2007).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00210 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.140 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43195

nondomestic passenger automobile this requirement in more detail in ‘‘Preferred (Alternative 3)’’ and fleets manufactured for sale in the Section V.A.1 below. calculates what those standards would United States by all manufacturers in The following table lists the proposed be under the no action alternative (as the model year, which projection shall minimum domestic passenger car issued in 2012, and as updated by be published in the Federal Register standards (which very likely will be today’s analysis) and under the other when the standard for that model year updated for the final rule as the agency alternatives described and discussed is promulgated.377 NHTSA discusses updates its overall analysis and further in Section IV, below. resultant projection), highlighted as

respectively, for MYs 2021–2026 that D. Light Truck Standards are defined by the following coefficients: For light trucks, NHTSA and EPA are proposing CAFE and CO2 standards,

377 49 U.S.C. 32902(b)(4).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00211 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.141 EP24AU18.142 43196 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

Section II.C above discusses in detail coefficients result in the footprint- 2017–2020 standards are also shown for how the coefficients in Table III–4 were dependent targets shown graphically comparison. developed for this proposal. The below for MYs 2021–2026. The MYs

378 Prior to MY 2021, average achieved CO2 levels technologies that reduce the refrigerant leaks. calculations after MY 2020, CO2 targets and include adjustments reflecting the use of Because EPA is today proposing to exclude air resultant fleet average requirements for MYs 2021 automotive refrigerants with reduced global conditioning refrigerants and leakage, and nitrous and beyond do not reflect these adjustments. warming potential (GWP) and/or the use of oxide and methane GHGs from average performance

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00212 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.143 EP24AU18.144 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43197

While we do not know yet with above would result in the following refrigerant leaks. Because EPA is today certainty what CAFE and CO2 levels average required fuel economy and CO2 proposing to exclude air conditioning will ultimately be required of individual emissions levels for individual refrigerants and leakage, and nitrous manufacturers, because those levels will manufacturers during MYs 2021–2026. oxide and methane GHGs from average depend on the mix of vehicles that they Prior to MY 2021, average required CO2 performance calculations after model produce for sale in future model years, levels reflect underlying target functions year 2020, CO2 targets and resultant based on the market forecast of future (specified above) that reflect the use of fleet average requirements for model sales that were used to examine today’s automotive refrigerants with reduced years 2021 and beyond do not reflect proposed standards, we currently global warming potential (GWP) and/or these adjustments. estimate that the target functions shown the use of technologies that reduce the

We emphasize again the values in As discussed above in Chapter II, baseline no-action alternative, all of the these tables are estimates and not today’s notice also presents the results alternatives are more stringent than the necessarily the ultimate levels with of analysis estimating impacts under a preferred alternative. which each of these manufacturers will range of other regulatory alternatives the EPA also seeks comment on retaining have to comply for reasons described agencies are considering. Aside from the the existing credit program for above. no-action alternative, NHTSA and EPA regulation of A/C refrigerant leakage, defined the different regulatory nitrous oxide, and methane emissions as IV. Alternative CAFE and GHG alternatives in terms of percent- part of the CO2 standard. Standards Considered for MYs 2021/ increases in CAFE and GHG stringency The agencies have examined these 22–2026 from year to year. Under some alternatives because the agencies intend alternatives, the rate of increase is the to continue considering them as options Agencies typically consider regulatory same for both passenger cars and light for the final rule. The agencies seek alternatives in proposals as a way of trucks; under others, the rate of increase comment on these alternatives and on evaluating the comparative effects of differs. Two alternatives also involve a the analysis presented here, seek any different potential ways of gradual discontinuation of CAFE and relevant data and information, and will accomplishing their desired goal.379 average GHG adjustments reflecting the review responses. That review could Alternatives analysis begins with a ‘‘no- application of technologies that improve lead the agencies to select one of the action’’ alternative, typically described air conditioner efficiency or, in other as what would occur in the absence of ways, improve fuel economy under to calculate tailpipe CO2 for its standards. In any regulatory action. Today’s proposal conditions not represented by long- addition, under the no action alternative EPA adds CO2 equivalent (using Global Warming Potential includes a no-action alternative, standing fuel economy test procedures. (GWP) adjustment) for AC refrigerant leakage and described below, as well as seven For increased harmonization with nitrous oxide and methane emissions. The CAFE ‘‘action alternatives’’ besides the NHTSA CAFE standards, which cannot program does not include A/C refrigerant leakage, nitrous oxide and methane emissions because they proposal. The proposal may, in places, account for such issues, under do not impact fuel economy. Under Alternatives 1– be referred to as the ‘‘preferred Alternatives 1–8, EPA would regulate 8, the standards are completely aligned for gasoline alternative,’’ which is NEPA parlance, tailpipe CO2 independently of A/C because compliance is based on tailpipe CO2, CH4 refrigerant leakage, nitrous oxide and and CO for both programs and not emissions but NHTSA and EPA intend ‘‘proposal,’’ unrelated to fuel economy. Diesel and alternative ‘‘proposed action,’’ and ‘‘preferred methane emissions. Under the no action fuel vehicles would continue to be treated alternative’’ to be used interchangeably alternative, EPA would continue to differently between the CAFE and CO2 programs. for purposes of this rulemaking. regulate A/C refrigerant leakage, nitrous While harmonization would be significantly oxide and methane emissions under the improved, standards would not be fully aligned 380 because of the small fraction of the fleet that uses 379 As Section V.A.3 explains, NEPA requires overall CO2 standard. Like the diesel and alternative fuels (e.g., about four percent agencies to compare the potential environmental of the MY 2016 fleet), as well as differences impacts of their proposed actions to those of a 380 For the CAFE program, carbon-based tailpipe involving EPCA/EISA provisions EPA, lacking any reasonable range of alternatives. Executive Orders emissions (including CO2, CH4 and CO) are specific direction under the CAA, has declined to 12866 and 13563 and OMB Circular A–4 also measured, and fuel economy is calculated using a adopt, such as minimum standards for domestic encourage agencies to evaluate regulatory carbon balance equation. EPA uses carbon-based passenger cars and limits on credit transfers alternatives in their rulemaking analyses. emissions (CO2, CH4 and CO, the same as for CAFE) between regulated fleets.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00213 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.146 43198 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

other regulatory alternatives for the final A. What alternatives did NHTSA and rule. EPA consider? The table below shows the different alternatives evaluated in this proposal.

Also, as mentioned previously in whether to proceed with this proposal better harmony with the CAFE program Section III.B., EPA seeks comments on to discontinue accounting for A/C or whether to continue to consider these leakage, methane emissions, and nitrous factors toward compliance and retain 381 Carbon dioxide equivalent of air conditioning oxide emissions as part of the CO2 that as a feature that differs between the refrigerant leakage, nitrous oxide and methane emissions standards to provide for programs. EPA seeks comment on emissions are included for compliance with the whether to change existing methane and EPA standards for all MYs under the baseline/no emissions may be regulated independently by EPA. nitrous oxide standards that were action alternative. Carbon dioxide equivalent is The GWP equivalent of each of the emissions would calculated using the Global Warming Potential finalized in the 2012 rule. Specifically, no longer be included with the tailpipe CO2 for EPA seeks information from the public (GWP) of each of the emissions. compliance with tailpipe CO2 standards. A 382 Beginning in MY 2021, air conditioning lengthier discussion of this issue can be found in on whether the existing standards are refrigerant leakage, nitrous oxide, and methane Section III.B. appropriate, or whether they should be

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00214 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.147 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43199

revised to be less stringent or more draws attention the discussion of For MY 2026, this alternative applies stringent based on any updated data. ‘‘enhanced flexibilities’’ in Section X.C. the same targets as for MY 2025. Carbon Additionally, the agencies note that B. Definition of Alternatives dioxide equivalent of air conditioning this proposal also seeks comment on a refrigerant leakage, nitrous oxide, and 1. No-Action Alternative number of additional compliance methane emissions are included for flexibilities for the programs. See The No-Action Alternative applies the compliance with the EPA standards for Section X below, and EPA specifically augural CAFE and final GHG targets all model years under the baseline/no announced in 2012 for MYs 2021–2025. action alternative.

2. Alternative 1 (Proposed) tailpipe CO2 standards. Section III, cars and 0.5% for light trucks. Section above, defines this alternative in greater III describes the proposed standards Alternative 1 holds the stringency of detail. included in the preferred alternative. targets constant and MY 2020 levels 3. Alternative 2 Beginning in MY 2021, air conditioning through MY 2026. Beginning in MY refrigerant leakage, nitrous oxide, and 2021, air conditioning refrigerant Alternative 2 increases the stringency methane emissions are no longer leakage, nitrous oxide, and methane of targets annually during MYs 2021– included with the tailpipe CO for emissions are no longer included with 2026 (on a gallon per mile basis, starting 2 compliance with tailpipe CO2 standards. the tailpipe CO2 for compliance with from MY 2020) by 0.5% for passenger

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00215 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.148 EP24AU18.149 43200 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

4. Alternative 3 trucks. The cap on adjustments for AC 4, 2, and 0 grams per mile in MYs 2022, efficiency improvements declines from 2023, 2024, 2025, and 2026, Alternative 3 phases out A/C and off- 6 grams per mile in MY 2021 to 5, 4, 3, respectively. Beginning in MY 2021, air cycle adjustments and increases the 2, and 0 grams per mile in MYs 2022, conditioning refrigerant leakage, nitrous stringency of targets annually during 2023, 2024, 2025, and 2026, oxide, and methane emissions are no MYs 2021–2026 (on a gallon per mile respectively. The cap on adjustments for longer included with the tailpipe CO2 basis, starting from MY 2020) by 0.5% off-cycle improvements declines from for compliance with tailpipe CO2 for passenger cars and 0.5% for light 10 grams per mile in MY 2021 to 8, 6, standards.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00216 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.150 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43201

5. Alternative 4 2026 (on a gallon per mile basis, starting refrigerant leakage, nitrous oxide, and from MY 2020) by 1.0% for passenger methane emissions are no longer Alternative 4 increases the stringency cars and 2.0% for light trucks. included with the tailpipe CO2 for of targets annually during MYs 2021– Beginning in MY 2021, air conditioning compliance with tailpipe CO2 standards.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00217 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.151 43202 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

6. Alternative 5 from MY 2021) by 1.0% for passenger included with the tailpipe CO2 for cars and 2.0% for light trucks. compliance with tailpipe CO2 standards, Alternative 5 increases the stringency Beginning in MY 2021, air conditioning and MY 2021 CO2 targets are adjusted of targets annually during MYs 2022– refrigerant leakage, nitrous oxide, and accordingly. 2026 (on a gallon per mile basis, starting methane emissions are no longer

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00218 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.152 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43203

7. Alternative 6 2026 (on a gallon per mile basis, starting refrigerant leakage, nitrous oxide, and from MY 2020) by 2.0% for passenger methane emissions are no longer Alternative 6 increases the stringency cars and 3.0% for light trucks. included with the tailpipe CO2 for of targets annually during MYs 2021– Beginning in MY 2021, air conditioning compliance with tailpipe CO2 standards.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00219 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.153 43204 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

8. Alternative 7 trucks. The cap on adjustments for AC 4, 2, and 0 grams per mile in MYs 2022, efficiency improvements declines from 2023, 2024, 2025, and 2026, Alternative 7 phases out A/C and off- 6 grams per mile in MY 2021 to 5, 4, 3, respectively. Beginning in MY 2021, air cycle adjustments and increases the 2, and 0 grams per mile in MYs 2022, conditioning refrigerant leakage, nitrous stringency of targets annually during 2023, 2024, 2025, and 2026, oxide, and methane emissions are no MYs 2021–2026 (on a gallon per mile respectively. The cap on adjustments for longer included with the tailpipe CO2 basis, starting from MY 2020) by 1.0% off-cycle improvements declines from for compliance with tailpipe CO2 for passenger cars and 2.0% for light 10 grams per mile in MY 2021 to 8, 6, standards.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00220 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.154 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43205

9. Alternative 8 from MY 2021) by 2.0% for passenger included with the tailpipe CO2 for cars and 3.0% for light trucks. compliance with tailpipe CO2 standards, Alternative 8 increases the stringency Beginning in MY 2021, air conditioning and MY 2021 CO2 targets are adjusted of targets annually during MYs 2022– refrigerant leakage, nitrous oxide, and accordingly. 2026 (on a gallon per mile basis, starting methane emissions are no longer

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00221 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.155 43206 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

V. Proposed Standards, the Agencies’ model year.385 In determining the determination of what standards can be Statutory Obligations, and Why the maximum feasible level achievable by considered maximum feasible involves Agencies Propose To Choose Them the manufacturers, EPCA requires that a weighing and balancing of these Over the Alternatives NHTSA consider the four statutory factors, and the balance may shift factors of technological feasibility, depending on the information before A. NHTSA’s Statutory Obligations and economic practicability, the effect of NHTSA about the expected Why the Proposed Standards Appear to other motor vehicle standards of the circumstances in the model years be Maximum Feasible Government on fuel economy, and the covered by the rulemaking. The 1. EPCA, as Amended by EISA need of the United States to conserve agency’s decision must also support the 386 EPCA, as amended by EISA, contains energy. In addition, NHTSA has the overarching purpose of EPCA, energy authority to (and traditionally does) conservation, while balancing these a number of provisions regarding how 388 NHTSA must set CAFE standards. consider other relevant factors, such as factors. NHTSA must establish separate CAFE the effect of the CAFE standards on Besides the requirement that the motor vehicle safety and consumer standards be maximum feasible for the standards for passenger cars and light 387 trucks 383 for each model year,384 and preferences. The ultimate fleet in question and the model year in each standard must be the maximum question, EPCA/EISA also contain 385 feasible that NHTSA believes the Id. 386 49 U.S.C. 32902(f) (2007). 388 Center for Biological Diversity v. NHTSA, 538 manufacturers can achieve in that 387 Both of these additional considerations also F. 3d 1172, 1197 (9th Cir. 2008) (‘‘Whatever method relate, to some extent, to economic practicability, it uses, NHTSA cannot set fuel economy standards 383 49 U.S.C. 32902(b)(1) (2007). but NHTSA also has the authority to consider them that are contrary to Congress’ purpose in enacting 384 49 U.S.C. 32902(a) (2007). independently of that statutory factor. the EPCA—energy conservation.’’)

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00222 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.156 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43207

several other requirements as explained cargo-carrying capability, etc., need to standard for that model year will also be below. use more fuel per mile to perform those evaluated or reevaluated and jobs than vehicles without these established accordingly. NHTSA (a) Lead Time characteristics. Thus, regardless of the explained this in the rulemaking to EPCA requires that NHTSA prescribe plain language of the statute, NHTSA establish standards for MYs 2017 and new CAFE standards at least 18 months believes that the different fuel economy beyond and received no comments.394 before the beginning of each model capabilities of cars and trucks would The 2016 Alliance/Global petition for year.389 For light-duty vehicles, NHTSA generally make separate standards rulemaking asked NHTSA to has consistently interpreted the appropriate for these different types of retroactively revise the 92-percent ‘‘beginning of each model year’’ as vehicles. minimum domestic passenger car September 1 of the CY prior, such that EPCA, as amended by EISA, also standards for MYs 2012–2016 ‘‘to reflect the beginning of MY 2019 would be requires another separate standard to be 92 percent of the required average September 1, 2018. Thus, if the first year set for domestically-manufactured 392 passenger car standard taking into for which NHTSA is proposing to set passenger cars. Unlike under the account the fleet mix as it actually new standards in this NPRM is MY standards for passenger cars and light occurred, rather than what was 2022, NHTSA interprets this provision trucks described above, the compliance forecast.’’ The petitioners stated that as requiring us to issue a final rule burden of the minimum domestic doing so would be ‘‘fully consistent covering MY 2022 standards no later passenger car standard is the same for with the statute.’’ 395 than April 1, 2020. all manufacturers; the statute clearly NHTSA understands that determining For amendments to existing states that any manufacturer’s the 92 percent value ahead of the model standards, EPCA requires that if the domestically-manufactured passenger year to which it applies, based on the amendments make an average fuel car fleet must meet the greater of either information then available to the economy standard more stringent, at 27.5 mpg on average, or agency, results in a different mpg least 18 months of lead time must be number than if NHTSA determined the provided.390 EPCA contains no lead . . . 92 percent of the average fuel economy 92 percent value based on the projected by the Secretary for the combined time requirement unless amendments domestic and non-domestic passenger information available at the end of the make an average fuel economy standard automobile fleets manufactured for sale in model year in question. NHTSA further less stringent. NHTSA therefore the United States by all manufacturers in the understands that determining the 92 interprets EPCA as allowing model year, which projection shall be percent value ahead of time can make amendments to reduce a standard’s published in the Federal Register when the the domestic minimum passenger car stringency up until the beginning of the standard for that model year is promulgated standard more stringent than it could be model year in question. In this in accordance with [49 U.S.C. 32902(b)].393 if it were determined at the end of the rulemaking, NHTSA is proposing to Since that requirement was model year, if manufacturers end up amend the standards for model year promulgated, the ‘‘92 percent’’ has producing more larger-footprint 2021. Since the agency proposes to always been greater than 27.5 mpg. passenger cars than NHTSA originally reduce these standards, this action is NHTSA published the 92-percent anticipated. not subject to a lead time requirement. minimum domestic passenger car Accordingly, NHTSA seeks comment standards for model years 2017–2025 at on this request by Alliance/Global. (b) Separate Standards for Cars and 49 CFR 531.5(d) as part of the 2012 final Additionally, recognizing the Trucks, and Minimum Standards for rule. For MYs 2022–2025, 531.5(e) states uncertainty inherent in projecting Domestic Passenger Cars that these were to be applied if, when specific mpg values far into the future, As discussed above, EPCA requires actually proposing MY 2022 and it is possible that NHTSA could define NHTSA to set separate CAFE standards subsequent standards, the previously the mpg values associated with a CAFE for passenger cars and light trucks for identified standards for those years are standard (i.e., the footprint curve) as a each model year.391 NHTSA interprets deemed maximum feasible, but if range rather than as a single number. this requirement as preventing the NHTSA determines that the previously For example, the sensitivity analysis agency from setting a single combined identified standards are not maximum included in this proposal and in the CAFE standard for cars and trucks feasible, the 92-percent minimum accompanying PRIA could provide a together, based on the plain language of domestic passenger car standards would basis for such an mpg range ‘‘defining’’ the statute. Congress originally intended also change. This is consistent with the the passenger car standard in any given separate CAFE standards for cars and statutory language that the 92-percent model year. If NHTSA took that trucks to reflect the different fuel standards must be determined at the approach, 92 percent of that ‘‘standard’’ economy capabilities of those different time an overall passenger car standard would also, necessarily, be a range. We types of vehicles, and over the history is promulgated and published in the also seek comment on this or other of the CAFE program, has never revised Federal Register. Thus, any time similar approaches. this requirement. Even as many cars and NHTSA establishes or changes a (c) Attribute-Based and Defined by trucks have come to resemble each other passenger car standard for a model year, Mathematical Function more closely over time—many crossover the minimum domestic passenger car and sport-utility models, for example, EISA requires NHTSA to set CAFE come in versions today that may be 392 In the CAFE program, ‘‘domestically- standards that are ‘‘based on 1 or more subject to either the car standards or the manufactured’’ is defined by Congress in 49 U.S.C. § 32904(b). The specifics of the definition are too 394 truck standards depending on their 77 FR 62624, 63028 (Oct. 15, 2012). many for a footnote, but roughly, a passenger car 395 Automobile Alliance and Global Automakers characteristics—it is still accurate to say is ‘‘domestically manufactured’’ as long as at least Petition for Direct Final Rule with Regard to that vehicles with truck-like 75% of the cost to the manufacturer is attributable Various Aspects of the Corporate Average Fuel characteristics such as 4 wheel drive, to value added in the United States, Canada, or Economy Program and the Greenhouse Gas Program Mexico, unless the assembly of the vehicle is (June 20, 2016) at 5, 17–18, available at https:// completed in Canada or Mexico and the vehicle is www.epa.gov/sites/production/files/2016-09/ 389 49 U.S.C. 32902(a) (2007). imported into the United States more than 30 days documents/petition_to_epa_from_auto_alliance_ 390 49 U.S.C. 32902(g)(2) (2007). after the end of the model year. and_global_automakers.pdf [hereinafter Alliance/ 391 49 U.S.C. 32902(b)(1) (2007). 393 49 U.S.C. § 32902(b)(4) (2007). Global Petition].

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00223 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43208 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

attributes related to fuel economy and 32902(c) and 32902(g). We therefore feasibility and economic practicability express[ed] . . . in the form of a believe that it is reasonable to interpret are often conflated, as will be covered mathematical function.’’ 396 NHTSA has section 32902(b)(3)(B) as applying only further in the following section. To be thus far based standards on vehicle to the establishing of new standards clear, whether a fuel-economy- footprint and proposes to continue to do rather than to the combined action of improving technology does or will exist so for all the reasons described in establishing new standards and (technological feasibility) is a different previous rulemakings. As in previous amending existing standards. question from what economic rulemakings, NHTSA proposes to define Moreover, we believe it would be an consequences could ensue if NHTSA the standards in the form of a absurd result not intended by Congress effectively requires that technology to constrained linear function that if the five year maximum limitation become widespread in the fleet and the generally sets higher (more stringent) were interpreted to prevent NHTSA economic consequences of the absence targets for smaller-footprint vehicles and from revising a previously-established of consumer demand for technology that lower (less stringent) targets for larger- standard that we have determined to be are projected to be required (economic footprint vehicles. These footprint beyond maximum feasible, while practicability). It is therefore possible curves are discussed in much greater concurrently setting five years of for standards to be technologically detail in Section II.C above. We seek standards not so distant from today. The feasible but still beyond the level that comment both on the choice of footprint concerns Congress sought to address are NHTSA determines to be maximum as the relevant attribute and on the much starker when NHTSA is trying to feasible due to consideration of the rationale for the constrained linear determine what standards would be other relevant factors. functions chosen to represent the maximum feasible 10 years from now as (2) Economic Practicability standards. compared to three years from now. ‘‘Economic practicability’’ has (d) Number of Model Years for Which (e) Maximum Feasible traditionally referred to whether a Standards May Be Set at a Time As discussed above, EPCA requires standard is one ‘‘within the financial EISA also states that NHTSA shall NHTSA to consider four factors in capability of the industry, but not so ‘‘issue regulations under this title determining what levels of CAFE stringent as to’’ lead to ‘‘adverse prescribing average fuel economy standards would be maximum feasible, economic consequences, such as a standards for at least 1, but not more and NHTSA presents in the sections significant loss of jobs or unreasonable than 5, model years.’’ 397 In the 2012 below its understanding of what those elimination of consumer choice.’’ 401 In final rule, NHTSA interpreted this four factors mean. All factors should be evaluating economic practicability, provision as preventing the agency from considered, in the manner appropriate, NHTSA considers the uncertainty setting final standards for all of MYs and then the maximum feasible surrounding future market conditions 2017–2025 in a single rulemaking standards should be determined. and consumer demand for fuel economy action, so the MYs 2022–2025 standards alongside consumer demand for other (1) Technological Feasibility were termed ‘‘augural,’’ meaning ‘‘that vehicle attributes. NHTSA has they represent[ed] the agency’s current ‘‘Technological feasibility’’ refers to explained in the past that this factor can judgment, based on the information whether a particular method of be especially important during available to the agency [then], of what improving fuel economy is available for rulemakings in which the auto industry levels of stringency would be maximum deployment in commercial application is facing significantly adverse economic feasible in those model years.’’ 398 That in the model year for which a standard conditions (with corresponding risks to said, NHTSA also repeatedly clarified is being established. Thus, NHTSA is jobs). Consumer acceptability is also a that the augural standards were in no not limited in determining the level of major component to economic way final standards and that a future de new standards to technology that is practicability,402 which can involve novo rulemaking would be necessary in already being commercially applied at consideration of anticipated consumer order to both propose and promulgate the time of the rulemaking. For this responses not just to increased vehicle final standards for MYs 2022–2025. proposal, NHTSA is considering a wide cost, but also to the way manufacturers Today, NHTSA proposes to establish range of technologies that improve fuel may change vehicle models and vehicle new standards for MYs 2022–2026 and economy, subject to the constraints of sales mix in response to CAFE to revise the previously-established final EPCA regarding how to treat alternative standards. In attempting to determine standards for MY 2021. Legislative fueled vehicles, and considering the the economic practicability of attribute- history suggests that Congress included need to account for which technologies based standards, NHTSA considers a the five year maximum limitation so have already been applied to which wide variety of elements, including the NHTSA would issue standards for a vehicle model/configuration, and the annual rate at which manufacturers can period of time where it would have need to realistically estimate the cost increase the percentage of their fleet that reasonably realistic estimates of market and fuel economy impacts of each employs a particular type of fuel-saving conditions, technologies, and economic technology. NHTSA has not attempted technology,403 the specific fleet mixes of practicability (i.e., not set standards too to account for every technology that far into the future).399 However, the might conceivably be applied to has considered a range of hybrid vehicle concerns Congress sought to address by improve fuel economy and considers it technologies that do so. imposing those limitations are not unnecessary to do so given that many 401 67 FR 77015, 77021 (Dec. 16, 2002). 402 See, e.g., Center for Auto Safety v. NHTSA present for nearer model years where technologies address fuel economy in (CAS), 793 F.2d 1322 (D.C. Cir. 1986) 400 NHTSA already has existing standards. similar ways. Technological (Administrator’s consideration of market demand as Revisiting existing standards is component of economic practicability found to be contemplated by both 49 U.S.C. 400 For example, NHTSA has not considered high- reasonable); Public Citizen v. NHTSA, 848 F.2d 256 speed flywheels as potential energy storage devices (Congress established broad guidelines in the fuel for hybrid vehicles; while such flywheels have been economy statute; agency’s decision to set lower 396 49 U.S.C. 32902(b)(3)(A). demonstrated in the laboratory and even tested in standards was a reasonable accommodation of 397 49 U.S.C. 32902(b)(3)(B). concept vehicles, commercially available hybrid conflicting policies). 398 77 FR 62623, 62630 (Oct. 15, 2012). vehicles currently known to NHTSA use chemical 403 For example, if standards effectively require 399 See 153 Cong. Rec. 2665 (Dec. 28, 2007). batteries as energy storage devices, and the agency manufacturers to widely apply technologies that

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00224 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43209

different manufacturers, and standards. Economic practicability is increasing vehicle weight thereby lower assumptions about the cost of standards complex, and like the other factors must fuel economy capability, thus to consumers and consumers’ valuation also be considered in the context of the decreasing the level of average fuel of fuel economy, among other things. overall balancing and EPCA’s economy that NHTSA can determine to Prior to the MYs 2005–2007 overarching purpose of energy be feasible. NHTSA has considered the rulemaking under the non-attribute- conservation. Depending on the additional weight that it estimates based (fixed value) CAFE standards, conditions of the industry and the would be added in response to new NHTSA generally sought to ensure the assumptions used in the agency’s safety standards during the rulemaking economic practicability of standards in analysis of alternative standards, timeframe.405 NHTSA has also part by setting them at or near the NHTSA could well find that standards accounted for EPA’s ‘‘Tier 3’’ standards capability of the ‘‘least capable that maximize net benefits, or that are for criteria pollutants in its estimates of manufacturer’’ with a significant share higher or lower, could be at the limits technology effectiveness.406 of the market, i.e., typically the of economic practicability, and thus In the 2012 final rule establishing manufacturer whose fleet mix was, on potentially the maximum feasible level, CAFE standards for MYs 2017–2021, average, the largest and heaviest, depending on how the other factors are NHTSA also discussed whether EPA generally having the highest capacity balanced. GHG standards and California GHG and capability so as to not limit the While we discuss safety as a separate standards should be considered and availability of those types of vehicles to consideration, NHTSA also considers accounted for as ‘‘other motor vehicle consumers. In the first several safety as closely related to, and in some standards of the Government.’’ NHTSA rulemakings establishing attribute-based circumstances a subcomponent of recognized that ‘‘To the extent the GHG standards, NHTSA applied marginal economic practicability. On a broad standards result in increases in fuel cost-benefit analysis, considering both level, manufacturers have finite economy, they would do so almost overall societal impacts and overall resources to invest in research and exclusively as a result of inducing consumer impacts. Whether the development. Investment into the manufacturers to install the same types standards maximize net benefits has development and implementation of of technologies used by manufacturers thus been a touchstone in the past for fuel saving technology necessarily in complying with the CAFE NHTSA’s consideration of economic comes at the expense of investing in standards.’’ 407 NHTSA concluded that practicability. Executive Order 12866, as other areas such as safety technology. ‘‘the agency had already considered amended by Executive Order 13563, On a more direct level, when making EPA’s [action] and the harmonization states that agencies should ‘‘select, in decisions on how to equip vehicles, benefits of the National Program in choosing among alternative regulatory manufacturers must balance cost developing its own [action],’’ and that approaches, those approaches that considerations to avoid pricing further ‘‘no further action was needed.’’ 408 maximize net benefits . . .’’ In practice, consumers out of the market. As Considering the issue afresh in this however, agencies, including NHTSA, manufacturers add technology to proposal, and looking only at the words must consider situations in which the increase fuel efficiency, they may in the statute, obviously EPA’s GHG modeling of net benefits does not decide against installing new safety standards applicable to light-duty capture all of the relevant equipment to reduce cost increases. And vehicles are literally ‘‘other motor considerations of feasibility. Therefore, as the price of vehicles increase beyond vehicle standards of the Government,’’ as in past rulemakings, NHTSA is the reach of more consumers, such in that they are standards set by a considering net societal impacts, net consumers continue to drive or Federal agency that apply to motor consumer impacts, and other related purchase older, less safe vehicles. In vehicles. Basic chemistry makes fuel elements in the consideration of assessing practicability, NHTSA also economy and tailpipe CO2 emissions economic practicability. considers the harm to the nation’s two sides of the same coin, as discussed NHTSA’s consideration of economic economy caused by highway fatalities at length above, and when two agencies practicability depends on a number of and injuries. functionally regulate both (because by elements. Expected availability of regulating fuel economy, you regulate capital to make investments in new (3) The Effect of Other Motor Vehicle Standards of the Government on Fuel CO2 emissions, and vice versa), it would technologies matters; manufacturers’ be absurd not to link their standards.409 expected ability to sell vehicles with Economy The global warming potential of N2O, certain technologies matters; likely ‘‘The effect of other motor vehicle CH4, and HFC emissions are not closely consumer choices matter and so forth. standards of the Government on fuel linked with fuel economy, but neither NHTSA’s analysis of the impacts of this economy’’ involves analysis of the do they affect fuel economy capabilities. proposal incorporates assumptions to effects of compliance with emission, How, then, should NHTSA consider capture aspects of consumer safety, noise, or damageability standards EPA’s various GHG standards? preferences, vehicle attributes, safety, on fuel economy capability and thus on NHTSA is aware that some and other elements relevant to an average fuel economy. In many past stakeholders believe that NHTSA’s impacts estimate; however, it is difficult CAFE rulemakings, NHTSA has said obligation to set maximum feasible to capture every such constraint. that it considers the adverse effects of CAFE standards can best be executed by Therefore, it is well within the agency’s other motor vehicle standards on fuel letting EPA decide what GHG standards discretion to deviate from the level at economy. It said so because, from the which modeled net benefits are CAFE program’s earliest years 404 until 405 PRIA, Chapter 5. maximized if the agency concludes that recently, the effects of such compliance 406 PRIA, Chapter 6. that level would not represent the on fuel economy capability over the 407 77 FR 62624, 62669 (Oct. 15, 2012). maximum feasible level for future CAFE history of the CAFE program have been 408 Id. 409 negative ones. For example, safety In fact, EPA includes tailpipe CH4, CO, and consumers do not want, or to widely apply CO2 in the measurement of tailpipe CO2 for GHG technologies before they are ready to be standards that have the effect of compliance using a carbon balance equation so that widespread, NHTSA believes that these standards the measurement of tailpipe CO2 exactly aligns with could potentially be beyond economically 404 42 FR 63184, 63188 (Dec. 15, 1977). See also the measurement of fuel economy for the CAFE practicable. 42 FR 33534, 33537 (June 30, 1977). compliance.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00225 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43210 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

are appropriate and reasonable under but is not labeled as one (i.e., a State unnecessary provisions were deleted. the CAA. NHTSA disagrees. While EPA tailpipe CO2 standard or prohibition on Specifically, the recodification and NHTSA consider some similar CO2 emissions). eliminated subsection (d). The House factors under the CAA and EPCA/EISA, NHTSA and EPA agree that state report on the recodification declared respectively, they are not identical. tailpipe greenhouse gas emissions that the subdivision was ‘‘executed,’’ Standards that are appropriate under the standards do not become Federal and described its purpose as CAA may not be ‘‘maximum feasible’’ standards and qualify as ‘‘other motor ‘‘[p]rovid[ing] for modification of under EPCA/EISA, and vice versa. vehicle standards of the Government,’’ average fuel economy standards for Moreover, considering EPCA’s language when subject to a CAA preemption model years 1978, 1979, and 1980.’’ 415 in the context in which it was written, waiver. EPCA’s legislative history It is generally presumed, when Congress it seems unreasonable to conclude that supports this position. includes text in one section and not in Congress intended EPA to dictate CAFE EPCA, as initially passed in 1975, another, that Congress knew what it was stringency. In fact, Congress clearly mandated average fuel economy doing and made the decision separated NHTSA’s and EPA’s standards for passenger cars beginning deliberately. responsibilities for CAFE under EPCA with model year 1978. The law required NHTSA has previously considered the by giving NHTSA authority to set the Secretary of Transportation to impact of California’s Low Emission standards and EPA authority to measure establish, through regulation, maximum Vehicle standards in establishing fuel and calculate fuel economy. If Congress feasible fuel economy standards 412 for economy standards and occasionally had wanted EPA to set CAFE standards, model years 1981 through 1984 with the has done so under the ‘‘other standards’’ it could have given that authority to intent to provide steady increases to sections.416 During the 2012 EPA in EPCA or at any point since achieve the standard established for rulemaking, NHTSA sought comment Congress amended EPCA.410 1985 and thereafter authorized the on the appropriateness of considering NHTSA and EPA are obligated by Secretary to adjust that standard. California’s tailpipe GHG emission Congress to exercise their own For the statutorily-established standards in this section and concluded independent judgment in fulfilling their standards for model years 1978–1980, that doing so was unnecessary.417 In statutory missions, even though both EPCA provided each manufacturer with light of the legislative history discussed agencies’ regulations affect both fuel the right to petition for changes in the above, however, NHTSA now economy and CO2 emissions. Because of standards applicable to that determines that this was not this relationship, it is incumbent on manufacturer. A petitioning appropriate. Notwithstanding the both agencies to coordinate and look to manufacturer had the burden of improper categorization of such demonstrating a ‘‘Federal fuel economy one another’s actions to avoid discussions, NHTSA may consider standards reduction’’ was likely to exist unreasonably burdening industry elements not specifically designated as for that manufacturer in one or more of through inconsistent regulations, but factors to be considered under EPCA, those model years and that it had made both agencies must be able to defend given the breadth of such factors as reasonable technology choices. ‘‘Federal their programs on their own merits. As technological feasibility and economic standards,’’ for that limited purpose, with other recent CAFE and GHG practicability, and such consideration included not only safety standards, rulemakings, the agencies are was appropriate.418 continuing do all of these things in this noise emission standards, property loss proposal. reduction standards, and emission (4) The Need of the United States To With regard to standards issued by the standards issued under various Federal Conserve Energy State of California, State tailpipe statutes, but also ‘‘emissions standards ‘‘The need of the United States to standards (whether for greenhouse gases applicable by reason of section 209(b) of 413 conserve energy’’ means ‘‘the consumer or for other pollutants) do not qualify as [the CAA].’’ (Emphasis added). cost, national balance of payments, ‘‘other motor vehicle standards of the Critically, all definitions, processes, and environmental, and foreign policy Government’’ under 49 U.S.C. 32902(f); required findings regarding a Federal implications of our need for large fuel economy standards reduction were therefore, NHTSA will not consider quantities of petroleum, especially located within a single self-contained them as such in proposing maximum imported petroleum.’’ 419 feasible average fuel economy subsection of 15 U.S.C. 2002 that standards. States may not adopt or applied only to model years 1978– (i) Consumer Costs and Fuel Prices 414 enforce tailpipe greenhouse gas 1980. Fuel for vehicles costs money for emissions standards when such In 1994, Congress recodified EPCA. vehicle owners and operators. All else standards relate to fuel economy As part of this recodification, the CAFE equal, consumers benefit from vehicles standards and are therefore preempted provisions were moved to Title 49 of the that need less fuel to perform the same under EPCA, regardless of whether EPA United States Code. In doing so, amount of work. Future fuel prices are granted any waivers under the Clean Air a critical input into the economic 412 Act (CAA).411 As is the case today, EPCA required the Secretary to determine ‘‘maximum feasible average Preempted standards of a State or a fuel economy’’ after considering technological 415 H.R. Rep. No. 103–180, at 583–584, tbl. 2A. political subdivision of a State include, feasibility, economic practicability, the effect of 416 See, e.g., 68 FR 16896, 71 FR 17643. for example: other Federal motor vehicle standards on fuel 417 See 77 FR 62669. (1) A fuel economy standard; and economy, and the need of the Nation to conserve 418 See, e.g., discussion in Center for Automotive (2) A law or regulation that has the energy. 15 U.S.C. 2002(e) (recodified July 5, 1994). Safety v. National Highway Traffic Safety 413 Section 202 of the CAA (42 U.S.C. 7521) Administration, et al., 793 F.2d. 1322 (D.C. Cir. direct effect of a fuel economy standard, requires EPA to prescribe air pollutant emission 1986) at 1338, et seq., providing that NHTSA may standards for new vehicles; Section 209 of the CAA consider consumer demand in establishing 410 We note, for instance, that EISA was passed (42 U.S.C. 7543) preempts state emissions standards standards, but not ‘‘to such an extent that it ignored after the Massachusetts v. EPA decision by the but allows California to apply for a waiver of such the overarching goal of fuel conservation. At the Supreme Court. If Congress had wanted to amend preemption. other extreme, a standard with harsh economic EPCA in light of that decision, they would have 414 As originally enacted as part of Public Law consequences for the auto industry also would done so at the time. They did not. 94–163, that subsection was designated as section represent an unreasonable balancing of EPCA’s 411 This topic is discussed further in Section VI 502(d) of the Motor Vehicle Information and Cost policies.’’ below. Savings Act. 419 42 FR 63184, 63188 (Dec. 15, 1977).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00226 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43211

analysis of potential CAFE standards raised concerns about potential NHTSA defined ‘‘the need of the United because they determine the value of fuel economic consequences for automaker States to conserve energy’’ in the late savings both to new vehicle buyers and and supplier operations in the U.S. due 1970s as including, among other things, to society, the amount of fuel economy to disparities between CAFE standards environmental implications. In 1988, that the new vehicle market is likely to at home and their counterpart fuel NHTSA included climate change demand in the absence of new economy/efficiency and GHG standards concepts in its CAFE notices and standards, and they inform NHTSA abroad. NHTSA finds these concerns prepared its first environmental about the ‘‘consumer cost . . . of our more relevant to technological assessment addressing that subject.424 It need for large quantities of petroleum.’’ feasibility and economic practicability cited concerns about climate change as In this proposal, NHTSA’s analysis than to the national balance of one of its reasons for limiting the extent relies on fuel price projections from the payments. Moreover, to the extent that of its reduction of the CAFE standard for U.S. Energy Information an automaker decides to globalize a MY 1989 passenger cars.425 Since then, Administration’s (EIA) Annual Energy vehicle platform to meet more stringent NHTSA has considered the effects of Outlook (AEO) for 2017. Federal standards in other countries, that reducing tailpipe emissions of CO2 in its government agencies generally use EIA’s automaker would comply with United fuel economy rulemakings pursuant to price projections in their assessment of States’s standards and additionally the need of the United States to future energy-related policies. generate overcompensation credits that conserve energy by reducing petroleum it can save for future years if facing consumption. (ii) National Balance of Payments compliance concerns,or sell to other (iv) Foreign Policy Implications Historically, the need of the United automakers. While CAFE standards are States to conserve energy has included set at maximum feasible rates, efforts of U.S. consumption and imports of consideration of the ‘‘national balance manufacturers to exceed those standards petroleum products impose costs on the of payments’’ because of concerns that are rewarded not only with additional domestic economy that are not reflected importing large amounts of oil created a credits but a market advantage in that in the market price for crude petroleum significant wealth transfer to oil- consumers who place a large weight on or in the prices paid by consumers for exporting countries and left the U.S. fuel savings will find such vehicles that petroleum products such as gasoline. economically vulnerable.420 As recently much more attractive. These costs include (1) higher prices for as 2009, nearly half the U.S. trade petroleum products resulting from the deficit was driven by petroleum,421 yet (iii) Environmental Implications effect of U.S. oil demand on world oil this concern has largely laid fallow in Higher fleet fuel economy can reduce prices, (2) the risk of disruptions to the more recent CAFE actions, arguably in U.S. emissions of various pollutants by U.S. economy caused by sudden part because other factors besides reducing the amount of oil that is increases in the global price of oil and petroleum consumption have since produced and refined for the U.S. its resulting impact of fuel prices faced played a bigger role in the U.S. trade vehicle fleet but can also increase by U.S. consumers, and (3) expenses for deficit. Given significant recent emissions by reducing the cost of maintaining the strategic petroleum increases in U.S. oil production and driving, which can result in increased reserve (SPR) to provide a response corresponding decreases in oil imports, vehicle miles traveled (i.e., the rebound option should a disruption in this concern seems likely to remain effect). Thus, the net effect of more commercial oil supplies threaten the fallow for the foreseeable future.422 stringent CAFE standards on emissions U.S. economy, to allow the U.S. to meet Increasingly, changes in the price of fuel of each pollutant depends on the part of its International Energy Agency have come to represent transfers relative magnitudes of its reduced obligation to maintain emergency oil between domestic consumers of fuel emissions in fuel refining and stocks, and to provide a national and domestic producers of petroleum distribution and increases in its defense fuel reserve.426 Higher U.S. rather than gains or losses to foreign emissions from vehicle use. Fuel consumption of crude oil or refined entities. Some commenters have lately savings from CAFE standards also petroleum products increases the necessarily results in lower emissions of magnitude of these external economic 420 See 42 FR 63184, 63192 (Dec. 15, 1977) ‘‘A CO2, the main GHG emitted as a result costs, thus increasing the true economic major reason for this need [to reduce petroleum of refining, distribution, and use of cost of supplying transportation fuels consumption] is that the importation of large transportation fuels. Reducing fuel quantities of petroleum creates serious balance of above the resource costs of producing payments and foreign policy problems. The United consumption directly reduces CO2 them. Conversely, reducing U.S. States currently spends approximately $45 billion emissions because the primary source of consumption of crude oil or refined annually for imported petroleum. But for this large transportation-related CO2 emissions is petroleum products (by reducing motor expenditure, the current large U.S. trade deficit fuel combustion in internal combustion would be a surplus.’’ fuel use) can reduce these external 421 See Today in Energy: Recent improvements in engines. costs. petroleum trade balance mitigate U.S. trade deficit, NHTSA has considered While these costs are considerations, U.S. Energy Information Administration (July 21, environmental issues, both within the the United States has significantly 2014), https://www.eia.gov/todayinenergy/ context of EPCA and the context of the increased oil production capabilities in detail.php?id=17191. National Environmental Policy Act 422 For an illustration of recent increases in U.S. production, see, e.g., U.S. crude oil and liquid fuels (NEPA), in making decisions about the 424 53 FR 33080, 33096 (Aug. 29, 1988). production, Short-Term Energy Outlook, U.S. setting of standards since the earliest 425 53 FR 39275, 39302 (Oct. 6, 1988). Energy Information Administration (June 2018), days of the CAFE program. As courts of 426 While the U.S. maintains a military presence https://www.eia.gov/outlooks/steo/images/ appeal have noted in three decisions in certain parts of the world to help secure global fig13.png. While it could be argued that reducing 423 access to petroleum supplies, that is neither the oil consumption frees up more domestically- stretching over the last 20 years, primary nor the sole mission of U.S. forces produced oil for exports, and thereby raises U.S. overseas. Additionally, the scale of oil consumption GDP, that is neither the focus of the CAFE program 423 CAS, 793 F.2d 1322, 1325 n. 12 (D.C. Cir. reductions associated with CAFE standards would nor consistent with Congress’ original intent in 1986); Public Citizen, 848 F.2d 256, 262–63 n. 27 be insufficient to alter any existing military EPCA. EIA’s Annual Energy Outlook (AEO) series (D.C. Cir. 1988) (noting that ‘‘NHTSA itself has missions focused on ensuring the safe and provides midterm forecasts of production, exports, interpreted the factors it must consider in setting expedient production and transportation of oil and imports of petroleum products, and is available CAFE standards as including environmental around the globe. See Chapter 7 of the PRIA for at https://www.eia.gov/outlooks/aeo/. effects’’); CBD, 538 F.3d 1172 (9th Cir. 2007). more information on this topic.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00227 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43212 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

recent years to the extent that the U.S. technologies as available in the analysis. evaluating what levels of stringency is currently producing enough oil to Thus, today’s analysis includes would result in maximum feasible satisfy nearly all of its energy needs and assumptions about manufacturers’ use standards, NHTSA assesses the is projected to continue to do so or of those technologies, as detailed in potential safety impacts and considers become a net energy exporter. This has Section X.B.1.c)(4) them in balancing the statutory added new stable supply to the global considerations and to determine the (f) EPCA/EISA Requirements That No oil market and reduced the urgency of maximum feasible level of the Longer Apply Post-2020 the U.S. to conserve energy. We discuss standards. this issue in more detail below. Congress amended EPCA through The attribute-based standards that EISA to add two requirements not yet Congress requires NHTSA to set help to (5) Factors That NHTSA Is Prohibited discussed in this section relevant to mitigate the negative safety effects of the From Considering determination of CAFE standards during historical ‘‘flat’’ standards originally EPCA also provides that in the years between MY 2011 and MY required in EPCA, in recent determining the level at which it should 2020 but not beyond. First, Congress rulemakings, NHTSA limited the set CAFE standards for a particular stated that, regardless of NHTSA’s consideration of mass reduction in model year, NHTSA may not consider determination of what levels of lower weight vehicles in its analysis, the ability of manufacturers to take standards would be maximum feasible, which impacted the resulting advantage of several EPCA provisions standards must be set at levels high assessment of potential adverse safety that facilitate compliance with CAFE enough to ensure that the combined effects. That analytical approach did not standards and thereby reduce the costs U.S. passenger car and light truck fleet reflect, however, the likelihood that of compliance.427 As discussed further achieves an average fuel economy level automakers may pursue the most cost in Section X.B.1.c) below, NHTSA of not less than 35 mpg no later than effective means of improving fuel cannot consider compliance credits that MY 2020.428 And second, between MYs efficiency to comply with CAFE manufacturers earn by exceeding the 2011 and 2020, the standards must requirements. For this rulemaking, the CAFE standards and then use to achieve ‘‘increase ratably’’ in each model modeling does not limit the amount of compliance in years in which their year.429 Neither of these requirements mass reduction that is applied to any measured average fuel economy falls apply after MY 2020, so given that this segment but rather considers that below the standards. NHTSA also rulemaking concerns the standards for automakers may apply mass reduction cannot consider the use of alternative MY 2021 and after, they are not relevant based upon cost-effectiveness, similar to fuels by dual fuel vehicles nor the to this rulemaking. most other technologies. NHTSA does availability of dedicated alternative fuel (g) Other Considerations in Determining not, of course, mandate the use of any vehicles in any model year. EPCA Maximum Feasible Standards particular technology by manufacturers encourages the production of alternative in meeting the standards. The current fuel vehicles by specifying that their NHTSA has historically considered proposal, like the Draft TAR, also fuel economy is to be determined using the potential for adverse safety considers the safety effect associated a special calculation procedure that consequences in setting CAFE with the additional vehicle miles results in those vehicles being assigned standards. This practice has been traveled due to the rebound effect. a higher fuel economy level than they consistently approved in case law. As In this rulemaking, NHTSA is actually achieve. courts have recognized, ‘‘NHTSA has considering the effect of additional The effect of the prohibitions against always examined the safety expenses in fuel savings technology on considering these statutory flexibilities consequences of the CAFE standards in the affordability of vehicles—the in setting the CAFE standards is that the its overall consideration of relevant likelihood that increased standards will flexibilities remain voluntarily- factors since its earliest rulemaking result in consumers being priced out of employed measures. If NHTSA were under the CAFE program.’’ Competitive the new vehicle market and choosing to instead to assume manufacturer use of Enterprise Institute v. NHTSA, 901 F.2d keep their existing vehicle or purchase those flexibilities in setting new 107, 120 n. 11 (D.C. Cir. 1990) (‘‘CEI–I’’) a used vehicle. Since new vehicles are standards, higher standards would (citing 42 FR 33534, 33551 (June 30, significantly safer than used vehicles, appear less costly and therefore more 1977)). The courts have consistently slowing fleet turnover to newer vehicles feasible, which would thus tend to upheld NHTSA’s implementation of results in older and less safe vehicles require manufacturers to use those EPCA in this manner. See, e.g., remaining on the roads longer. This flexibilities in order to meet higher Competitive Enterprise Institute v. significantly affects the safety of the standards. By keeping NHTSA from NHTSA, 956 F.2d 321, 322 (D.C. Cir. United States light duty fleet, as including them in our stringency 1992) (‘‘CEI–II’’) (in determining the described more fully in Section 0 above determination, the provision ensures maximum feasible fuel economy and in Chapter 11 of the PRIA that these statutory credits remain true standard, ‘‘NHTSA has always taken accompanying this proposal. compliance flexibilities. passenger safety into account’’) (citing Furthermore, as fuel economy standards Additionally, for non-statutory CEI–I, 901 F.2d at 120 n. 11); become more stringent, and more fuel incentives that NHTSA developed by Competitive Enterprise Institute v. efficient vehicles are introduced into the regulation, NHTSA does not consider NHTSA, 45 F.3d 481, 482–83 (D.C. Cir. fleet, fueling costs are reduced. This these subject to the EPCA prohibition on 1995) (‘‘CEI–III’’) (same); Center for results in consumers driving more considering flexibilities, either. EPCA is Biological Diversity v. NHTSA, 538 F.3d miles, which results in more crashes very clear as to which flexibilities are 1172, 1203–04 (9th Cir. 2008) and increased highway fatalities. not to be considered. When the agency (upholding NHTSA’s analysis of vehicle 2. Administrative Procedure Act has introduced additional flexibilities safety issues associated with weight in such as A/C efficiency and ‘‘off-cycle’’ connection with the MYs 2008–2011 To be upheld under the ‘‘arbitrary and technology fuel economy improvement light truck CAFE rulemaking). Thus, in capricious’’ standard of judicial review values, NHTSA has considered those in the APA, an agency rule must be 428 49 U.S.C. 32902(b)(2)(A). rational, based on consideration of the 427 49 U.S.C. 32902(h). 429 49 U.S.C. 32902(b)(2)(C). relevant factors, and within the scope of

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00228 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43213

the authority delegated to the agency by considerations be integrated into that the 2012 final rule that established the statute. The agency must examine process.432 To accomplish that purpose, standards for MY 2021 and identified the relevant data and articulate a NEPA requires an agency to compare ‘‘augural’’ standard levels for MYs satisfactory explanation for its action the potential environmental impacts of 2022–2025. Not only do we believe that including a ‘‘rational connection its proposed action to those of a the facts before us have changed, but we between the facts found and the choice reasonable range of alternatives. believe that those facts have changed made.’’ Burlington Truck Lines, Inc., v. To explore the environmental sufficiently that the balancing of the United States, 371 U.S. 156, 168 (1962). consequences of this proposed rule in EPCA factors and other considerations Statutory interpretations included in depth, NHTSA has prepared a Draft must also change. The standards we are an agency’s rule are subject to the two- Environmental Impact Statement proposing today reflect that balancing. step analysis of Chevron, U.S.A. v. (‘‘DEIS’’). The purpose of an EIS is to The overarching purpose of EPCA is Natural Resources Defense Council, 467 ‘‘provide full and fair discussion of energy conservation; that fact remains U.S. 837 (1984). Under step one, where significant environmental impacts and the same. Examining that phrasing a statute ‘‘has directly spoken to the [to] inform decisionmakers and the afresh, Merriam-Webster states that to precise question at issue,’’ id. at 842, the public of the reasonable alternatives ‘‘conserve’’ means, in relevant part, ‘‘to court and the agency ‘‘must give effect which would avoid or minimize adverse keep in a safe or sound state; especially, to the unambiguously expressed intent impacts or enhance the quality of the to avoid wasteful or destructive use of Congress,’’ id. at 843. If the statute is human environment.’’ 433 of.’’ 434 This is consistent with our silent or ambiguous regarding the NEPA is ‘‘a procedural statute that understanding of Congress’ original specific question, the court proceeds to mandates a process rather than a intent for the CAFE program: To raise step two and asks ‘‘whether the agency’s particular result.’’ Stewart Park & fleet-wide fuel economy levels in answer is based on a permissible Reserve Coal., Inc. v. Slater, 352 F.3d response to the Arab oil embargo in the construction of the statute.’’ Id. 545, 557 (2d Cir. 2003). The agency’s 1970s and protect the country from If an agency’s interpretation differs overall EIS-related obligation is to ‘‘take further gasoline price shocks and supply from the one that it has previously a ‘hard look’ at the environmental shortages. Those price shocks, while adopted, the agency need not consequences before taking a major they were occurring, were disruptive to demonstrate that the prior position was action.’’ Baltimore Gas & Elec. Co. v. the U.S. economy and significantly wrong or even less desirable. Rather, the Natural Resources Defense Council, affected consumers’ daily lives. agency would need only to demonstrate Inc., 462 U.S. 87, 97 (1983). Congress therefore sought to keep U.S. that its new position is consistent with Significantly, ‘‘[i]f the adverse energy consumption in a safe and sound the statute and supported by the record environmental effects of the proposed state for the sake of consumers and the and acknowledge that this is a departure action are adequately identified and economy and avoid such shocks in the from past positions. The Supreme Court evaluated, the agency is not constrained future. emphasized this in FCC v. Fox by NEPA from deciding that other Today, the conditions that led both to Television, 556 U.S. 502 (2009). When values outweigh the environmental those price shocks and to U.S. energy an agency changes course from earlier costs.’’ Robertson v. Methow Valley vulnerability overall have changed regulations, ‘‘the requirement that an Citizens Council, 490 U.S. 332, 350 significantly. In the late 1970s, the U.S. agency provide a reasoned explanation (1989). was a major oil importer and changes for its action would ordinarily demand The agency must identify the (intentional or not) in the global oil that it display awareness that it is ‘‘environmentally preferable’’ supply had massive domestic changing position,’’ but ‘‘need not alternative but need not adopt it. consequences, as Congress saw. While demonstrate to a court’s satisfaction that ‘‘Congress in enacting NEPA . . . did oil consumption exceeded domestic the reasons for the new policy are better not require agencies to elevate production for many years after that, net than the reasons for the old one; it environmental concerns over other energy imports peaked in 2005, and suffices that the new policy is appropriate considerations.’’ Baltimore since then, oil imports have declined permissible under the statute, that there Gas & Elec. Co. v. Natural Resources while exports have increased. are good reasons for it, and that the Defense Council, Inc., 462 U.S. 87, 97 The relationship between the U.S. and agency believes it to be better, which the (1983). Instead, NEPA requires an the global oil market has changed for conscious change of course adequately agency to develop alternatives to the two principal reasons. The first reason indicates.’’ 430 The APA also requires proposed action in preparing an EIS. 42 is that the U.S. now consumes a that agencies provide notice and U.S.C. 4322(2)(C)(iii). The statute does significantly smaller share of global oil comment to the public when proposing not command the agency to favor an production than it did in the 1970s. At regulations,431 as we are doing today. environmentally preferable course of the time of the Arab oil embargo, the U.S. consumed about 17 million barrels 3. National Environmental Policy Act action, only that it make its decision to proceed with the action after taking a per day of the globe’s approximately 55 As discussed above, EPCA requires hard look at the environmental million barrels per day.435 While OPEC NHTSA to determine the level at which consequences. (particularly Saudi Arabia) still has the to set CAFE standards for each model We seek comment on the DEIS ability to influence global oil prices by year by considering the four factors of associated with this NPRM. imposing discretionary supply technological feasibility, economic restrictions, the greater diversity of both practicability, the effect of other motor 4. Evaluating the EPCA Factors and suppliers and consumers since the vehicle standards of the Government on Other Considerations To Arrive at the 1970s has reduced the degree to which fuel economy, and the need of the Proposed Standards United States to conserve energy. The NHTSA well recognizes that the 434 ‘‘Conserve,’’ Merriam-Webster, available at National Environmental Policy Act decision it proposes to make in today’s https://www.merriam-webster.com/dictionary/ (NEPA) directs that environmental NPRM is different from the one made in conserve (last visited June 25, 2018). 435 Short-Term Energy Outlook, U.S. Energy Information Administration (June 2018), available 430 Ibid., 1181. 432 NEPA is codified at 42 U.S.C. 4321–47. at https://www.eia.gov/outlooks/steo/pdf/steo_ 431 5 U.S.C. 553. 433 40 CFR 1502.1. full.pdf.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00229 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43214 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

a single actor (or small collection of rising global oil prices themselves made commenters raise the possibility that actors) can impact the welfare of using those technologies more feasible. even if the most productive deposits are individual consumers. Oil is a fungible As a hypothetical example, if it costs already tapped, any rises in global oil global commodity, though there are $79 per barrel to extract oil from a shale prices should spur technology limits to the substitutability of different play, when the market price for that oil development that improves output of types of crude for a given application. is $60 per barrel, it is not worth the less productive deposits.442 Moreover, The global oil market can, to a large producer’s cost to extract the oil; when even if U.S. production increases more extent, compensate for any producer the market price is $80 per barrel, it slowly than, for example, EIA currently that chooses not to sell to a given buyer becomes cost-effective. estimates, all increases in U.S. by shifting other supply toward that Recent analysis further suggests that production help to temper global prices buyer. And while regional proximity, the U.S. oil supply response to a rise in and the risk of oil shocks because they comparability of crude oil, and foreign global prices is much larger now due to reduce the influence of other producing policy considerations can make some the shale revolution, as compared to countries who might experience supply transactions more or less attractive, as what it was when U.S. production interruptions due to geopolitical long as exporters have a vested interest depended entirely on conventional instability or deliberately reduce supply in preserving the stability (both in terms wells. Unconventional wells may be not in an effort to raise prices.443 of price and supply) of the global oil only capable of producing more oil over market, coordinated, large-scale actions time but also may be capable of These changes in U.S. oil intensity, (like the multi-nation sanctions against responding faster to price shocks. One production, and capacity cannot Iran in recent years) would be required 2017 study concluded that ‘‘The long- entirely insulate consumers from the to impose costs or welfare losses on one run price responsiveness of supply is effects of price shocks at the gas pump, specific player in the global market. As about 6 times larger for tight oil on a per because although domestic production a corollary to the small rise in U.S. well basis, and about 9 times larger may be able to satisfy domestic energy petroleum consumption over the last when also accounting for the rise in demand, we cannot predict whether few decades, the oil intensity of U.S. unconventional-directed drilling.’’ That domestically produced oil will be GDP has continued to decline since the same study further found that ‘‘Given a distributed domestically or more Arab oil embargo, suggesting that U.S. price rise to $80 per barrel, U.S. oil broadly to the global market. But it GDP is less susceptible to increases in production could rise by 0.5 million appears that domestic supply may global petroleum prices (sudden or barrels per day in 6 months, 1.2 million dampen the magnitude, frequency, and otherwise) than it was at the time of in 1 year, 2 million in 2 years, and 3 duration of price shocks. As global per- EPCA’s passage or when these policies million in 5 years.’’ 439 Some analysts barrel oil prices rise, U.S. production is were last considered in 2012. While the suggest that shale drillers can respond now much better able to (and does) U.S. still has a higher energy intensity more quickly to market conditions ramp up in response, pulling those of GDP than some other developed because, unlike conventional drillers, prices back down. Corresponding per- nations, our energy intensity has been they do not need to spend years looking gallon gas prices may not fall declining since 1950 (shrinking by for new deposits, because there are overnight,444 but it is foreseeable that about 60% since 1950 and almost 30% simply so many shale oil wells being they could moderate over time and between 1990 and 2015).436 drilled, and because they are more likely respond faster than prior to the The second factor that has changed productive (although their supply may shale revolution. EIA’s Annual Energy the United States’ relationship to the be exhausted more quickly than a Outlook for 2018 acknowledges global oil market is the changing U.S. conventional well, the sheer numbers uncertainty regarding these new oil reliance on imported oil over the last appear likely to make up for that sources but projects that while retail decade. U.S. domestic oil production concern).440 Some commenters disagree prices of gasoline and diesel will began rising in 2009 with more cost- and suggest that the best deposits are increase between 2018 and 2050, annual effective drilling and production already known and tapped.441 Other average gasoline prices would not technologies.437 Domestic oil exceed $4/gallon (in real dollars) during production became more cost-effective downhole drilling motors and the invention of that timeframe under EIA’s ‘‘reference for two basic reasons. First, technology other necessary supporting equipment, materials, and technologies (particularly, downhole telemetry improved: The use of horizontal drilling equipment) had brought some applications within 442 LeBlanc, R. In the Sweet Spot: The Key to in conjunction with hydraulic fracturing the realm of commercial viability. EIA’s AEO 2018 Shale, Wall Street Journal (Mar. 6, 2018), available has greatly expanded the ability of also projects that by the early 2040s, tight oil at http://partners.wsj.com/ceraweek/connection/ producers to profitably recover natural production will account for nearly 70% of total U.S. sweet-spot-key-shale/. production, up from 54% of the U.S. total in 2017. 443 Alessi, C. & Sider, A. U.S. Oil Output Expected gas and oil from low-permeability See also, Tight oil remains the leading source of to Surpass Saudi Arabia, Rivaling Russia for Top geologic plays—particularly, shale future U.S. crude oil production, U.S. Energy Spot, Wall Street Journal (Jan. 19, 2018), available plays—and consequently, oil Information Administration (Feb. 22, 2018), https:// at https://www.wsj.com/articles/u-s-crude- www.eia.gov/todayinenergy/detail.php?id=35052. production-expected-to-surpass-saudi-arabia-in- production from shale plays has grown 439 rapidly in recent years.438 And second, Newell, R. G. & Prest, B.C. The 2018-1516352405. Unconventional Oil Supply Boom: Aggregate Price 444 To be clear, the fact that the risk of gasoline Response from Microdata, Working Paper 23973, price shocks may now be lower than in the past is 436 Today in Energy: Global energy intensity National Bureau of Economic Research (Oct. 2017), different from arguing that gasoline prices will continues to decline, U.S. Energy Information available at http://www.nber.org/papers/w23973 never rise again at all. The Energy Information Administration (July 12, 106), https://www.eia.gov/ (last visited June 25, 2018). Administration tracks and reports on pump prices todayinenergy/detail.php?id=27032. 440 Ip, G. America’s Emerging Petro Economy around the country, and we refer readers to their 437 Energy Explained, U.S. Energy Information Flips the Impact of Oil, Wall Street Journal (Feb. 21, website for the most up-to-date information. EIA Administration, https://www.eia.gov/energy 2018), available at https://www.wsj.com/articles/ projects under its ‘‘reference case’’ assumptions that explained/index.cfm (last visited June 25, 2018). americas-emerging-petro-economy-flips-the-impact- the structural changes in the oil market will keep 438 Review of Emerging Resources: U.S. Shale Gas of-oil-1519209000 (last visited June 25, 2018). prices below $4/gallon through 2050. Prices will and Shale Oil Plays, U.S. Energy Information 441 Olson, B. Shale Trailblazer Turns Skeptic on foreseeably continue to rise and fall with supply Administration (July 8, 2011), https://www.eia.gov/ Soaring U.S. Oil Production, Wall Street Journal and demand changes; the relevant question for the analysis/studies/usshalegas/. Practical application (Mar. 5, 2018), available at https://www.wsj.com/ need of the U.S. to conserve energy is not whether of horizontal drilling to oil production began in the articles/shale-trailblazer-turns-skeptic-on-soaring- there will be any movement in prices but whether early 1980s, by which time the advent of improved u-s-oil-production-1520257595. that movement is likely to be sudden and large.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00230 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43215

case’’ projection.445 The International in the 1970s when making fuel- prosperity as it was when EPCA was Energy Agency (IEA)’s Oil 2018 report efficiency trade-offs within their passed. The national balance of suggests some concern that excessive household fleets (or when replacing payments considerations are likely focus on investing in U.S. shale oil household vehicles at the time of drastically less important than they production may increase price volatility purchase). On a longer-term basis, if oil were in the 1970s, at least in terms of after 2023 if investment is not applied prices rise, consumers have more oil imports and vehicle fuel economy. more broadly but also states that U.S. options to invest in additional fuel Foreign policy considerations appear to shale oil is capable of and expected to economy when purchasing new vehicles have shifted along with the supply respond quickly to rising prices in the than at any other time in history. shifts also discussed above. future, and that American influence on Global oil demand conditions are also Whether and how environmental global oil markets is expected to different than in previous years. considerations create a need for CAFE continue to rise.446 From the supply Countries that had very small markets standards is, perhaps, more side, it is possible that the oil market for new light-duty vehicles in the 1970s complicated. As discussed earlier in this conditions that created the price shocks are now driving global production as document, carbon dioxide is a direct in the 1970s may no longer exist. their economies improve and growing byproduct of the combustion of carbon- Regardless of changes in the oil numbers of middle-class consumers are based fuels in vehicle engines.448 Many supply market, on the demand side, able to purchase vehicles for personal argue that it is likely that human conditions are also significantly use. The global increase in drivers activities, especially emissions of different from the 1970s. If gasoline inevitably affects global oil demand, greenhouse gases like carbon dioxide, prices increase suddenly and which affects oil prices. However, these contribute to the observed climate dramatically, in today’s market changes generally occur gradually over warming since the mid-20th century.449 American consumers have more options time, unlike a disruption that causes a Even taking that premise as given, it is for fuel-efficient new vehicles. Fuel- gasoline price shock. Market growth reasonable to ask whether rapid ongoing efficient vehicles were available to happens relatively gradually and is increases in CAFE stringency (or even, purchasers in the 1970s, but they were subject to many different factors. Oil for that matter, electric vehicle generally small entry-level vehicles with supply markets likely have time to mandates) can sufficiently address features that did not meet the needs and adjust to increases in demand from climate change to merit their costs. To preferences of many consumers. Today, higher vehicle sales in countries like ‘‘conserve,’’ again, means ‘‘to avoid most U.S. households maintain a China and India, and in fact, those wasteful or destructive use of.’’ household vehicle fleet that serves a increases in demand may temper global Some commenters have argued variety of purposes and represents a prices by keeping production increasing essentially that any petroleum use is variety of fuel efficiency levels. more steadily than if demand was less destructive because it all adds Manufacturers have responded to fuel certain; clear demand rewards increased incrementally to climate change. They economy standards and to consumer production and encourages additional argue that as CAFE standards increase, demand over the last decade to offer a resource development over time. It petroleum use will decrease; therefore wide array of fuel-efficient vehicles in therefore seems unlikely that growth in CAFE standard stringency should different segments and with a wide these vehicle markets could lead to increase as rapidly as possible. Other range of features. A household may now gasoline price shocks. Moreover, even as commenters, recognizing that economic respond to short-term increases in fuel these vehicle markets grow, it is practicability is also relevant, have price by shifting vehicle miles traveled possible that these and other vehicle argued essentially that because more within their household fleet away from markets may be moving away from stringent CAFE standards produce less less-efficient vehicles and toward petroleum usage under the direction of CO2 emissions, NHTSA should simply models with higher fuel economy. A their governments.447 If this occurs, set CAFE standards to increase at the similar option existed in the 1970s, global oil production will fall in most rapid of the alternative rates that though not as widely as today, and response to reduced global demand, but NHTSA cannot prove is economically vehicle owners in 2018 do not have to latent production capacity would exist impracticable. The question here, again, sacrifice as much utility as owners did to offset the impacts of unexpected is whether the additional fuel saved supply interruptions and maintain a (and CO2 emissions avoided) by more 445 Annual Energy Outlook 2018, U.S. Energy rapidly increasing CAFE standards Information Administration (Feb. 6, 2018) at 57, 58, level of global production that is available at https://www.eia.gov/outlooks/aeo/pdf/ accessible to petroleum consumers. better satisfies the U.S.’s need to avoid AEO2018.pdf. The U.S. Energy Information This, too, would seem likely to reduce destructive or wasteful use of energy Administration (EIA) is the statistical and analytical the risk of gasoline price shocks. than more moderate approaches that agency within the U.S. Department of Energy more appropiately balance other (DOE). EIA is the nation’s premier source of energy Considering all of the above factors, if information and every fuel economy rulemaking gasoline price shocks are no longer as statutory considerations. since 2002 (and every joint CAFE and CO2 much of a threat as they were when In the context of climate change, rulemaking since 2009) has applied fuel price EPCA was originally passed, it seems NHTSA believes it is hard to say that projections from EIA’s Annual Energy Outlook increasing CAFE standards is necessary (AEO). AEO projections, documentation, and reasonable to consider what the need of underlying data and estimates are available at the United States to conserve oil is to avoid destructive or wasteful use of https://www.eia.gov/outlooks/aeo/. today and going forward. Looking to the energy as compared to somewhat-less- 446 See Oil 2018: Analysis and Forecasts to 2023 discussion above on what factors are rapidly-increasing CAFE standards. The Executive Summary, International Energy Agency most stringent of the regulatory (2018), available at http://www.iea.org/Textbase/ relevant to the need of the United States npsum/oil2018MRSsum.pdf (last visited June 25, to conserve oil, one may conclude that 2018). See also Kent, S. & Puko, T. U.S. Will Be the the U.S. is no longer as dependent upon 448 Depending on the energy source, it may also World’s Largest Oil Producer by 2023, Says IEA, be a byproduct of consumption of electricity by Wall Street Journal (Mar. 5, 2018), available at petroleum as the engine of economic vehicles. https://www.wsj.com/articles/u-s-will-be-the- 449 Climate Science Special Report: Fourth worlds-largest-oil-producer-by-2023-says-iea- 447 Lynes, M. Plug-in electric vehicles: future National Climate Assessment, Volume I (Wuebbles, 1520236810 (reporting on remarks at the 2018 market conditions and adoption rates, U.S. Energy D.J. et al., eds. 2017), available at https:// CERAWeek energy conference by IEA Executive Information Administration (Oct. 23, 2017), https:// science2017.globalchange.gov/ (last accessed Feb. Director Fatih Birol). www.eia.gov/outlooks/ieo/pev.php. 23, 2018).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00231 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43216 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

alternatives considered in the 2012 final much more likely today that oil prices vehicles. Any of the alternatives could rule and FRIA (under much more will rise only moderately in the future thus be achieved on a technical basis optimistic assumptions about and that price shocks are less likely. alone but only if the level of resources technology effectiveness), which would Accordingly, it is reasonable to believe that might be required to implement the have required a seven percent average that U.S. consumers value future fuel technologies is not considered. annual fleetwide increase in fuel savings accurately and choose new However, as discussed above, we no economy for MYs 2017–2025 compared vehicles based on that view. This is longer view the need of the U.S. to to MY 2016 standards, was forecast to particularly true, since Federal law conserve energy as nearly infinite, only decrease global temperatures in requires that new vehicles be posted which means that it no longer combines 2100 by 0.02 °C in 2100. Under with a window sticker providing with boundless technological feasibility NHTSA’s current proposal, we estimated costs or savings over a five to quickly push stringency upward. anticipate that global temperatures year period compared to average new The effect of other motor vehicle would increase by 0.003 °C in 2100 vehicles.451 Even if consumers do not standards of the Government on fuel compared to the augural standards. As explicitly think to themselves ‘‘this new economy is similarly not limiting during reported in NHTSA’s Draft EIS, car will save me $5,000 in fuel costs this rulemaking time frame. As compared to the average global mean over its lifetime compared to that other discussed above, the analysis projects surface temperature for 1986–2005, new car,’’ gradual and relatively that safety standards will add some global surface temperatures are still predictable fuel price increases in the mass to new vehicles during this time forecast to increase by 3.484–3.487 °C, foreseeable future allow consumers to frame and accounts for Tier 3 depending on the alternative. Because roughly estimate the comparative value compliance in estimates of technology the impacts of any standards are small, of fuel savings among vehicles and effectiveness, but neither of these things and in fact several-orders-of-magnitude choose the amount of fuel savings that appear likely to make it significantly smaller, as compared to the overall they want, in light of the other vehicle harder for industry to comply with more forecast increases, this makes it hard for attributes they value. It seems, then, that stringent CAFE standards. In terms of NHTSA to conclude that the climate consumer cost as an element of the need EPA’s GHG standards, as also discussed change effects potentially attributable to of the U.S. to conserve energy is also above, NHTSA and EPA’s coordination the additional energy used, even over less urgent in the context of the in this proposal should make the two the full lifetimes of the vehicles in structural changes in oil markets over sets of standards similarly binding, question, is ‘‘destructive or wasteful’’ the last several years. although differences in compliance enough that the ‘‘need of the U.S. to Given the discussion above, NHTSA provisions remain such that which conserve energy’’ requires NHTSA to tentatively concludes that the need of standards are more binding will vary place an outsized emphasis on this the U.S. to conserve energy may no somewhat between manufacturers and consideration as opposed to others.450 longer function as assumed in previous over time. Consumer costs are the remaining considerations of what CAFE standards The remaining factor to consider is issue considered in the context of the would be maximum feasible. The economic practicability. NHTSA has need of the U.S. to conserve energy. overall risks associated with the need of typically defined economic NHTSA has argued in the past, the U.S. to conserve oil have entered a practicability, as discussed above, as somewhat paternalistically, that CAFE new paradigm with the risks whether a given CAFE standard is standards help to solve consumers’ substantially lower today and projected ‘‘within the financial capability of the ‘‘myopia’’ about the value of fuel into the future than when CAFE industry but not so stringent as’’ to lead savings they could receive, when buying standards were first issued and in the to ‘‘adverse economic consequences, a new vehicle if they chose a more fuel- recent past. The effectiveness of CAFE such as a significant loss of jobs or efficient model. There has been standards in reducing the demand for unreasonable elimination of consumer extensive debate over how much fuel combined with the increase in choice.’’ As part of that definition, consumers do (and/or should) value fuel domestic oil production have NHTSA looks at a variety of elements savings and fuel economy as an attribute contributed significantly to the current that can lead to adverse economic in new vehicles, and that debate is situation and outlook for the near- and consequences. All of the alternatives addressed in Section II.E. For purposes mid-term future. The world has considered today arguably raise of considering the need of the U.S. to changed, and the need of the U.S. to economic practicability issues. NHTSA conserve energy, the question of conserve energy, at least in the context believes there could be potential for consumer costs may be closer to of the CAFE program, has also changed. unreasonable elimination of consumer whether U.S. consumers so need to save Of the other factors under 32902(g), choice, loss of U.S. jobs, and a number money on fuel that they must be the changes are perhaps less significant. of adverse economic consequences required to save substantially more fuel We continue to believe that under nearly all if not all of the (through purchasing a new vehicle technological feasibility, per se, is not regulatory alternatives considered made more fuel-efficient by more limiting during this rulemaking time today. stringent CAFE standards) than they frame. The technologies considered in If a potential CAFE standard requires would otherwise choose. this analysis either are already in manufacturers to add technology to new Again, when EPCA originally passed, commercial production or likely will be vehicles that consumers do not want, or Congress was trying to protect U.S. by MY 2021—some at great expense. to skip adding technology to new consumers from the negative effects of Based on our analysis, all of the vehicles that consumers do want, it another gasoline price shock. It appears alternatives appear as though they could would seem to present issues with narrowly be considered technologically elimination of consumer choice. 450 The question of whether or how rapidly to feasible, in that they could be achieved Depending on the extent and expense of increase CAFE stringency is different from the based on the existence or the projected required fuel saving technology, that question of whether to set CAFE standards at all. future existence of technologies that elimination of consumer choice could Massachusetts v. EPA, 549 U.S. 497 (2007) (‘‘Agencies, like legislatures, do not generally could be incorporated on future be unreasonable. resolve massive problems in one fell regulatory When deciding on which new vehicle swoop.’’) 451 49 CFR 575.401; 40 CFR 600.302–12. to purchase, American consumers

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00232 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43217

generally tend not to be interested in can increase sales of vehicles that do not fuel economy over other attributes, In better fuel economy above other help manufacturers meet the standards addition, EIA’s latest AEO 2018 attributes, particularly when gasoline (such as vehicles that fall significantly suggests, based on current assumptions, prices are low.452 Manufacturers have short of their fuel economy targets, due that per-gallon prices are likely to stay repeatedly indicated to the agencies that to higher levels of performance (e.g., under $4 through 2050.455 It therefore new vehicle buyers are only willing to larger, less efficient engines) or other seems unlikely that consumer pay for fuel economy-improving features). Over the last several years, preferences are going to change technology if it pays back within the despite record sales overall, most dramatically in the foreseeable future first two to three years of vehicle manufacturers have been managing their and certainly not within the time frame ownership.453 NHTSA has therefore CAFE compliance obligations through of the standards covered by this incorporated this assumption (of use of credits,454 because many proposal. willingness to pay for technology that consumers have chosen to buy vehicles Thus, if manufacturers are not pays back within 30 months) into that do not improve manufacturers’ currently able to sell higher-fuel today’s analysis. As a result, NHTSA’s compliance positions. analysis finds that the most cost- economy vehicles without heavy Consumer decisions to purchase subsidization, particularly HEVs, effective technology is applied with or relatively low-fuel economy vehicles PHEVs, and EVs, it seems unlikely that without CAFE (or CO2) standards, might seem irrational if gasoline prices their ability to do so will improve diminishing somewhat the incremental were expected to rebound in the future, unless consumer preferences change or cost-effectiveness of new CAFE but current indicators suggest this is not fuel prices rise significantly, either of standards. particularly likely. Although we know which seem unlikely. Today’s analysis Consumers not being interested in of no clear ‘‘tipping point’’ for gasoline indicates, perhaps predictably, that better fuel economy can take two forms: prices at which American consumers electrification rates must increase as First, it can dampen sales of vehicles suddenly become more interested in with the additional technology required stringency increases among the options the agencies are considering. to meet the standards, and second, it 454 See CAFE Public Information Center, National Highway Traffic Safety Administration, https:// 452 See, e.g., Comment by Global Automakers, one.nhtsa.gov/cafe_pic/CAFE_PIC_Mfr_LIVE.html 455 As noted elsewhere in this proposal, the Docket ID NHTSA–2016–0068–0062 (citing a 2014 (last visited June 25, 2018). Readers can examine agencies based analysis on AEO 2017 projections of, study by Strategic Vision that found that ‘‘. . . achieved versus required fuel economy by model for instance, fuel prices, as it was the best available generally, customers as a whole place a higher year and by individual manufacturer or by entire information at the time the analysis was conducted. priority on handling and ride than fuel economy.’’). fleets. When a manufacturer’s achieved fuel As such, where possible, the agency incorporated 453 This is supported by the 2015 NAS study, economy falls short of required fuel economy but latest AEO 2018 projections into the discussion, in which found that consumers seek to recoup added the manufacturer has not paid civil penalties, the effort to re-confirm no discernible impact to upfront purchasing costs within two or three years. manufacturer is using credits somehow to make up analysis results or to provide the best possible See 2015 NAS Report, at pg. 317. the shortfall. information for the discussion.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00233 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43218 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00234 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.157 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43219

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00235 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.158 43220 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00236 Fmt 4701 Sfmt 4725 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.159 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43221

Manufacturers have commented to the offering more of these models every options, sales of these vehicles are not agencies that ‘‘Although automakers are year, with improved technology and growing,’’ noting that even for hybrid

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00237 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.160 43222 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

vehicles, which require no adaptation additional costs projected for HEVs, vehicle, or a pickup made by a different by consumers (for example, to range PHEVs, and EVs. This trend may well manufacturer rather than pay the extra limits or refueling by charging), sales extend beyond electrification cost that the manufacturer is trying to ‘‘have declined from a peak of a 3.1 technologies to other technologies. recoup from higher-fuel economy percent share of the market (in 2013) to When costs for fuel economy-improving vehicles that had to be artificially . . . 1.8 percent [in 2016].’’ 456 The same technology exceed the fuel savings, discounted to be sold. We seek source further stated that this decline consumers may very well be unwilling comment on the effect of fuel economy was despite the technology being to pay the full cost for vehicles with standards on cross-subsidization across available in the market for more than 15 higher fuel economy that would be models. years, and that in 2016, ‘‘close to 75 increasingly needed as to comply as the Moreover, assuming that percent of the people who have traded stringency of the alternatives increases. manufacturers try to pass the costs of in a hybrid or electric car to a dealer If consumers are not willing to pay those technologies on to consumers in have replaced it with a conventional the full cost for vehicles with higher the form of higher new vehicle prices, (non-hybrid) gasoline-powered car.’’ 457 fuel economy, it seems reasonably rather than absorbing them and hurting While some consumers continue to seek foreseeable that they will consider profitability, this can affect consumers’ out hybrid and electric vehicles, then, vehicles made more expensive by higher ability to afford new vehicles. The many other consumers seem CAFE standards to be not ‘‘available’’ to analysis assumes that the increased cost uninterested in them, even given the them to purchase—or put more simply, of meeting standards is passed on to generous incentives and subsidies often that they will be turned off by more consumers through higher new vehicle available for consumers in the form of expensive vehicles with technologies prices, and looks at those increases as a tax credits, government rebates, High they do not want, and seek instead to one-time payment. In the context of, for Occupancy Vehicle Lane access, purchase cheaper vehicles without that example, a $30,000 new vehicle, preferred and/or subsidized parking, technology (or with different another $2,000 may not seem significant among others. Despite this broad technologies, such as those that improve to some readers. Yet manufacturers and ongoing lack of consumer interest, a performance or safety). Manufacturers dealers have repeatedly commented to number of manufacturers nonetheless have long cross-subsidized vehicle NHTSA that the overall price of the models in their lineups in order to continue to increase their offerings of vehicle is less relevant to the majority recoup costs in cases where they do not these vehicles. At best, this trend seems of consumers than the monthly payment believe they can pass the full costs of economically inefficient; more amount, which is a significant factor in development and production forward as concerningly for economic consumers’ ability to purchase or lease price increases for the vehicle model in practicability, it seems likely to impact a new vehicle. Amortizing a $2,000 question. Given that this cross- consumer choice (as discussed further price increase over, for example, 48 subsidization is ongoing, however, and below) in ways that could weigh heavily months may also not seem like a large possibly deepening as manufacturers on sales, jobs, and consumers amount to some readers, even have had to meet increasingly stringent themselves. We seek comment on this accounting for interest payments. Yet issue. CAFE standards over the past several years, it is unclear how much additional the corresponding up-front and monthly If the evidence indicates that hybrid costs may pose a challenge to low- sales are declining as gasoline prices distribution of costs could be supported by the market. Certainly, if CAFE income or credit-challenged purchasers. remain low, it seems reasonable to standards continue to increase in As discussed previously, such increased conclude that consumers will not stringency as gasoline prices stay costs will price many consumers out of choose to buy more of them going relatively low and consumer willingness the market—leaving them to continue forward as gasoline prices are forecast to to pay for significant additional fuel driving an older, less safe, less efficient, remain low. This is consistent with the economy improvements remains and more polluting vehicle, or analysis discussed in Section II.E, that correspondingly low, then additional purchasing another used vehicle that even while some consumers may be cross-subsidization of products to try to would likewise be less safe, less willing to pay between $2,000 and ease those products into consumer efficient, and more polluting than an $3,000 more for vehicles with electrified acceptance seems likely to impair equivalent new vehicle. technologies, that incremental consumer choice, insofar as the vehicles For example, the average MY 2025 willingness-to-pay falls well short of the they want to buy will cost more and prices estimated here under the baseline may have technology for which they are and proposed CAFE standards are about 456 Comment by Global Automakers, Docket ID $34,800 and $32,750, respectively (and NHTSA–2016–0068–0062, citing IHS Global New unwilling to pay. Models that have Vehicle Registration Data for 2013, 2015, and historically been able to bear higher $34,500 and $32,550 under the baseline January–June 2016. percentages of the cross-subsidization and proposed GHG standards). The 457 Id. at B–6 and B–7, citing Matt Richtel, burden may not be able to bear much buyer of a new MY 2025 vehicle might American Drivers Regain Appetite for Gas Guzzlers, thus avoid the following purchase and New York Times (June 24, 2016), https:// more—a pickup truck buyer, for www.nytimes.com/2016/06/28/science/cars-gas- example, may eventually decide to first-year ownership costs under the global-warming.html. purchase a used vehicle, another type of proposed standards:

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00238 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43223

While the buyer of the average vehicle more and more difficult for foreseeably increase, and the ability to would also purchase somewhat more manufacturers and dealers to continue offset such increases by extending fuel under the proposed standards, this creating loan terms that both keep finance terms to account for increased difference might average only five monthly payments low and do not finance charges and vehicle prices due gallons per month during the first year result in consumers still owing to CAFE standards is limited by the fact of ownership.462 Some purchasers may significant amounts of money on the that doing so increases the amount of consider it more important to avoid vehicle by the time they can be expected time before consumers will have these very certain (e.g., being reflected to be ready for a new vehicle. positive equity in their vehicles (and Over the last decade, as vehicle sales in signed contracts) cost savings than able to trade in the vehicle for a newer have rebounded in the wake of the the comparatively uncertain (because, model). This reduces the mechanisms recession, historically low interest rates e.g., some owners drive considerably and increases in the average duration of that manufacturers, captive finance less than others, and may purchase fuel financing terms have helped companies, dealers, and independent in small increments as needed) fuel manufacturers and dealers keep lenders have in order to maintain sales savings. For some low-income consumers’ monthly payments low. at comparable levels. In other words, if purchasers or credit-challenged These trends (low interest rates and vehicle sales have not already hit the purchasers, the cost savings may make longer loan periods), along with pent-up breaking point, they may be close.463 the difference between being able or not demand for new vehicles, have helped The agencies seek comment on the to purchase the desired vehicle. As keep vehicle sales high. As interest rates impact that increased prices, interest vehicles get more expensive in response have increased, and most predict will rates, and financing terms are likely to to higher CAFE standards, it will get continue to rise, monthly payments will have on the new vehicle market.

458 Using down payment assumption of $4,056. 461 Using average rate of 1.83% (discussed above (‘‘Current sales are a poor predictor of future sales. See Press Release, Edmunds, New Vehicle Prices in Section II.E). Many of the macroeconomic factors that have Climb to All-Time High in December (Jan. 3, 2018), 462 Based on estimated sales volumes and average contributed to the current boom may not exist six https://www.edmunds.com/about/press/new- fuel consumption discussed below in Section VI, to nine years into the future [i.e., during the mid- vehicle-prices-climb-to-all-time-high-in- and on average vehicle survival and mileage 2020s]. The low interest loans and extended time accumulation rates (discussed above in Section II.E) december.html. loans that are now readily available may not be indicating that the average vehicle delivers about 459 Using average rate of 5.46% (discussed above 11% of it lifetime service (i.e., distance driven) available then. The automotive industry is a in Section II.E). during the first year of operation. cyclical business, and it appears to be near the top 460 Using average rate of 4.25% (discussed above 463 See, e.g., Comment by Global Automakers, of a cycle now.’’) in Section II.E). Docket ID NHTSA–2016–0068–0062, at 10

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00239 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.161 43224 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

The increasing risk that turn to the used vehicle market, where model. As Section VII.C on today’s sales manufacturers and dealers will hit a levels of safety may not be comparable. and employment analysis indicates, the wall in their ability to keep monthly We seek comment on these sales function of the CAFE model payments low may fall considerations. appears fairly accurate at predicting disproportionately on new and low- Alternatively, rather than or in sales trends but does not presume that income buyers. To build on the addition to continuing to cross- sales are particularly responsive to discussion above, manufacturers often subsidize fuel economy improvements changes in vehicle price. We are purposely cross-subsidize the prices of that consumers are unwilling to pay for concerned, however, that the sales entry-level vehicles to keep monthly directly, manufacturers may choose to model as it currently functions may payments low and attract new and try to improve their compliance position miss two key points about potential young consumers to their brand. Higher under higher CAFE standards by future sales and employment effects. CAFE standard stringency leads to restricting sales of certain vehicle First, the analysis does not account higher costs for technology across models or options. If consumers tend to for the risk discussed above that manufacturers’ fleets, meaning that want the 6-cylinder engine version of a manufacturers and dealers may not be more and more cross-subsidization vehicle rather than the 4-cylinder able to continue keeping monthly new becomes necessary to maintain version, for example, the manufacturer vehicle payments low, for a variety of affordability for entry-level vehicle may choose to make fewer 6-cylinders reasons. Interest rates and inflation may purchasers. While this is clearly an available. This solution, if chosen, rise; further lengthening loan terms may economic issue for industry, it may also would directly impact consumer choice. not be practical as they increase the slow fleet-wide improvement in vehicle It seems increasingly likely that this period of time that the purchaser has characteristics like safety—both in terms solution could be chosen as CAFE negative equity (which has secondary of manufacturers having to divert stringency increases. impacts described above). While these resources to adding technology to In terms of risks to employment, may be not-entirely-negative things for vehicles that consumers do not want today’s analysis focuses on employment the economy as a whole, they would and then figuring out how to get as a function of estimated changes in create negative pressure on vehicle sales consumers to buy them and in terms of vehicle price in response to different or employment associated with those new vehicles potentially becoming levels of standards and assumes that all sales. unaffordable for certain groups of cost increases to vehicle models are Second, as the cost of compliance consumers, meaning that they must passed forward to consumers in the increases with CAFE stringency, it is either defer new vehicle purchases or form of price increases for that vehicle possible that manufacturers may shift

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00240 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.162 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43225

production overseas to locations where losing sales by absorbing the additional all vehicles covered by the rulemaking labor is cheaper. The CAFE program costs of meeting higher CAFE standards, time frame), higher stringency standards contains no mandates with regard to it is foreseeable that absorbing those are increasingly less cost-beneficial. As where vehicles are manufactured and costs would hurt company profits. If shown and discussed in Section VII.C, arguably disincentivizes domestic manufacturers choose that approach the analysis of consumer impacts shows production of passenger cars through year after year to avoid losing market that consumers recoup only a portion of the minimum domestic passenger car share, continued falling profits would the costs associated with increasing standard. If it becomes substantially lead to negative earnings reports and stringency under all of the alternatives. more expensive for manufacturers to risks to companies’ long-term viability. The fuel savings resulting from each of meet their CAFE obligations, they may Thus, even if sales levels are maintained the alternatives is substantially less than seek to cut costs wherever they can, despite higher standards, it seems the costs associated with the alternative, which could include layoffs or changing possible that industry could face meaning that net savings for consumers production locations. adverse economic consequences. There may be other adverse economic More broadly, when gasoline prices improves as stringency decreases. consequences besides those discussed stay relatively low (as they are expected Figure V–2 below illustrates this above. If manufacturers seek to avoid to remain through the lifetime of nearly trend.464

We recognize that this is a • In 2012, we assumed in the main analysis fleet reflected. When significantly different analytical result analysis that manufacturers would add no technology is used by the analysis fleet, from the 2012 final rule, which showed more technology than needed for it is ‘‘unavailable’’ to be used again for compliance, while today’s analysis assumes the opposite trend. Using the compliance with future standards logically that manufacturers will add because the same technology cannot be projections available to the agencies for technologies that pay for themselves within the 2012 rulemaking, all of the 2.5 years, consistent with manufacturer used twice (once by a manufacturer for alternatives considered in that information on payback period. its own reasons and then again by the rulemaking were projected to have net • In 2012, we measured impacts of the model to simulate manufacturer savings to consumers and to society post-2017 standards relative to compliance responses to higher standards). Some of with pre-2017 standards, which meant that a this would happen necessarily in an overall, and those net savings improved lot of cost-effective technology attributable to updated rulemaking because a later-in- as stringency increased. Put simply, the the 2017–2020 standards was ‘‘counted’’ time analysis fleet inevitably includes result is different today from what it toward the 2025 standards. • more technology; in this particular case, was in 2012 because the facts and the In 2012, we used analysis fleets based on 2016 happened to be a somewhat analysis are also different. While the 2008 or 2010 technology. Today’s analysis technology-heavy year, and 2008 and uses a 2016-based analysis fleet. differences in the facts and the analysis 2010 (the fleets used in 2012) arguably are described extensively in Section II These three points above mean that, did not reflect the state of technology in above and in the PRIA accompanying overall, the current analysis fleet reflects 2012 well. this proposal, a few noteworthy ones the application of much additional Furthermore, readers should note the include: technology than the 2012-final-rule following changes:

464 For the reader’s reference, Alternatives 3 and unchanged. Phasing out these procedures increases to leaving the procedures unchanged, net savings to 7 phase out A/C and off-cycle procedures, while the compliance costs and reduces net savings relative consumer with seven percent discount rate. other alternatives leave those procedures

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00241 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 EP24AU18.163 43226 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

• Estimates of effectiveness and cost are fulfill EPCA’s overarching purpose of motor vehicle standards of the different for a number of technologies, as energy conservation in light of the facts Government is minimal when the two discussed in Section II above and in Chapter before the agency today and as we agencies regulating the same aspects of 6 of the PRIA, and indirect costs are expect them to be in the rulemaking vehicle performance are working determined using the RPE rather than the time frame. In the 2012 final rule that together to develop those regulations. ICM; • Fuel prices forecasts are considerably established CAFE standards for MYs Therefore, NHTSA views the lower in AEO 2017 than they were in AEO 2017–2021, and presented augural determination of maximum feasible 2012; CAFE standards for MYs 2022–2025, standards as a question of the • The current analysis uses a rebound NHTSA stated that ‘‘maximum feasible appropriateness of standards given that effect value of 20% instead of 10%; standards would be represented by the their need—either from the societal- • The current analysis newly accounts for mpg levels that we could require of the benefits perspective in terms of risk price impacts on fleet turnover; industry before we reach a tipping point associated with gasoline price shocks or • The social cost of carbon is different and that presents risk of significantly other related catastrophes, or from the accounts only for domestic (not adverse economic consequences.’’ 467 private-benefits perspective in terms of international) impacts; However, the context of that rulemaking consumer willingness to purchase new • The current analysis does not attempt to purposely limit the appearance of potential was meaningfully different from the vehicles with expensive technologies safety effects, and the value of a statistical current context. At that time, NHTSA that may allow them to save money on life is higher than in 2012. understood the need of the U.S. to future fuel purchases—seems likely to conserve energy as necessarily pushing remain low for the foreseeable future. All of these changes, together, mean the agency toward setting stricter and When determining the maximum that the standards under any of the stricter standards. Combining a then- feasible standards, and in particular the regulatory alternatives (compared to the paramount need of the U.S. to conserve economic practicability of higher preferred alternative) are more energy with the perception that standards, we also note that the expensive and have lower benefits than technological feasibility should no proposed standards have the most if they had been calculated using the longer be seen as an important limiting positive effect on on-road safety as inputs and assumptions of the 2012 factor, NHTSA then concluded that only compared to the alternatives considered. analysis. This, in turn, helps lead the significant economic harm would be a The analysis indicates that, compared to agency to a different conclusion about basis for controlling the pace at which the baseline standards defining the No- what standards might be maximum CAFE stringency increased over time. Action alternative, any regulatory feasible in the model years covered by Today, the relative importance of the alternatives under consideration would the rulemaking. NHTSA has thus both need of the U.S. to conserve energy has improve overall highway safety. Some relied on new facts and circumstances changed when compared to the of this estimated reduction is in developing today’s proposal and beginning of the CAFE program and a attributable to vehicles, themselves, reasonably rejected prior facts and great deal even since the 2012 being generally safer if they do not analyses relied on in the 2012 final rulemaking. As discussed above, the apply as much mass reduction to rule.465 effectiveness of CAFE standards in passenger cars as might be applied By directing NHTSA to determine reducing the demand for fuel combined under the baseline standards. maximum feasible standards by with the increase in domestic oil Additionally, the analysis estimates that considering the four factors, Congress production have contributed the alternatives to the baseline recognized that ‘‘maximum feasible’’ significantly to the current situation and standards would cause the fleet to turn may change over time as the agency outlook for the near- and mid-term over to newer and safer vehicles, which assessed the relative importance of each future. The world has changed, and the will also be more fuel efficient than the factor.466 If one factor appears to be need of the U.S. to conserve energy may vehicles being replaced, more quickly more important than the others in the no longer disproportionately outhweigh than otherwise anticipated. time frame to be covered by the other statutorily-mandated Furthermore, the analysis estimates that standards, it makes sense to give it more considerations such as economic the alternatives to the baseline standard weight in the agency’s determination of practicability—even when considering would involve reduced overall demand maximum feasible standards for those fuel savings from potentially more- for highway travel. As discussed above model years. If no factor appears to be stringent standards. in Section II.F, and in Chapter 11 of the particularly paramount, it makes sense Thus, while more stringent standards accompanying PRIA, most of the to determine maximum feasible may be possible, insofar as production- estimated overall improvement in standards by more generally weighing ready technology exists that the highway safety from this proposal is each factor, as long as EPCA’s direction industry could physically employ to attributable to reduced travel demand to establish maximum feasible standards reach higher standards, it is not clear (attributable to the rebound effect) and continues to be fulfilled in a manner that higher standards are now accelerated turnover to safer vehicles. that does not undermine energy economically practicable in light of The trend in these results is clear, with conservation. current U.S. consumer needs to the less stringent alternatives producing NHTSA tentatively concludes that conserve energy. While vehicles can be the greatest estimated improvement in proposing CAFE standards that hold the built with advanced fuel economy- highway safety and the proposed MY 2020 curves for passenger cars and improving technology, this does not standards producing the most favorable light trucks constant through MY 2026 mean that consumers will buy the new outcomes from a highway safety would be the maximum feasible vehicles that might be required to perspective. These considerations standards for those fleets and would include such technology; that industry bolster our determination that the could continue to subsidize their proposed standards are maximum 465 See Fox v. FCC, 556 U.S. at 514–515; see also production and sale; or that adverse feasible based upon current and NAHB v. EPA, 682 F.3d 1032 (D.C. Cir. 2012). economic consequences would not projected technology for the model 466 If this were not accurate, it seems illogical that Congress would have, at various times, set specific result from doing so. The effect of other years in question. mpg goals for the CAFE program (e.g., 35 mpg by Standards that retain the MY 2020 2020). 467 77 FR 62624, 63039 (Oct. 15, 2012). curves through MY 2026 will save fuel

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00242 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43227

beyond what the market would achieve CAFE standard and in the ‘‘flat’’ reasonably be anticipated to endanger on its own for vehicles manufactured portions of the footprint curves at the public health or welfare.’’ If EPA makes during the rulemaking time frame and larger-footprint end, designing an the appropriate endangerment and will result in the highest net benefits appropriate backstop was likely to be cause or contribute findings, then both for consumers and for society. fairly complex and likely to undermine section 202(a) authorizes EPA to issue Such standards would avoid the risks Congress’ objective in requiring standards applicable to emissions of identified in the discussion of economic attribute-based standards. See, those pollutants. Indeed, EPA’s practicability for more stringent particularly, 75 FR at 25369–70 (May 7, obligation to do so is mandatory: standards and are consistent with the 2010). Coalition for Responsible Regulation, relatively lower need of the United As in 2010, NHTSA believes that 684 F.3d at 114; Massachusetts v. EPA, States to conserve energy and the additional backstop standards are not 549 U.S. at 533. Moreover, EPA’s impact that has on consumer choice. necessary. The current proposal is based mandatory legal duty to promulgate Moreover, as the fuel economy of the on the agency’s best current these emission standards derives from new vehicle fleet improves over time, understanding of the need of the U.S. to ‘‘a statutory obligation wholly the marginal benefits of continued conserve energy now and going forward, independent of DOT’s mandate to improvements diminish, making the in light of changed circumstances and promote energy efficiency.’’ consumer willingness to bear them and balanced against the other EPCA factors. Massachusetts, 549 U.S. at 532. the economic practicability of them We seek comment on how an additional Consequently, EPA has no discretion to diminish. It is much more expensive, backstop standard might be constructed decline to issue greenhouse standards and saves much less fuel, for a vehicle that addresses the concerns raised in the under section 202(a) or to defer issuing to improve from 40 to 50 mpg, than for 2010 final rule and that also does not such standards due to NHTSA’s a vehicle to improve from 15 to 20 obviate the agency’s assessment of what regulatory authority to establish fuel mpg.468 If obtaining the marginal CAFE levels would be maximum economy standards. Rather, ‘‘[j]ust as benefits of new cars and their fuel feasible. EPA lacks authority to refuse to regulate economy technologies becomes too We seek comment on all aspects of on the grounds of NHTSA’s regulatory expensive for consumers, some the above discussion. authority, EPA cannot defer regulation consumers will choose to drive less on that basis.’’ Coalition for Responsible B. EPA’s Statutory Obligations and Why efficient used vehicles longer. Regulation, 684 F.3d at 127. the Proposed Standards Appear To Be NHTSA recognizes that the Ninth Appropriate and Reasonable Any standards under CAA section Circuit has previously held that NHTSA 202(a)(1) ‘‘shall be applicable to such must consider whether a ‘‘backstop’’ is 1. Basis for the CO2 Standards Under vehicles . . . for their useful life.’’ necessary for the CAFE standards based Section 202(a) of the Clean Air Act Emission standards set by the EPA on the EPCA factors in 49 U.S.C. Title II of the Clean Air Act (CAA) under CAA section 202(a)(1) are 32902(f), given that the overarching provides for comprehensive regulation technology-based, as the levels chosen purpose of EPCA is energy must be premised on a finding of 469 of mobile sources, authorizing EPA to conservation. NHTSA and EPA regulate emissions of air pollutants from technological feasibility. Thus, discussed the concept of backstops in all mobile source categories. Under standards promulgated under CAA the context of the modern CAFE Section 202(a) 470 and relevant case law, section 202(a) are to take effect only program (as opposed to the CAFE as discussed below, EPA considers such after providing ‘‘such period as the program at issue in the Ninth Circuit issues as technology effectiveness, its Administrator finds necessary to permit decision) in the 2010 final rule cost (both per vehicle, per manufacturer, the development and application of the establishing CAFE and GHG standards and per consumer), the lead time requisite technology, giving appropriate for MYs 2012–2016. In that document, necessary to implement the technology, consideration to the cost of compliance the agencies explained that even if the and based on this the feasibility and within such period’’ (CAA section 202 statute did not preclude a backstop practicability of potential standards; the (a)(2); see also NRDC v. EPA, 655 F. 2d beyond what was already provided for impacts of potential standards on 318, 322 (D.C. Cir. 1981)). EPA must in the minimum domestic passenger car emissions reductions of both GHGs and consider costs to those entities which non-GHGs; the impacts of standards on are directly subject to the standards. 468 As the base level of fuel economy improves, Motor & Equipment Mfrs. Ass’n Inc. v. there are fewer gallons to be saved from improving oil conservation and energy security; the further. A typical assumption is that vehicles are impacts of standards on fuel savings by EPA, 627 F. 2d 1095, 1118 (D.C. Cir. driven 15,000 miles per year. A vehicle that consumers; the impacts of standards on 1979). Thus, ‘‘the [s]ection 202(a)(2) improves from 30 mpg to 40 mpg reduces its annual the auto industry; other energy impacts; reference to compliance costs fuel consumption from 500 gallons/year to 375 encompasses only the cost to the motor- gallons/year at 15,000 miles/year or by 125 gallons. as well as other relevant factors such as A vehicle that improves from 15 mpg to 20 mpg, impacts on safety. vehicle industry to come into on the other hand, reduces its annual fuel This proposed rule would implement compliance with the new emission consumption from 1,000 gallons/year to 750 a specific provision from Title II, section standards.’’ Coalition for Responsible gallons/year—twice as much as the first example, 471 Regulation, 684 F.3d at 128; see also id. even though the mpg improvement is only half as 202(a). Section 202(a)(1) of the Clean large. Going from 40 to 50 mpg would save only 75 Air Act (CAA) states that ‘‘the at 126–27 (rejecting arguments that EPA gallons/year at 15,000 miles/year. If fuel prices are Administrator shall by regulation was required to consider or should have high, the value of those gallons may be sufficient prescribe (and from time to time revise) considered costs to other entities, such to offset the cost of improving further, but (1) EIA as stationary sources, which are not does not currently anticipate particularly high fuel . . . standards applicable to the prices in the foreseeable future, and (2) as the emission of any air pollutant from any directly subject to the emission baseline level of fuel economy continues to class or classes of new motor vehicles standards). EPA is afforded considerable increase, the marginal cost of the next gallon saved . . . , which in his judgment cause, or discretion under section 202(a) when similarly increases with the cost of the technologies assessing issues of technical feasibility required to meet the savings. contribute to, air pollution which may 469 CBD v. NHTSA, 508 F.3d 508, 537 (9th Cir. and availability of lead time to 2007), opinion vacated and superseded on denial of 470 42 U.S.C. 7521(a). implement new technology. Such reh’g, 538 F.3d 1172 (9th Cir. 2008). 471 42 U.S.C. 7521(a). determinations are ‘‘subject to the

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00243 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43228 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

restraints of reasonableness,’’ which F.3d 195 (D.C. Cir. 2001) (upholding EPA’s defined as the aggregate group of these ‘‘does not open the door to ‘crystal ball’ promulgation of technology-based standards same six greenhouse gases from new inquiry.’’ NRDC, 655 F. 2d at 328 for small non-road engines under section motor vehicles and new motor vehicle (quoting International Harvester Co. v. 213(a)(3) of the CAA). However, EPA is not engines contribute to air pollution. As a compelled under section 231 to obtain the Ruckelshaus, 478 F. 2d 615, 629 (D.C. ‘‘greatest degree of emission reduction result of these findings, section 202(a) Cir. 1973)). In developing such achievable’’ as per sections 213 and 202 of requires EPA to issue standards technology-based standards, EPA has the CAA, and so EPA does not interpret the applicable to emissions of that air the discretion to consider different Act as requiring the agency to give pollutant. New motor vehicles and standards for appropriate groupings of subordinate status to factors such as cost, engines emit CO2, CH4, N2O, and HFC. vehicles (‘‘class or classes of new motor safety, and noise in determining what EPA has established standards and other vehicles’’), or a single standard for a standards are reasonable for aircraft engines. provisions that control emissions of Rather, EPA has greater flexibility under larger grouping of motor vehicles CO2, HFCs, N2O, and CH4. EPA has not (NRDC, 655 F. 2d at 338). Finally, with section 231 in determining what standard is set any standards for PFCs or SF as most reasonable for aircraft engines, and is 6 respect to regulation of vehicular not required to achieve a ‘‘technology they are not emitted by motor vehicles. greenhouse gas emissions, EPA is not 473 forcing’’ result. 2. EPA’s Tentative Conclusion That the ‘‘required to treat NHTSA’s . . . Proposed CO Standards Are regulations as establishing the baseline This interpretation was upheld as 2 Appropriate and Reasonable for the [section 202(a) standards].’’ reasonable in NACAA v. EPA (489 F.3d Coalition for Responsible Regulation, 1221, 1230 (D.C. Cir. 2007)). CAA In this section, EPA discusses the 684 F.3d at 127 (noting further that ‘‘the section 202(a) does not specify the factors, data and analysis the [section 202 (a)standards] provid[e] degree of weight to apply to each factor, Administrator has considered in the benefits above and beyond those and EPA accordingly has discretion in selection of the EPA’s proposed revised resulting from NHTSA’s fuel-economy choosing an appropriate balance among GHG emission standards for MYs 2021 standards’’). factors. See Sierra Club v. EPA, 325 F.3d and later. EPA requests comment on all Although standards under CAA 374, 378 (D.C. Cir. 2003) (even where a aspects of the proposed revised section 202(a)(1) are technology-based, provision is technology-forcing, the standards, including all Alternatives they are not based exclusively on provision ‘‘does not resolve how the discussed in this section and section IV technological capability. EPA has the Administrator should weigh all [the of this preamble. discretion to consider and weigh statutory] factors in the process of As discussed in Sections I and V.B of various factors along with technological finding the ‘greatest emission reduction this preamble, the primary purpose of feasibility, such as the cost of achievable’ ’’); see also Husqvarna AB v. Title II of the Clean Air Act is the compliance (see section 202(a)(2)), lead EPA, 254 F. 3d 195, 200 (D.C. Cir. 2001) protection of public health and welfare. time necessary for compliance (section (great discretion to balance statutory EPA’s light-duty vehicle GHG standards 202(a)(2)), safety (see NRDC, 655 F.2d at factors in considering level of serve this purpose, as the GHG 336 n. 31) and other impacts on technology-based standard, and emissions from light-duty vehicles have consumers,472 and energy impacts statutory requirement ‘‘[to give] been found by EPA to endanger public associated with use of the technology appropriate consideration to the cost of health and welfare (see EPA’s 2009 (see George E. Warren Corp. v. EPA, 159 applying . . . technology’’ does not Endangerment Finding for on-highway F.3d 616, 623–624 (D.C. Cir. 1998) mandate a specific method of cost motor vehicles), and the goal of these (ordinarily permissible for EPA to analysis); Hercules Inc. v. EPA, 598 F. standards is to reduce these emissions consider factors not specifically 2d 91, 106–07 (D.C. Cir. 1978) (‘‘In that contribute to climate change. enumerated in the Act)). reviewing a numerical standard, we CAA section 202(a)(2) states when In addition, EPA has clear authority to must ask whether the agency’s numbers setting emission standards for new set standards under CAA section 202(a) are within a ‘zone of reasonableness,’ motor vehicles, the standards ‘‘shall that are technology forcing when EPA not whether its numbers are precisely take effect after such period as the considers that to be appropriate but is right’’); Permian Basin Area Rate Cases, Administrator finds necessary to permit not required to do so (as compared to 390 U.S. 747, 797 (1968) (same); Federal the development and application of the standards set under provisions such as Power Commission v. Conway Corp., requisite technology, giving appropriate section 202(a)(3) and section 213(a)(3)). 426 U.S. 271, 278 (1976) (same); Exxon consideration to the cost of compliance EPA has interpreted a similar statutory Mobil Gas Marketing Co. v. FERC, 297 within such period.’’ 42 U.S.C. provision, CAA section 231, as follows: F. 3d 1071, 1084 (D.C. Cir. 2002) (same). 7521(a)(2). That is, when establishing As noted above, EPA has found that emissions standards, the Administrator While the statutory language of section 231 the elevated concentrations of must consider both the lead time is not identical to other provisions in title II of the CAA that direct EPA to establish greenhouse gases in the atmosphere may necessary for the development of technology-based standards for various types reasonably be anticipated to endanger technology which can be used to of engines, EPA interprets its authority under public health and welfare.474 EPA achieve the emissions standards and the section 231 to be somewhat similar to those defined the ‘‘air pollution’’ referred to in resulting costs of compliance on those provisions that require us to identify a CAA section 202(a) to be the combined entities that are directly subject to the reasonable balance of specified emissions mix of six long-lived and directly standards. reduction, cost, safety, noise, and other emitted GHGs: Carbon dioxide (CO ), The Administrator is not limited to factors. See, e.g., Husqvarna AB v. EPA, 254 2 methane (CH4), nitrous oxide (N2O), consideration of the factors specified in hydrofluorocarbons (HFCs), CAA section 202(a)(2) when 472 Since its earliest Title II regulations, EPA has considered the safety of pollution control perfluorocarbons (PFCs), and sulfur establishing standards for light-duty technologies. See 45 FR 14496, 14503 (March 5, hexafluoride (SF6). The EPA further vehicles. In addition to feasibility and 1980). (‘‘EPA would not require a particulate found under CAA section 202(a) that cost of compliance, the Administrator control technology that was known to involve emissions of the single air pollutant may (and historically has) considered serious safety problems. If during the development of the trap-oxidizer safety problems are discovered, such factors as safety, energy use and EPA would reconsider the control requirements 473 70 FR 69664, 69676 (Nov. 17, 2005). security, degree of reduction of both implemented by this rulemaking.’’) 474 74 FR 66496 (Dec. 15, 2009). GHG and non-GHG pollutants,

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00244 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43229

technology cost-effectiveness, and costs vehicle fleet of a wide range of below, the Administrator weighs and other impacts on consumers. As technologies that both reduce CO2 and technology availability along with discussed in prior rulemakings setting improve fuel economy (see PRIA several other factors, including costs, GHG standards,475 EPA may establish Chapter 6). The majority of these emissions impacts, safety, and technology-forcing standards under technologies have already been consumer impacts in determining the section 202(a), but when it does so it developed, have been commercialized, appropriate standards under the Clean must provide sufficient basis for its and are in-use on vehicles today. These Air Act. belief that the industry can develop the technologies include, but are not limited (b) Consideration of the Cost of needed technology in the available time. to, engine and transmission Compliance However, EPA is not required to set technologies, vehicle mass reduction technology-forcing standards under technologies, technologies to reduce the EPA is required to consider costs in section 202(a). Rather, because section vehicles’ aerodynamic drag, and a range compliance before setting standards 202(a), unlike the text of section of electrification technologies. The under section 202(a). Compared to the 202(a)(3) and section 213(a)(3),476 does electrification technologies include 12- proposed standards, the EPA MY 2020– not specify that standards shall obtain Volt stop-start systems, 48-Volt mild 2025 standards announced in 2012 ‘‘the greatest degree of emission hybrids, strong hybrid systems, plug-in would cost the automotive industry an reduction achievable,’’ EPA retains hybrid electric vehicles, and dedicated estimated total of $260 billion for the considerable discretion under section electric vehicles. vehicles produced from MY 2016 through MY 2029, as shown in Table 202(a) in deciding how to weigh the If the Administrator’s consideration of various factors, consistent with the VIII–9. The additional per-vehicle the appropriateness of the standards technology costs for these previously- language and purpose of the Clean Air were based solely on an assessment of Act, to determine what standards are issued standards would be an estimated technology availability and $2,260 in MY 2030, relative to the appropriate. development, the Administrator might The analysis of alternatives supports proposed standards, as shown in Table consider a wide range of standards to be the Administrator’s consideration of a VIII–31 and Table VIII–32. Especially appropriate. As shown in Sections range of alternative standards, from the considering the change in reference VII.B.2 and VIII.B.1.b), and in PRIA existing standards to several alternatives point, these costs are considerably larger Chapter 6.3.2, the projected penetration that are less stringent. Specifically, the than EPA projected in 2012. Less of technologies varies across the analysis supports the consideration of stringent standards would be less this range of alternative standards due Alternatives presented in today’s burdensome. For example, compared to to factors relevant under the EPA’s proposal. In general, the existing EPA the proposed standards, Alternative 8 is authority pursuant to section 202(a), standards are projected to result in the projected to increase the per-vehicle such as GHG emissions reductions, the highest penetration of advanced cost by $1,510 (also in MY 2030), necessary technology and associated technologies, in particular mild hybrid Alternative 6 increases the per-vehicle lead-time, the costs of compliance on and strong hybrid technologies. Lower costs by $1,120, and Alternative 4 automakers, the impact on consumers stringency Alternatives in general are increases the per-vehicle costs by $490. projected to result in lower penetration with respect to cost and vehicle choice, (c) Consideration of Costs to Consumers and effects on safety. These factors, and of technologies, in particular for the the Administrator’s proposed mild hybrid and strong hybrid In addition to the costs to the conclusion, after consideration of these technologies, with the Preferred automotive industry described above, factors, indicate that Alternative 1 Alternative projected to result in the which could be passed on to consumers, represents the most appropriate lowest level of electrification technology the analysis estimates increased costs standards for model years 2021 and penetration. For example, the existing for the consumer for changes in beyond are discussed further below. CO2 standards are projected to require a maintenance, financing, insurance, combined passenger car and truck fleet taxes, and other fees, as shown in Table (a) Consideration of the Development penetration of mild hybrids plus strong VIII–31 and Table VIII–32. Considering and Application of Technology To hybrids of 58% of new vehicle sales in these additional costs, EPA’s Reduce CO2 Emissions MY 2030, while Alternative 8 projects a previously-issued standards for MYs When EPA establishes emissions 34% penetration, Alternative 6 projects 2020–2025 would increase the projected standards under section 202, it a 22% penetration, Alternative 4 per-vehicle costs in MY 2030 to an considers both what technologies are projects an 8% penetration, and the estimated $2,810 relative to the currently available and what Proposed Alternative (Alternative 1) proposed standards, at a seven percent technologies under development may projects a 4% penetration. These discount rate. The lower the increased become available. For today’s proposal, technologies are available and in stringency of the Alternative, the lower EPA takes note of the analysis of the production today, and MY 2020 through the total per-vehicle costs increase for potential penetration into the future MY 2025 standards are still a number of the consumer. For example, Alternative years away. In light of the wide range 8 increases the total costs for the 475 See, e.g., 77 FR 62624, 62673 (Oct. 15, 2012). of existing technologies that have consumer on a per-vehicle basis by 476 Section 202(a)(3) provides that regulations already been developed, have been $2,270 (in MY 2030 compared to the applicable to emissions of certain specified pollutants from heavy-duty vehicles or engines commercialized, and are in-use on costs of the proposed standards), ‘‘shall contain standards which reflect the greatest vehicles today, including those Alternative 6 increases the costs to the degree of emission reduction achievable through developed since the 2012 rule, consumer by $1,400 per-vehicle, and the application of technology which the technology availability, development Alternative 4 increases the costs by $610 Administrator determines will be available . . . giving appropriate consideration to cost, energy, and application, if it were considered in per-vehicle, all at a seven percent and safety factors associated with the application of isolation, is not necessarily a limiting discount rate. such technology.’’ 42 U.S.C. 7521(a)(3). Section factor in the Administrator’s selection of The analysis also projects the fuel 213(a)(3) contains a similar provision for new which standards are appropriate within savings for the vehicle owner over the nonroad engines and new nonroad vehicles (other than locomotives or engines used in locomotives). the range of the Alternatives presented life of the vehicles that come with lower 42 U.S.C. 7547(a)(3). in this proposal. However, as described levels of CO2 emissions. For example, as

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00245 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43230 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

477 shown in Table VIII–32 (at a seven CO2. As noted above, the purpose of emissions averaging, banking, and percent discount rate), for the Title II emissions standards is to protect trading program; air-conditioning previously-announced EPA standards the public health and welfare, and in program credits; flexibility in how to for MYs 2021–2025 (in MY 2030 establishing emissions standards the comply with the N2O and methane compared to the costs of the proposed Administrator is cognizant of the standard; off-cycle credit program, etc.) standards), the analysis projects a per- importance of this goal. At the same was to maintain consumer choice. The vehicle life-time fuel savings, including time, as discussed above, unlike other EPA standards are designed to require retail taxes, of $1,510 per vehicle, as provisions in Title II, Section 202(a) reductions of CO2 emissions over time well as an additional savings to the does not require the Administrator to set from the vehicle fleet as a whole but consumer from rebound driving and standards which result in the greatest also to provide sufficient flexibility to time saved refueling the vehicle of $610 degree of emissions control achievable, the automotive manufacturers so that per vehicle, for a total savings of $2,120. though the Administrator has the firms can produce vehicles which serve However, these savings to the consumer discretion to do so. Thus, in setting the needs of their customers. EPA are not enough to offset the these standards, the Administrator takes believes the past several model years in accompanying projected $2,810 increase into consideration other factors the market place show the benefits of in consumer costs. Compared to the discussed above and below, including this approach. Automotive companies proposed standards, the previously- not only technological feasibility, lead- have been able to reduce their fleet-wide issued EPA standards for MYs 2021– time, and the cost of compliance but CO2 emissions while continuing to 2025 would increase net costs to also potential impacts of vehicle produce and sell the many diverse consumers by $690 over the lifetime of emission standards on safety and other products that serve the needs of the MY 2030 vehicles. This imbalance impacts on consumers. Notwithstanding consumers in the market, e.g., full-size between costs and fuel savings contrasts the fact that GHG emissions reductions pick-up trucks with high towing sharply with what EPA projected in would be lower under today’s proposal capabilities, minivans, cross-over 2012 when setting those standards then, than for the existing EPA standards, in vehicles, SUVs, and passenger cars; and the fuel savings is considerably light of the new assessment indicating vehicles with off-road capabilities; smaller (this is due in large part to lower higher vehicle costs and associated luxury/premium vehicles, supercars, current and projected fuel prices). Also, impacts on consumers, and safety performance vehicles, entry level relative to the proposed standards, and impacts, the Administrator believes vehicles, etc. over the lifetime of MY 2030 vehicles, from a cost/benefit perspective that the At the same, the Administrator the projected net cost increase to foregone GHG emission reduction recognizes that automotive customers consumers from adopting Alternative 8 benefits from the proposed standards are a diverse group, that automotive is $300, Alternative 6 projects a net cost are warranted. companies do not all compete for the increase to consumers of $100, same segments of the market, and that (e) Consideration of Consumer Choice Alternative 4 projects a net savings to increasing stringency in the standards can be expected to have different effects consumers of $60, and Alternative 2 As discussed previously, the EPA CO2 projects a net savings to consumers of standards are based on vehicle footprint, not just on certain vehicle segments but $10. and in general smaller footprint vehicles on certain manufacturers who have developed market strategies around have individual CO2 targets that are (d) Consideration of GHG Emissions lower (more stringent) than larger those vehicle segments. The Administrator further recognizes that As discussed above, the purpose of footprint vehicles. The passenger car the diversity of the automotive customer CO standards established under CAA fleet has footprint curves that are 2 base, combined with the analysis, raises Section 202 is to reduce GHG emissions, distinct from the light-truck fleet. One of concerns that the existing standards, if which contribute to climate change. As the goals EPA had in designing the they are not adjusted, may not continue shown in Table VIII–34, the analysis program with footprint-based standards, to fulfill the agency’s goal of providing projects that, compared to the baseline in considering the shape, slope, and sufficient manufacturer flexibility to standards, the proposed CO standards stringency of the footprint standard 2 meet consumer needs and consumer for MYs 2021–2026 would increase curves, and in adopting many compliance flexibilities (e.g., the choice preferences. The analysis vehicle CO2 emissions by 713 million projects that high penetrations of metric tons (MMT) over the lifetime of 477 This preamble and the PRIA document hybridized vehicles would be required the vehicles produced from MY 1979 to achieve the previously-issued EPA through MY 2029, with an additional estimates annual GHG emissions from light-duty vehicles under the baseline CO2 standards, the MYs 2021–2025 standards, specifically 159 MMT in CO2 reduction from proposed standards, and the standards defined by 37% mild hybrid penetration and 21% upstream sources for a total increase of each of the other regulatory alternatives under consideration. For the final rule issued in 2012, strong hybrids for the new vehicle fleet 872 MMT. The modeling of proposed in MY 2030 (See Table VIII–24). For the revised and alternative standards EPA estimated changes in atmospheric CO2, global temperature, and sea level rise using GCAM and passenger car fleet, the projection is projects that more stringent standards MAGICC with outputs from its OMEGA model. 20% mild hybrid and 24% strong will result in smaller increases in GHG Because the agencies are now using the same model hybrid, and for the light-truck fleet 56% emissions (also compared to the and inputs, outputs from NHTSA’s DEIS (that also used GCAM and MAGICC) were analyzed. Today’s mild hybrid and 17% strong hybrid (See baseline standards. Compared to the analysis estimates that annual GHG emissions from Table VIII–26 and Table VIII–28). baseline standards, Alternative 8 is light-duty vehicles under the CO2 standards defined The Administrator is concerned that projected to increase CO2 emissions by by each regulatory alternative would be within this projected level of hybridization, 264 MMT from combined vehicle about one percent of emissions under the corresponding CAFE standards. Especially and the associated vehicle costs, arising tailpipe and upstream reductions over considering the uncertainties involved in estimating from the existing standards may be too the lifetime of the vehicles produced future climate impacts, the very similar estimates of high from a consumer-choice through MY 2029. Alternative 6 is future GHG emissions under CO2 standards and perspective. While consumers have projected to increase CO2 emissions by corresponding CAFE standards means that climate impacts presented in NHTSA’s draft EIS represent benefited from improvements over 422 MMT, Alternative 4 by 649 MMT of well the potential climate impacts of the proposed several decades in traditional vehicle CO2, and Alternative 2 by 825 MMT of and alternative CO2 standards. technologies, such as advancements in

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00246 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43231

transmissions and internal combustion The automotive industry is highly proposed standards, the previously- engines, advanced electrification competitive, and firms may be reluctant issued EPA standards would increase technologies are a departure from what to base their future product strategy on highway fatalities by 15,680 over the consumers have traditionally an uncertain future credit availability. lifetime of vehicles produced through purchased. Strong hybrid and other As can be seen in Table VIII–24, the MY 2029 (See Table VII–89). advanced electrification technologies analysis projects that lower levels of EPA views the potential impacts of have been available for many years (20 stringency (Alternatives 1–8) will emission standards on safety as an years for strong hybrids and eight years require lower penetrations of mild important consideration in determining for plug-in and all electric vehicles), and hybrids and strong hybrids as compared the appropriate standards under section sales levels have been relatively low, on to the 2012 EPA standards. For example, 202. The analysis projects adverse the order of two to three precent per Alternative 8 projects a 34% penetration impacts on safety that are significantly year for strong hybrids.478 As discussed of mild and strong hybrid new vehicle different from the analysis included and above, the analysis projects that the sales in MY 2030, Alternative 6 projects considered in the 2012 rule which 2012 EPA standards are projected to a 22% penetration of these technologies, established the MY 2021–25 GHG require a significant increase in Alternative 4 projects an eight percent standards and the 2016 Draft Technical hybridization over the next 7 to 12 penetration, and Alternative 2 projects a Assessment Report. As discussed model years. This large increase may four percent penetration of mild and previously in this document, previous require automotive companies to change strong hybrids in MY 2030. The EPA analyses limited the amount of mass the choice of vehicle types and the proposal, Alternative 1, projects a two reduction assumed for certain vehicles, utility of the vehicles available to percent penetration of mild hybrids and while acknowledging that customers from what the companies a two percent penetration of strong manufacturers would not necessarily would otherwise offer in the absence of hybrids. These are levels similar to what choose to avoid mass reductions in the the existing standards. auto manufacturers are selling today, ways that the agencies assumed. The EPA notes that in the EPA’s annual suggesting that auto companies will be current analysis eliminates this Manufacturer Performance Report on able to produce vehicles in the future constraint. The Administrator considers the compliance status of the automotive that meet the full range of needs from this difference to be a significant factor companies for the EPA GHG standards, consumers, thus preserving consumer indicating that it is appropriate to EPA has reported that emissions trading choice. consider a range of alternative revised standards, including Alternative 1, the has occurred a number of times in the (f) Consideration of Safety past several years.479 Through MY 2016, preferred alternative. EPA has long considered the effects these trades have included 12 firms, on safety of its emission standards. See (g) Balancing of Factors and EPA’s with five firms trading CO credits to 2 45 FR 14496, 14503 (1980) (‘‘EPA would Proposed Revised Standards for MY seven firms, and thus far in the EPA not require a particulate control 2021 and Later GHG program credits generated in MY technology that was known to involve As discussed in this section, the 2010 through MY 2016 have been serious safety problems.’’). More Administrator is required to consider a traded. This represents about one-half of recently, EPA has considered the number of factors when establishing the automotive companies selling potential impacts of emissions emission standards under Section vehicles in the U.S. market, but since standards on safety in past rulemakings 202(a)(2) of the Clean Air Act: The several of these firms are small players, on GHG standards, including the 2010 standards ‘‘shall take effect after such it is less than half of the volume. In rule which established the 2012–2016 period as the Administrator finds total, approximately 30 million light-duty vehicle GHG standards, and necessary to permit the development Megagrams of CO2 have been traded the 2012 rule which previously and application of the requisite between firms, which is approximately established the 2017–2025 light-duty technology, giving appropriate 10% of the MY 2016 industry-wide vehicle GHG standards. Indeed, section consideration to the cost of compliance bank of credits. Credit trading between 202(a)(4)(A) specifically prohibits the within such period.’’ 42 U.S.C. firms can lower the costs of compliance use of an emission control device, 7521(a)(2). For this proposal, the for firms, both for those selling and system or element of design that will Administrator has considered a wide those purchasing credits, and this cause or contribute to an unreasonable range of potential emission standards program compliance flexibility is risk to public health, welfare, or safety. (Alternatives 1 through 9), ranging from another tool by which auto firms can 42 U.S.C. 7521(a)(4)(A). the existing EPA MY 2021 to MY 2025 provide the types of vehicle offerings The proposal’s safety analysis projects standards, through a number of less that customers want. However, long- that the 2012 EPA GHG standards for stringent alternatives, including term planning is an important MYs 2021 and later would increase Alternative 1, the preferred Alternative. consideration for automakers, and an vehicle fatalities due to several reasons, In addition to technological feasibility, OEM who may want to purchase credits namely increased vehicle prices lead-time, and the costs of compliance, as part of a future compliance strategy resulting in delayed turnover of the the Administrator has also considered cannot be guaranteed they will be able vehicle fleet to newer, safer vehicles, the impact of various standards on to find credits. increased fatalities and accidents due projected emissions reductions, the rebound effect, and passenger car consumer choice, and vehicle safety. 478 Light-Duty Automotive Technology, Carbon mass reduction. The assessment is The Administrator believes the existing Dioxide Emissions, and Fuel Economy Trends: 1975 discussed in Section 0 of this preamble EPA standards for MY 2021 and later, Through 2017, U.S. EPA Table 5.1 (Jan. 2018), available at https://nepis.epa.gov/Exe/ZyPDF.cgi? and is detailed in Chapter 11 of the considered as a whole, are too stringent. Dockey=P100TGDW.pdf. PRIA. The assessment projects that The Administrator gives particular 479 See Greenhouse Gas Emission Standards for Alternative 1, which includes no change consideration to the high projected costs Light-Duty Vehicles: Manufacturer Performance in the GHG emissions standards for MY of the standards and the impact of the Report for the 2016 Model Year (EPA Report 420– R18–002), U.S. EPA (Jan. 2018), available at https:// 2021 and later, would yield the lowest standards on vehicle safety. The nepis.epa.gov/Exe/ZyPDF.cgi?Dockey= number of vehicle fatalities. The analysis projects that, compared to the P100TGIA.pdf. analysis projects that, compared to the proposed standards, the previously-

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00247 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43232 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

issued EPA standards for MYs 2021– VI. Preemption of State and Local Laws dioxide is an ‘‘air pollutant’’ within the 2025 would increase MY 2030 Accomplishing the goals of EPCA meaning of the CAA and thus compliance costs by nearly $1,900 per requires a set of uniform national fuel potentially subject to regulation under vehicle. Although EPA projected a economy standards. Achieving this that statute. The Supreme Court did not 480 similar cost increase in the 2012 rule national standard requires the agencies consider the issue of preemption under announcing standards through 2025, to clearly discuss the extent to which EPCA of state laws or regulations this prior estimate was relative to an state and local standards are expressly regulating CO2 tailpipe emissions from indefinite continuation of standards for or impliedly preempted. As described automobiles, but it did address the MY 2016, and assuming that absent herein, doing so is fundamental to the relationship between EPA and NHTSA 483 regulation, manufacturers would not effectiveness of the new proposed set of rulemaking obligations. Later that increase fuel economy at all. In fuel economy standards and to the year, two Federal district courts in addition, as mentioned above, the critical importance of ensuring that the Vermont and California ruled that the analysis projects that, compared to the proposed Federal standards will GHG motor vehicle emission standards proposed standards, the previously- constitute uniform national adopted by those states were not 484 issued EPA standards would increase requirements, as Congress intended. preempted under EPCA. Still later highway fatalities by 12,903 over the This is also a fundamental reason that that year, Congress enacted EISA, lifetime of vehicles produced through EPA is proposing the withdrawal of amending EPCA by mandating annual MY 2029. In evaluating the other CAA preemption waivers granted to increases in passenger car and light Alternatives under consideration, the California relating to its GHG standards truck CAFE standards through MY 2020 Administrator notes that Alternative 1 and Zero Emissions Vehicle (ZEV) and maximum feasible fuel economy 485 has the lowest cost of compliance and mandate. standards subsequently. the lowest number of fatalities. He also In March 2008, EPA denied notes that Alternative 1 will preserve A. Preemption Under the Energy Policy California’s request for a waiver of CAA consumer choice in the vehicle market and Conservation Act preemption.486 In May 2008, NHTSA and will provide a relatively high net 1. History of EPCA Preemption issued a proposal for MYs 2011–2015 savings to consumers, when assessing Discussions in Rulemakings standards, which included a significant the increased costs of vehicles against discussion of EPCA preemption and a fuel savings over the lifetime of the NHTSA has asserted the preemption proposed regulatory statement to of certain State emissions standards vehicle. provide that state vehicle tailpipe CO2 The Administrator recognizes that under EPCA a number of times in CAFE standards are related to fuel economy Alternative 1 is projected to result in rulemakings dating back to 2002.481 The and therefore expressly preempted less CO2 reductions compared to the initial rulemaking discussion was under EPCA, and that they conflict with existing EPA standards and is not prompted by a court filing by the State the goals and objectives of EPCA and projected to achieve additional GHG of California claiming that NHTSA did therefore also impliedly preempted.487 reductions beyond the MY 2020 not treat California’s Greenhouse Gas The Bush Administration did not issue standards. However, the Administrator Emissions regulation as preempted.482 a final rule for MYs 2011–2015. notes that, unlike other provisions in This continuous dialogue involves a A number of significant actions Title II referenced above, section 202(a) variety of parties (i.e., the states, the happened in quick succession at the does not require the Administrator to set Federal government—especially EPA— beginning of the prior Administration. standards which result in the ‘‘greatest and the general public) and occurs The first day post-inauguration, CARB degree of emissions control achievable.’’ through a variety of means, including petitioned for reconsideration of EPA’s In light of this statutory discretion and several rulemaking proceedings. After denial of a waiver of CAA preemption the range of factors that the statute NHTSA first raised the issue of for California’s GHG emissions authorizes and permits the preemption in 2002 when proposing standards for 2009 and later model year Administrator to consider, and his standards for MYs 2005–2007 light vehicles.488 Several days later, on consideration of the factors discussed trucks, the agency explored preemption January 26, 2009, President Obama above, the EPA proposes to conclude at great length in response to extensive issued a memorandum requesting, that maintaining the MY 2020 standards public comment in its August 2005 among other things (including going forward is an appropriate NPRM and its April 2006 final rule for approach under section 202(a). MYs 2008–2011 light trucks. 483 The Court reasoned that the fact that NHTSA Therefore, based on the data and During the period between the NPRM ‘‘sets mileage standards in no way licenses EPA to shirk its environmental responsibilities. EPA has analysis detailed in this proposal, the and the final rule for MYs 2008–2011 light trucks, California separately been charged with protecting the public’s ‘health’ Administrator is proposing that the and ‘welfare,’ . . . a statutory obligation wholly existing MY 2021 and later GHG requested that the EPA grant a waiver of independent of DOT’s mandate to promote energy standards are too stringent and is CAA preemption, pursuant to Section efficiency.... The two obligations may overlap, proposing to revise the MY 2021 and 209 of that act, for its Greenhouse Gas but there is no reason to think the two agencies Emissions regulation. If EPA granted the cannot both administer their obligations and yet later standards to maintain the MY 2020 avoid inconsistency.’’ Massachusetts v. EPA, 549 levels in subsequent model years. EPA waiver, the CAA would under certain U.S. 497, 532 (2007). requests comment on all aspects of this circumstances allow other states to 484 Green Mountain Chrysler v. Crombie, 508 proposal and supporting assessments, adopt the same regulation pursuant to F.Supp.2d 295 (D. Vt. 2007); Central Valley Chrysler-Jeep, Inc. v. Goldstene, 529 F.Supp.2d including the Administrator’s CAA Section 177, without being preempted by the CAA. 1151 (E.D. Cal. 2007), as corrected (Mar. 26, 2008). consideration of the relevant factors 485 In 2007, the Supreme Court ruled in Public Law 110–140 (2007). under section 202(a) of the Clean Air 486 73 FR 12156 (Mar. 6, 2008). Act, the proposed Alternative 1, the Massachusetts v. EPA that carbon 487 73 FR 24352 (May 2, 2008). previously-established EPA GHG 488 For background on CARB’s petition, see EPA’s 481 standards, and all of the Alternatives 67 FR 77025 (December 16, 2002). Notice of Decision Granting a Waiver of Clean Air 482 See Appellants Opening Brief filed on behalf Act Preemption for California’s 2009 and discussed in section IV of this preamble. Michael P. Kenny in Central Valley Chrysler- Subsequent Model Year Greenhouse Gas Emission Plymouth, Inc. et al. v. Michael P. Kenny, No. 02– Standards for New Motor Vehicles, 74 FR 32744 480 77 FR 62624, 62665 (Oct. 15, 2012). 16395, at p. 33 (9th Cir. 2002). (Jul. 8, 2009).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00248 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43233

consideration of EPCA preemption in requirements and be deemed to comply regulation. In a further indication of light of Massachusetts v. EPA and other with California’s standards, and that Congress’ intent to ensure that state laws), that NHTSA’s rulemaking be this amendment facilitated the National regulatory schemes do not impinge divided into two parts—one regulation Program by allowing a manufacturer to upon EPCA’s goals, the statute preempts establishing standards for model year ‘‘meet all standards with a single state laws merely related to fuel 2011 only, and another for subsequent national fleet.’’ 494 NHTSA, at the time, economy standards or average fuel years. Less than two months after that erroneously saw this as obviating economy standards. Here, NHTSA memorandum, on March 6, 2009, consideration of EPCA preemption. At intends to assert preemption only over NHTSA issued its final rule for MY the same time, the agency did not state requirements that directly affect 2011 vehicles and announced that it address whether California’s ZEV corporate average fuel economy. program would be preempted since it would consider EPCA preemption in The Supreme Court has interpreted 489 has never been part of the National subsequent rulemakings. Then, on similar statutory preemption language Program. May 19, 2009, the White House on several occasions, concluding that a announced a coordinated program 2. Preemption Analysis state law ‘‘relates to’’ a Federal law if it addressing motor vehicle fuel economy ‘‘has a connection with or refers to’’ the and greenhouse gas emissions, to be Present circumstances require NHTSA subject of the Federal law.498 The Court, known as the ‘‘National Program,’’ to address the issue of preemption. citing similar Federal statutory whereby NHTSA and EPA would jointly Despite past attempts by NHTSA and language, extended the application of establish rules to harmonize compliance EPA to harmonize their respective and the ‘‘related to’’ standard to the Airline requirements for manufacturers. As part related regulations, the automotive Deregulation Act in Morales v. Trans of the National Program, several industry and U.S. consumers now face World Airlines, Inc.,499 concluding manufacturers and their trade regulatory uncertainty and increased that,’’ [f]or purposes of the present case, associations announced their costs, in no small part as a result of the key phrase, obviously, is ‘relating commitment to take several actions, California’s separate GHG emissions and to.’ The ordinary meaning of these including agreeing not to contest ZEV program. NHTSA and EPA now words is a broad one—‘to stand in some forthcoming CAFE and GHG standards seek to address these concerns with this relation; to have bearing or concern; to for MYs 2012–2016; not to challenge rulemaking proposal, in the interest of pertain; refer; to bring into association any grant of a CAA preemption waiver regulatory certainty and the clear with or connection with,’ . . .—and the for California’s GHG standards for prospect for disharmony with conflicting state requirements.495 words thus express a broad pre-emptive certain model years; and to stay and NHTSA is also guided by a desire to purpose.’’ 500 Courts look ‘‘both to the then dismiss all pending litigation obtain comments from state and local objectives of the . . . statute as a guide challenging California’s regulation of officials and other members of the to the scope of the state law that GHG emissions, including litigation public to inform fully the agency’s Congress understood would survive, concerning EPCA preemption of state position on this important issue.496 [and] to the nature of the effect of the GHG standards.490 state law on [the Federal standards].’’ 501 Less than two months later, in July (a) EPCA Preemption One of Congress’ objectives in EPCA 2009, EPA granted California’s January EPCA’s express preemption language was to create a national fuel economy 2009 request for reconsideration of the is broad and clear: CAA preemption waiver denial, standard, as clearly expressed in 49 allowing California to establish its own When an average fuel economy standard U.S.C. 32919(a). In addition to the 491 prescribed under this chapter is in effect, a statute’s plain language, which controls, GHG standards under the CAA. In State or a political subdivision of a State may granting the preemption waiver, EPA not adopt or enforce a law or regulation the legislative history of that provision acknowledged that its analysis was related to fuel economy standards or average further confirms that Congress intended based solely on CAA considerations and fuel economy standards for automobiles the provision to be broadly preemptive. did not ‘‘attempt to interpret or apply covered by an average fuel economy standard As Congress debated proposals that EPCA,’’ concluding that ‘‘EPA takes no under this chapter.497 would eventually become EPCA, the position regarding whether or not Unlike the CAA, EPCA does not allow Senate bill 502 sought to preempt State California’s GHG standards are for a waiver of preemption. Nor does laws only if they were ‘‘inconsistent’’ preempted under EPCA.’’ 492 EPCA allow for states to establish or with Federal fuel economy standards, In the subsequent MYs 2012–2016 enforce an identical or equivalent labeling, or advertising, while the House CAFE rulemaking, NHTSA elected to bill 503 sought to preempt State laws defer consideration of EPCA preemption 494 76 FR 74854, 74863 (Dec. 1, 2011). only if they were not ‘‘identical to’’ a concerns because of the ‘‘consistent and 495 While California’s ‘‘deem to comply’’ Federal requirement. The express coordinated Federal standards that provision provided some temporary relief from preemption provision, as enacted, three different sets of standards, its regulations still preempts all State laws that relate to apply nationally under the National mandate that some manufacturers comply with 493 Program.’’ Later, in establishing MYs burdensome filing requirements and California may fuel economy standards. No exception is 2017–2021 CAFE standards, NHTSA act to revoke the provision. In fact, California is made for State laws on the ground that pointed out that after finalization of the already seeking comment on potentially changing the regulation to provide that manufacturers would 498 MYs 2012–2016 CAFE standards, only be deemed to comply with CARB requirements Shaw v. Delta Airlines, Inc., 463 U.S. 85, 97 California amended its GHG regulations if meeting the currently-final EPA standards. See (1983) (ERISA case). to provide that manufacturers could https://www.arb.ca.gov/msprog/levprog/leviii/ 499 504 U.S. 374, 383–84 (1992). elect to comply with the EPA GHG leviii_dtc_notice05072018.pdf (last accessed May 500 Id. at 383. 17, 2018). Moreover, the ‘‘deem to comply’’ 501 California Div. of Labor Standards provision applies only to tailpipe CO2 emissions Enforcement v. Dillingham Constr., N.A., Inc., 519 489 74 FR 14196 (Mar. 6, 2009). requirements—not to the ZEV program. U.S. 316, 325 (1997), (quoting N.Y Conference of 490 75 FR 25324, 25328 (May 7, 2010). 496 See also E.O. 13132 (Federalism); E.O. 12988 Blue Cross & Blue Shield Plans v. Travelers Ins. Co., 491 74 FR 32744 (Jul. 8, 2009). sec. 3(b)(1)(B) (Civil Justice Reform); 54 FR 11765 514 U.S. 645, 656 (1995)). 492 74 FR at 32783 (Jul. 8, 2009). (Mar. 22, 1989); 58 FR 68274 (Dec. 23, 1993); and 502 S. 1883, 94th Cong., 1st Sess., Section 509. 493 75 FR 25324, 25546 (May 7, 2010); see also 74 70 FR 21844 (Apr. 27, 2005). 503 H.R. 7014, 94th Cong., 1st Sess., Section 507 FR 49454, 49635 (Sep. 28, 2009). 497 49 U.S.C. 32919. as introduced, Section 509 as reported.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00249 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43234 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

they are consistent with or identical to linked to fuel consumption because CO2 preempting state standards that are Federal requirements.504 is a necessary and inevitable byproduct related to fuel economy standards, when In enacting EISA, Congress did not of burning gasoline. The more fuel a Federal fuel economy standards are in repeal or amend EPCA’s express vehicle burns or consumes, the more place. preemption provision. Congress did, CO2 it emits. To the extent that light In mandating Federal fuel economy however, adopt a savings provision vehicles are not powered by internal standards under EPCA, Congress has regarding the effect of EISA, and the combustion engines, their use generally expressly preempted any state laws or amendments made by it: involves some release of CO2 or other regulations relating to fuel economy Nothing in this Act or an amendment made GHG emissions, even if indirectly, standards. A state requirement limiting by this Act supersedes, limits the authority associated with the vehicle performing tailpipe CO2 emissions is such a law or provided or responsibility conferred by, or its work of traveling down the road. regulation because it has the direct authorizes any violation of any provision of CNG and LPG vehicles release CO2 effect of regulating fuel consumption. law (including a regulation), including any during combustion. Even for battery- Given that substantially reducing CO energy or environmental law or regulation. 2 electric vehicles, fossil fuels are used in tailpipe emissions from automobiles is We understand this statutory language at least some part of production of unavoidably and overwhelmingly to prevent EISA from limiting pre- electricity in virtually all parts of the dependent upon substantially existing authority or responsibility country, and that electricity is used to increasing fuel economy through conferred by any law or from move the vehicles. And with hydrogen installation of engine technologies, authorizing violation of any law. By the vehicles, methane remains a major part transmission technologies, accessory same token, the savings provision does of the generation of hydrogen fuel, technologies, vehicle technologies, and not purport to expand pre-existing which is also used to move those hybrid technologies, increases in fuel authority or responsibility. Thus, to the vehicles. Carbon dioxide is thus a economy inevitably produce extent that EPCA’s express preemption byproduct of moving virtually if not commensurate reductions in CO2 provision limited State authority and literally all light-duty vehicles, and the tailpipe emissions. Since there is but responsibility prior to the enactment of amount of CO2 released directly one pool of technologies 507 for reducing EISA, it continues to limit such correlates to the amount of fossil fuels tailpipe CO2 emissions and increasing authority and responsibility to the same used to power the vehicle so it can fuel economy available now and for the extent after the enactment of EISA. We move. foreseeable future, regulation of CO2 recognize that the Congressional Record EPCA has specified since its inception emissions and fuel consumption are contains statements regarding the that compliance with CAFE standards is inextricably linked. Such state savings provision indicating that certain to be determined in accordance with regulations are therefore unquestionably members of Congress may have test and calculation procedures ‘‘related’’ and expressly preempted considered this language as allowing established by EPA.506 More under 49 U.S.C. 32919. California to set tailpipe GHG emissions specifically, the tests are to be Moreover, state standards that have standards in contravention of EPCA’s performed using ‘‘the same procedures the effect of regulating tailpipe CO2 express preemption provision. Note, for passenger automobiles the emissions or fuel economy are likewise however, that statements made on the Administrator used for model year 1975 related to fuel economy standards and floor of the Senate or House before the . . . procedures that give comparable likewise preempted. For instance, if a votes on EISA cannot expand the scope results.’’ Under these procedures, state were to regulate all tailpipe GHG of the savings provision or even be used compliance with the CAFE standards is emissions from a vehicle, and not just to ‘‘clarify’’ it, given the unambiguous and has always been based on the rates CO2, the state would nonetheless plain meaning of both the savings of emission of CO2, CO, and regulate tailpipe CO2 emissions, since provision and EPCA’s express hydrocarbons from covered vehicles, CO2 emissions comprise the preemption provision. If Congress had but primarily on the emission rates of overwhelming majority of tailpipe wanted to narrow the express CO2. In the measurement and carbon emissions. EPCA preempts such preemption provision, it could have calculation of a given vehicle model’s a standard. chosen to include such an amendment fuel economy for purposes of Likewise, a state law prohibiting all in EISA. It did not. determining a manufacturer’s tailpipe emissions, carbon or otherwise, compliance with Federal fuel economy (b) Tailpipe CO2 Emissions Regulations from some or all vehicles sold in the standards, the role of CO2 is state, would relate to fuel economy or Prohibitions are Related to Fuel approximately 100 times greater than Economy Standards standards and be preempted by EPCA, the combined role of the other two since the majority of tailpipe emissions This broad statutory preemption relevant carbon exhaust gases. Given consist of CO2. We recognize that this provision also necessarily governs state that the amount of CO2, CO, and preempts state programs, such as regulations over greenhouse gas hydrocarbons emitted from a vehicle’s California’s ZEV mandate, that establish emissions. GHG emissions, and tailpipe relates directly to the amount of requirements that a portion of a particularly CO2 emissions, are fuel it consumes, EPA can reliably and vehicle’s fleet sold or purchased consist mathematically linked to fuel economy; accurately convert the amount of those of vehicles that produce no tailpipe therefore, regulations limiting tailpipe gases emitted by that vehicle into the emissions. CO2 emissions are directly related to miles per gallon achieved by that fuel economy.505 To summarize, most vehicle. In recognizing that 1975 test (c) Other GHG Emissions Requirements light vehicles are powered by gasoline procedures were sufficient to measure May Not Be Preempted by EPCA internal combustion engines. The fuel economy performance, Congress While EPCA expressly preempts state combustion of gasoline produces CO2 in recognized the direct relationship tailpipe CO2 emission limits, some GHG amounts that can be readily calculated. between CO2 emissions and fuel emissions from vehicles have no CO2 emissions are always and directly economy standards, while in the same piece of legislation expressly 507 With the minor exception of regulating the 504 See 71 FR 17566, 17657 (April 6, 2006). carbon intensity of fuels—an activity not preempted 505 71 FR at 17659, et seq. 506 49 U.S.C. 32904(c). by EPCA.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00250 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43235

relation to fuel economy and are NHTSA invites comments on the preemption, makes clear that waiver of therefore outside the scope of EPCA extent to which a state standard can preemption under that statute operates preemption. For instance, vehicle air have some incidental impact on fuel only to relieve ‘‘application of this conditioning units can cause GHG economy or CO2 emissions without section’’—the preemption provision of emissions by leaking refrigerants when being ‘‘related to’’ fuel economy the CAA—and not application of other the system is recharged or when it is standards. statutes.516 EPA and NHTSA tentatively crushed at the end of the vehicle’s life. (d) A Waiver of CAA Preemption Does agree that a waiver under the CAA does Since such emissions have no bearing Not Affect, in Any Way, EPCA not also waive EPCA preemption. on a vehicle’s fuel economy Preemption The Vermont and California Federal performance or tailpipe CO2 emissions, states can pass laws specifically When a state establishes a standard district court decisions mentioned regulating or even prohibiting such related to fuel economy, it does so in above involved challenges to a vehicular refrigerant leakage without violation of EPCA’s preemption statute California Air Resources Board relating to fuel economy if doing so and the standard is therefore void ab regulation establishing vehicle tailpipe would be otherwise consistent with initio. GHG emission standards. The courts Federal law. Therefore, EPCA would not Federal preemption is rooted in the concluded that EPCA did not preempt preempt such laws, if narrowly drafted Supremacy Clause of the U.S. such standards. In both decisions, the Constitution.509 Courts have long so as not to include tailpipe CO2 courts placed much weight upon the emissions. If, however, a state law recognized that the Supremacy Clause fact that California had petitioned EPA sought to limit the combined GHG of the Constitution gives Congress the for a waiver of CAA preemption emissions from a motor vehicle, in a power to specifically preempt State pursuant to 42 U.S.C. 7543(b). law.510 Broadly speaking, the United manner that would include tailpipe CO2 NHTSA and EPA do not agree with States Supreme Court has long held that emissions, EPCA would preempt that the district courts’ express preemption ‘‘an act done in violation of a statutory portion of the law limiting tailpipe CO2 analyses. EPCA preempts state laws and prohibition is void,’’ 511 and has emissions. regulations ‘‘related to fuel economy specifically noted that such acts are not Similarly, state safety requirements standards or average fuel economy may have a merely incidental impact on merely ‘‘voidable at the instance of the government’’ but void from the standards for automobiles covered by an fuel economy and not relate to fuel 517 512 average fuel economy standard.’’ The economy. For instance, a state may outset. The Ninth Circuit stated it more plainly: ‘‘Under Federal law, an courts in Green Mountain Chrysler and mandate that children traveling in Central Valley Chrysler-Jeep recognized motor vehicles sit in child safety seats. act occurring in violation of a statutory 513 the relationship between CO2 emissions Child safety seats add weight, and mandate is void ab initio.’’ Discussing the Supremacy Clause, the and fuel economy. Nonetheless, they added weight has an impact on fuel erroneously concluded that the ‘‘related economy. This impact is merely Supreme Court explicitly explained that, ‘‘[i]t is basic to this constitutional to’’ language in EPCA’s preemption incidental, however, and does not clause should be construed ‘‘very directly relate to fuel economy command that all conflicting state provisions be without effect.’’ 514 And at narrowly’’ and adopted a novel standards. interpretation of ‘‘related to.’’ 518 The Likewise, EPA has recognized that least one Federal Court of Appeals explicitly stated that the Supremacy courts failed to recognize precedent California may apply for a waiver of providing broad effect to other CAA preemption for vehicle emissions, Clause means ‘‘state laws that ‘interfere preemption statutes using terms similar which must be granted in certain with, or are contrary to the laws of 515 to ‘‘related to,’’ as discussed above. circumstances. That said, EPCA does Congress’ are void ab initio.’’ While both the CAA and EPCA may preempt any regulation limiting or (e) A Clean Air Act Waiver Does Not preempt state laws limiting GHG prohibiting CO2 emissions or all tailpipe ‘‘Federalize’’ EPCA-Preempted State emissions from motor vehicles, avoiding emissions, as such regulations have the Standards preemption (by waiver or otherwise) effect of regulating CO emissions and 2 under one Federal law has no bearing relate to fuel economy standards.508 The district court in Green Mountain on the other Federal law’s preemptive Chrysler concluded that it could resolve 508 NHTSA notes that over the last decade CARB effect. Section 209 of the CAA, which the challenge to Vermont’s regulations has complicated its regulation of smog-forming provides for the possible waiver of CAA without directly considering the emissions (the original purpose of the Section 209 application of EPCA’s preemption CAA waiver) by combining it with regulation of and which contain dense populations of allergy GHG and, principally, CO2 emissions as well as the sufferers. provision. The court said that the ZEV mandate. Since EPCA prohibits state 509 U.S. Const. art VI, cl. 2. dispute did not concern preemption but regulation of CO emissions, a state program that 2 510 See Gibbons v. Ogden, 22 U.S. 1 (1824). concerned reconciling two different combines regulation of the two groups of pollutants 511 is preempted to the extent that the program relates Ewert v. Bluejacket, 259 U.S. 129, 138 (1922), Federal statutes (EPCA and the CAA). In to fuel economy. A regulatory regime in which quoting Waskey v. Hammer, 223 U.S. 85, 94 (1912). this regard, the district court stated that smog-forming pollutants are addressed without also 512 Waskey, 223 U.S. at 92. if EPA approved California’s waiver 513 directly or indirectly regulating fuel economy is not Cabazon Band of Mission Indians v. City of petition (which had not yet occurred), preempted under EPCA. Indio, Cal., 694 F.2d 634, 637 (9th Cir. 1982). Additionally, NHTSA notes that some suggest 514 Maryland v. Louisiana, 451 U.S. 725, 746 then Vermont’s GHG regulations that insofar as carbon dioxide emissions cause (1981) (citing McCulloch v. Maryland, 4 Wheat. 316, become ‘‘other motor vehicle standards’’ global climate change, they indirectly worsen air 427 (1819)). Other courts have used similar that NHTSA must consider in setting quality by (1) increasing formation of smog, because language to describe the impact of preemption. See, the chemical process that forms ground-level ozone e.g., Nathan Kimmel, Inc. v. DowElanco, 275 F.3d occurs faster at higher temperatures, and (2) 1199, 1203 (9th Cir. 2002) (explaining preempted 516 42 U.S.C. 7543(b)(1) (emphasis added); see increasing ragweed pollen, which can cause asthma state laws are ‘‘without effect’’); Sweat v. Hull, 200 also 42 U.S.C. 7543(b)(3) (‘‘compliance with such attacks in allergy sufferers. Comment is sought on F.Supp.2d 1162, 1172 (D. Ariz. 2001) (explaining State standards shall be treated as compliance with the extent to which the zero-tailpipe-emissions preempted state laws are ‘‘ineffective.’’). applicable Federal standards for purposes of this vehicles compelled to be sold by California’s ZEV 515 Antilles Cement Corp. v. Fortuno, 670 F.3d subchapter’’) (emphasis added). program reduce temperatures in the parts of 310, 323 (1st Cir. 2012) (quoting Gibbons v. Ogden, 517 49 U.S.C. 32919(a) (emphasis added). California which are in non-attainment for ozone 22 U.S. (9 Wheat.) 1 (1824)). 518 E.g., 529 F.Supp.2d at 1176.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00251 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43236 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

CAFE standards.519 In the court’s view, automotive industry’s withdrawal of what level of regulation is necessary to once EPA grants a waiver, compliance appeals. As explained above, the secure public health and welfare.’’ 523 with California’s standards is deemed to withdrawal of those appeals was a pre- In support of its position, the Central satisfy all Federal standards—not just condition to the 2010 issuance of the Valley Chrysler-Jeep found persuasive those of the CAA. In states that adopt final rule establishing the ‘‘National the Green Mountain Chrysler court’s California’s standards, compliance with Program’’ of fuel economy standards view that California emissions that standard would be deemed to and GHG emission standards for MYs regulations under CAA Section 209 satisfy all Federal standards as well. 2012–2016. have always been considered ‘‘other With this Federal accommodation of standards’’ on fuel economy. As state standards, the court concluded, In their appeals of the Green mentioned previously in the discussion Vermont’s regulations would stand. Mountain Chrysler decision, the vehicle of the ‘‘other standards’’ to be The court’s premise that preemption manufacturer associations argued that considered as factors in establishing provisions and principles do not apply the operation of EPCA’s express maximum feasible fuel economy is not based on precedent and is not preemption provision does not require standards, EPCA, as originally enacted, supported by applicable law. In fact, the that a conflict be shown between the contained a specific self-contained district court in Central Valley Chrysler- Federal and state standards, that the provision that provided that any Jeep recognized that ‘‘[t]he Green Federal and state standards be identical, manufacturer could apply to DOT for Mountain court never actually offers a or that the Federal and state standards modification of an average fuel economy legal foundation for the conclusion that serve the same purpose. We agree. The standard for model years 1978 through a state regulation granted waiver under conflict principles of implied 1980 if it could show the likely [CAA] section 209 [42 U.S.C. 7543] is preemption do not apply in fields where existence of a ‘‘Federal standards fuel essentially a federal regulation such that Congress has enacted an express economy reduction,’’ defined to include any conflict between the state regulation preemption provision prohibiting even EPA-approved California emissions and EPCA is a conflict between federal standards that reduce fuel economy. the existence of state standards. The regulations.’’ 520 NHTSA and EPA The court reasoned that ‘‘in 1975 when statutory test, whether the state disagree with the conclusion of these EPCA was passed, Congress decisions and reaffirm the longstanding standards are ‘‘related to’’ the Federal unequivocally stated that federal position that state standards regulating standards, is met by showing that the standards included EPA-approved tailpipe GHG emissions, such as the state GHG emission standards are not California emissions standards.’’ 524 standards challenged in the California simply related to, but actually the However, when EPCA was recodified in and Vermont district court cases, are functional equivalent of, the Federal 1994, ‘‘all reference to the modification preempted by EPCA because they fuel economy standards. The district process applicable for model years 1978 ‘‘relate to’’ fuel economy standards. We court itself recognized that ‘‘there is a through 1980, including the categories also note that those courts failed to near-perfect correlation between fuel of federal standards, was omitted as consider, much less give any weight to, consumed and carbon dioxide executed.’’ 525 The court noted that the NHTSA’s views of preemption, as the released.’’ Neither the inclusion in the legislative intent of the 1994 expert agency with authority over the state standard of emissions for which recodification was not intended to make Federal fuel economy program.521 The that relationship does not exist, nor the a substantive change to the law.526 United States opposed, as amicus assigning to the state standard of a Thus, the court concluded that ‘‘[i]f the curiae, the Green Mountain Chrysler purpose other than energy conservation, recodification worked no substantive decision on appeal to the Second diminishes the statutory implications of change in the law, then the term ‘other Circuit, but the Second Circuit did not the state standard’s meeting the motor vehicle standards of the 522 issue a decision on appeal due to the relatedness test. Those unrelated types Government’ continues to include both of emissions constitute a very low emission standards issued by EPA and 519 Green Mountain Chrysler, 508 F.Supp.2d at emission standards for which EPA has percentage of the overall tailpipe 398. issued a waiver under Section 209(b) of 520 emissions. Finally, while there are Central Valley Chrysler-Jeep, 529 F.Supp.2d at the CAA, as it did when enacted in 1165. Congress must state its intention clearly to means of compliance with the state 527 accord a state law the status of Federal law, which 1975.’’ it did not do in either in Section 209(b) of the CAA standard other than improving fuel NHTSA believes that the district court or in EPCA. See, e.g., Indep. Cmty. Bankers Ass’n economy, their contributions to misread EPCA to the point of turning it v. Bd. of Governors, 820 F.2d 428, 436–37 (D.C. Cir. compliance are minor. Improving fuel on its head. As discussed previously in 1987) (recognizing that, although Congress ‘‘has the power to assimilate state law,’’ ‘‘[s]uch decisions economy is the only feasible method of this document, the ‘‘federal standards’’ require an unequivocal congressional expression’’ achieving full compliance. Again, definition discussed by the court existed because ‘‘some [state] restrictions would in all NHTSA and EPA agree. in a self-contained scheme allowing likelihood conflict with [other] existing Federal laws’’). The Central Valley Chrysler-Jeep court manufacturers to petition NHTSA for 521 See Geier v. American Honda Motor Co., 529 went further, noting that while NHTSA modification of the fuel economy U.S. 861, 883 (2000) (‘‘Congress has delegated to is required to give consideration to requirements only between 1978 and DOT authority to implement the statute; the subject ‘‘other standards,’’ including those matter is technical; and the relevant history and 523 Central Valley Chrysler-Jeep, 529 F.Supp.2d at background are complex and extensive. The agency ‘‘promulgated by EPA,’’ ‘‘[t]here is no 1168. is likely to have a thorough understanding of its corresponding duty by EPA to give 524 own regulation and its objectives and is ‘uniquely Central Valley Chrysler-Jeep, 529 F.Supp.2d at qualified’ to comprehend the likely impact of state consideration to EPCA’s regulatory 1173 (quoting Green Mountain Chrysler, 508 requirements.’’); Medtronic, Inc. v. Lohr, 518 U.S. scheme. This asymmetrical allocation F.Supp.2d at 345). EPCA Section 502(d)(3)(D)(i) provided: ‘‘Each of the following is a category of 470, 496 (1996) (‘‘agency is uniquely qualified to by Congress of the duty to consider determine whether a particular form of state law Federal standards: . . . Emissions standards under stands as an obstacle to the accomplishment and other governmental regulations Section 202 of the Clean Air Act, and emissions execution of the full purposes and objectives of indicates that Congress intended that standards applicable by reason of Section 209(b) of Congress’’) (internal quotation marks omitted). DOT, through NHTSA, is to have the such Act.’’ 522 525 Id. See Proof Brief for the United States as burden to conform its CAFE program Amicus Curiae, 07–4342–cv (2d Cir. filed Apr. 16, 526 Id. 2008). under EPCA to EPA’s determination of 527 Id.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00252 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43237

1980, and thus has no application either Thus, EPCA went on to include requirement to consider other standards. at the time of the decision or today. And ‘‘Emissions standards under Section 202 There is no hint in the histories of either even if that definition of ‘‘federal of the CAA, and emissions standards EPCA or EISA of an intent to give other standards’’ were applied to EPCA applicable by reason of Section 209(b) of standards special, much less superior, generally, NHTSA would balance that such Act’’ in its list of ‘‘categor[ies] of status under EPCA. The limited against other factors enumerated in Federal standards.’’ 531 concerns and purpose were to ensure EPCA that it ‘‘shall’’ consider in setting Because California standards to that any adverse effects of other maximum feasible fuel economy combat smog (not GHG regulations) ‘‘by standards on fuel economy considered standards. However, the district courts’ reason of section 209(b)’’ could be in connection with the fuel economy view is that this factor instead creates an considered to reduce federal fuel standards. Those concerns are evident ‘‘obligation’’ to ‘‘harmonize’’ CAFE economy standards for three years, the in a 1974 report, entitled ‘‘Potential for standards with state emissions district courts erroneously believed that Motor Vehicle Fuel Economy regulations under a CAA Section 209 state CO2 regulations are somehow now Improvement,’’ submitted to Congress waiver.528 In other words, under the ‘‘federal’’ standards under 49 U.S.C. by the Department of Transportation district courts’ opinions, a state 32902(f). On its face, this language and EPA.535 That report noted that the standard controls what NHTSA does, applied only to three long past model weight added by safety standards would and the agency therefore has no further years and only to reducing standards, and one set of emissions standards discretion to consider the other factors not setting them. ‘‘For purposes of this might temporarily reduce the level of Congress directed it to consider. subsection’’ referred to section 502(d) of achievable fuel economy.536 These Consistent with the legislative history EPCA—not EPCA section 502(e) [now concerns can also be found in the and NHTSA’s long-standing 49 U.S.C. 32902(f)] which sets forth the congressional reports on EPCA.537 interpretations, NHTSA interprets EPCA factor of ‘‘the effect of other (f) State Tailpipe GHG Emissions EPCA, a statute which it administers in Federal motor vehicle standards on fuel Standards Conflict With EPCA and are implementing the national fuel economy.’’ After MY 1980, section Therefore Preempted Impliedly economy program, as providing that the 502(d) became obsolete. When EPCA requirement to ‘‘consider’’ the four was recodified in 1994, section 502(d) Notwithstanding that state standards EPCA statutory factors set forth in 49 was dropped as executed and therefore limiting or prohibiting tailpipe CO2 U.S.C. 32902(f) does not mean the surplusage. As the listing of Federal emissions are expressly preempted by agency is obligated to harmonize CAFE standards in 502(d) never had any EPCA, they also clearly conflict with the standards with state tailpipe CO2 application outside that subsection and objectives of EPCA and would therefore emissions standards. EPA concurs that a ceased to have significance when that also be impliedly preempted. CAA waiver does not also waive the subsection became obsolete, it had and State regulation of CO2 emissions effect of any other Federal law, has no bearing on the recodified version would frustrate Congress’ objectives in including EPCA. of EPCA. The recodification to rescind establishing the CAFE program and As discussed above in the ‘‘other this subsection, which had no conflict with NHTSA’s efforts to standards’’ section of this rulemaking, substantive significance for 14 years, implement the program in a manner NHTSA further believes that the district was entirely non-substantive.532 consistent with EPCA. While the courts in Green Mountain Chrysler and NHTSA believes that the district overarching purpose of EPCA may be Central Valley Chrysler-Jeep courts in Green Mountain Chrysler and energy conservation, Congress directed misconstrued the provision in EPCA as Central Valley Chrysler-Jeep sought to NHTSA to consider four factors in enacted in 1975 that allowed give a CAA waiver for the California establishing maximum feasible fuel manufacturers to petition NHTSA to GHG regulation an effect far beyond the economy standards. NHTSA balances reduce CAFE standards that Congress terms of the CAA provision authorizing these factors to determine, through the had set for model years 1978, 1979, and such a waiver. As discussed previously, CAFE program, the amount of energy 1980 if there was a ‘‘Federal standards the courts overlooked the fact that the the light-duty vehicle fleet should fuel economy reduction.’’ 529 This CAA itself makes clear that waiver of conserve. Allowing a state to make a provision did not involve a factor to be preemption under that statute operates state-specific determination for how balanced in determining fuel economy only to relieve application of the CAA much energy should be conserved (in standards. It provided for a reduction in preemption statute.533 State GHG the same way that the CAFE program fuel economy standards for cars at a regulations, even if subject to an EPA conserves energy) necessarily frustrates time when only conventional pollutants waiver, would remain regulations NHTSA’s efforts to make that were regulated. The provision was ‘‘adopt[ed] or enforc[ed]’’ by ‘‘a State or determination for the country as a specifically designed to address political subdivision of a State’’ and whole because it sends the industry in California’s then-existing smog therefore would be subject to different directions in order to try to regulations, particularly with regard to preemption by EPCA.534 meet multiple standards at once rather the additional weight (which other The courts’ view suggests an apparent than allowing the industry to focus its things being equal reduces fuel misunderstanding of the underlying resources and efforts on the path laid economy) associated with catalytic concerns and purposes of the out at the Federal level. This is converters. In so doing, Congress particularly true when considering that recognized the potential interplay for 531 Id. § 502(d)(3)(D). when California sets standards, other three model years between California’s 532 The recodification was ‘‘[t]o revise, codify, states can choose to adopt those smog regulations and the possibility that and enact without substantive change’’ laws related it could reduce Federal fuel economy to transportation. Public Law 103–272 (emphasis 535 standards for those model years.530 added). This report was prepared in compliance with 533 42 U.S.C. 7543(b)(1) (emphasis added); see Section 10 of the Energy Supply and Environmental also 42 U.S.C. 7543(b)(3) (‘‘compliance with such Coordination Act of 1974, Public Law 93–319. 528 Id. at 1170. State standards shall be treated as compliance with 536 See id. at 6–8 and 91–93. 529 Public Law 94–163 sec. 502(d), 89 Stat. 904– applicable Federal standards for purposes of this 537 See page 22 of Senate Report 94–179, pages 88 05. subchapter’’) (emphasis added). and 90 of House Report 94–340, and pages 155–7 530 See H.R. No. 94–340, at 87. 534 49 U.S.C. 32919(a). of the Conference Report, Senate Report 94–516.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00253 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43238 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

standards and thereby further increase rejecting the manufacturers’ arguments, feasible means to eliminate tailpipe CO2 the compliance complexity. the court held that the Vermont emissions is by eliminating the use of A critical objective of EPCA was to standards do not create an obstacle to petroleum fuel (i.e., electric or fuel cell establish a single national program to achieving EPCA’s goals because the propulsion), and because the purpose of regulate vehicle fuel economy. Vermont standards are, in the court’s the ZEV program is to affect fuel Congress, in passing EPCA, judgment, consistent with EPCA’s economy,539 ZEV mandates directly accomplished this objective by standard setting criteria. In reaching that relate to fuel economy and are thereby providing broad preemptive power conclusion, the court did not consider expressly preempted. ZEV mandates are established in the language codified at the impact of the Vermont standards on also intended to force the development 49 U.S.C. 32919(a). Other congressional the balancing done by NHTSA in setting and commercial deployment of ZEVs— objectives underlying EPCA include CAFE standards. For its part, the court regardless of the technological avoiding serious adverse economic in Central Valley Chrysler-Jeep feasibility or economic practicability of effects on manufacturers and concluded that there was no conflict doing so—putting the program entirely maintaining a reasonable amount of preemption, since if California’s at odds with critical factors that consumer choice among a broad variety standards were granted a waiver under Congress required NHTSA to consider of vehicles. To guide the agency toward CAA section 209 by EPA, they would in establishing fuel economy standards. the selection of standards meeting these satisfy CAA objectives and be consistent Therefore, ZEV mandates also interfere competing objectives, Congress with EPCA.538 The court simply with achieving the goals of EPCA and specified four factors that NHTSA must assumed consistency. If this assumption are therefore impliedly preempted. consider in determining the maximum proved incorrect, to the extent of any California’s ZEV mandate represents feasible level of average fuel economy incompatibility between the two the most prominent example. California and thus the level at which each regimes, ‘‘NHTSA is empowered to initially launched its ZEV mandate in standard must be set. As discussed revise its standards’’ to take into 1990 to force the development and above, since the only practical way to account California’s regulations, deployment of ZEVs to reduce smog- reduce tailpipe CO2 emissions is to according to that court. forming emissions. As California’s Low improve fuel economy, it would be NHTSA disagreed with the two Emission Vehicle and EPA’s Tier 3 impossible for a state tailpipe CO2 district court rulings at the time and standards for criteria pollutant emissions standard to be adopted continues to do so now. We note that emissions have become increasingly without interfering with CAFE the Vermont decision was appealed and stringent, the greater impact of standards. If a state were to establish briefed (including an Amicus Brief filed California’s ZEV mandate is the standards that have the effect of by the United States) prior to the stay reduction of tailpipe GHG emissions. In requiring a lower level of fuel economy and withdrawal of the litigation its latest iteration the ZEV mandate no than CAFE standards, those standards pursuant to the National Program longer focuses on tailpipe smog forming would be meaningless since they would arrangement described previously. emissions, a fact that CARB not reduce CO2 emissions. Instead, a NHTSA was not a party to those cases acknowledged in 2012 when applying State could only establish a standard and is not bound by these decisions. for a waiver for its Advanced Clean Car that has the effect of requiring a higher Those erroneous decisions further Program, in stating ‘‘[t]here is no criteria level of average fuel economy. Setting support the need for NHTSA, as the emissions benefit from including the standards that are more stringent than agency with expert authority to interpret ZEV proposal in terms of vehicle (tank- the fuel economy standards EPCA, to reaffirm its longstanding view to-wheel or TTW) emissions. The LEV promulgated under EPCA would upset of the preemption provision. Moreover, III criteria pollutant fleet standard is the efforts of NHTSA to balance and EPA, as the agency charged with responsible for those emission achieve Congress’s competing goals. administering the CAA, further reductions in the fleet; the fleet would Setting a standard above the level determines that CAA waivers do not become cleaner regardless of the ZEV judged by NHTSA to be consistent with ‘‘federalize’’ state standards; therefore, regulation because manufacturers would the statutory consideration after careful state standards directly affecting fuel adjust their compliance response to the consideration of these issues in a economy are subject to EPCA standard by making less polluting rulemaking proceeding would negate preemption even if there is a CAA conventional vehicles.’’ 540 the agency’s careful analysis and waiver in place. In its current configuration, the ZEV decision-making. (g) ZEV Mandates mandate requires manufacturers to For the same reasons, a state generate credits based upon the number regulation having the effect of regulating Another form of EPCA-preempted of vehicles delivered for retail sale. tailpipe carbon dioxide emissions or state regulation is a zero-emission Vehicles earn varying amounts of ZEV fuel economy is likewise impliedly vehicle (ZEV) mandate. Such laws credits depending upon technology and preempted under 49 U.S.C. Chapter 329. require that a certain number or range, with some vehicles earning The Vermont and California district percentage of vehicles sold or delivered several credits. Manufacturers court decisions discussed above for sale within a state must be ZEVs, delivering for sale certain plug-in hybrid addressed conflict preemption. The vehicles that produce neither smog- Green Mountain Chrysler court forming nor CO2 tailpipe emissions. 539 See, e.g., Fact Sheet: 2003 Zero Emission concluded that the Vermont GHG ZEV mandates may require either that Vehicle Program, California Air Resources Board standards presented no conflict actual ZEVs be sold or delivered for sale (March 18, 2004), available at https:// preemption concerns and rejected the or provide for generation and www.arb.ca.gov/msprog/zevprog/factsheets/ 2003zevchanges.pdf (stating that one of the contention that Vermont’s GHG application of ZEV credits, which may ‘‘significant features of the April 2003 changes to regulations would conflict with or may not be traded. While NHTSA has the ZEV regulation’’ included removal of ‘‘all Congress’ intent that there be a single, not previously commented on the references to fuel economy or efficiency,’’ after a nationwide fuel economy standard and relationship between the ZEV mandates 2002 lawsuit asserting that AT PZEV provisions pertaining to the fuel economy of hybrid electric that those regulations upset NHTSA’s and the CAFE program because the only vehicles were preempted by EPCA). careful balancing of the EPCA statutory 540 Docket No. EPA–HQ–OAR–2012–0562, Pp. factors in its rulemaking proceedings. In 538 529 F.Supp.2d at 1179. 15–16.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00254 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43239

vehicles earn some limited ZEV credits, This involves implementation of some based rather than design mandates. even though they are not truly ZEVs, but of the most expensive and advanced Moreover, by forcing manufacturers to such credits can only satisfy a portion technologies in the automotive industry, design, produce, and deliver for sale of a manufacturer’s ZEV credit regardless of consumer demand (which vehicles that produce no tailpipe CO2 requirements. The credit requirements tends to be lower during periods of emissions, the ZEV mandate forces increase annually, with the number of sustained relatively-low gasoline further expensive investments in fuel- required credits equaling 4.5% of a prices). The California Air Resources saving technology than NHTSA has manufacturer’s light duty vehicle sales Board’s own midterm review report for determined appropriate to require in 541 in 2018, rising to 22% in 2025. To hit their Advanced Clean Car program cites setting fuel economy standards.547 We this 22% credit requirement, a estimates from the 2016 Draft Technical seek comment on the extent to which manufacturer would need to deliver for Assessment Report relating to the compliance with the ZEV mandate sale ZEVs totaling somewhere between incremental vehicle costs of ZEVs over less than eight percent and 15.4% of 2016 vehicles with internal combustion frustrates manufacturers’ efforts to their light duty sales in California, per engines.545 While stating marginal comply with CAFE standards. various projections.542 With advance increased costs have fallen when For the reasons outlined above, the notice, manufacturers may elect to use compared to previous estimates, CARB California ZEV mandate is expressly credits earned from over-complying nevertheless still shows battery electric and impliedly preempted by EPCA. with vehicle tailpipe GHG emission subcompact vehicles with 75 miles of While EPA had previously granted a requirements toward partial satisfaction range, for which consumer demand waiver of CAA preemption for of the ZEV mandate. remains very low, as costing $7,505 California’s Advanced Clean Car The EPA has granted a waiver of CAA more than ones with an internal Program, which includes the California preemption under Section 209 of the combustion engine, with large cars ZEV mandate, this waiver has no effect CAA for California’s Advanced Clean costing $11,355 more. Battery electric on EPCA preemption of the ZEV Car program, which includes subcompacts with a 200-mile range, for mandate, as described above. California’s ZEV mandate in addition to which consumer demand is slightly California’s GHG regulation and LEV higher than a 75-mile range, were 3. Conclusion and Severability program. Nine other states have elected estimated to cost $12,001 more than to adopt the ZEV mandate pursuant to comparable vehicles with internal Given the importance of an effective, Section 177 of the CAA 543—which, combustion engines, and large cars smooth functioning national program to combined with California, represent $16,746 more. Even subcompact plug-in regulate fuel economy and in light of the approximately 30% of United States hybrids with 40 miles of electric range failure of two Federal district courts to light duty vehicle sales annually.544 cost $9,260 more than internal consider NHTSA’s analysis and Manufacturers must satisfy the ZEV combustion engine equivalents, and carefully crafted position on mandate for each state. While, $13,991 more for large cars. And as preemption, NHTSA is considering traditionally, manufacturers could apply discussed above, consumers have not taking the further step of summarizing credits earned in one state to satisfy the been willing to pay the full cost of this that position in an appendix to be added requirements of another state, this technology—meaning manufacturers are to the parts in the Code of Federal ‘‘travel’’ provision is limited only to fuel likely to spread the costs of the ZEV Regulations setting forth the passenger cell electric vehicles beginning with MY mandate to non-ZEV vehicles (and to car and light truck CAFE standards. 2018. vehicles sold in other states). This That proposed regulatory text may be Accordingly, manufacturers must expensive and market-distorting found at the end of this preamble. endeavor to design, produce, and mandate for manufacturers to eliminate deliver for sale significant numbers of vehicle tailpipe CO2 emissions (and NHTSA considers its proposed vehicles that produce zero tailpipe CO2 thus petroleum fuel use) for part of their decision on the maximum feasible emissions within each state that has fleets has always interfered with CAFE standards for MY 2021–2026 to be adopted the California ZEV mandate. NHTSA’s balancing of statutory factors severable from its decision on EPCA in establishing maximum feasible fuel preemption. Our proposed 541 Cal. Code Regs. tit.13, sec. 1962.2(b). economy standards, and increasing ZEV interpretation of 49 U.S.C. 32919 does 542 The Air Resources Board initially projected credit requirements through 2025 make not depend on our decision to finalize that 15.4% of new vehicles delivered for sale would consist of ZEVs. See., e.g., Staff Report: Initial it all-the-more of an obstacle to and a court’s decision to uphold, the Statement of Reasons 2012 Proposed Amendments accomplishing EPCA’s goal of CAFE standards being proposed today to the California Zero Emission Vehicle Program establishing a coherent national fuel under 49 U.S.C. 32902. NHTSA solicits Regulations, California Air Resources Board at 48 economy program. Unlike NHTSA’s (Dec. 7, 2011), available at https://www.arb.ca.gov/ comment on the severability of these regact/2012/zev2012/zevisor.pdf (stating ‘‘[b]y CAFE program, the ZEV mandate forces actions. model year 2025, staff expects 15.4 percent of new investment in specific technology sales will be ZEVs and [Plug-In Hybrids].’’) (electric and fuel cell technology) rather However, an increased supply of credits and than allowing manufacturers to improve projected increases in battery electric range has resulted in others projecting reduced required ZEV fuel economy through more cost- fleet penetration. See, e.g., What is ZEV?, Union of effective technologies that better reflect Concerned Scientists (Oct. 31, 2016), https:// consumer demand.546 This appears to 547 See, e.g., Alan, J., Hardman, S. & Carley, S. www.ucsusa.org/clean-vehicles/california-and- conflict directly with Congress’ intent Cost implications for automakers’ compliance with western-states/what-is-zev (projecting ‘‘about 8 emission standards from Zero Emissions Vehicle percent of sales to be ZEVs’’ in 2025). that CAFE standards be performance- mandate, TRB 2018 Annual Meeting paper 543 These states are Connecticut, Maine, submittal, https://trid.trb.org/view/1495714 (last Maryland, Massachusetts, New Jersey, New York, 545 California Air Resources Board, California’s accessed June 28, 2018) (finding based on Oregon, Rhode Island, and Vermont. Advanced Clean Cars Midterm Review, Appendix independent research that in 2025, costs reach 544 See Automotive Retailing: State by State, C, Zero Emission Vehicle and Plug-in Hybrid National Automobile Dealers Association, https:// Electric Vehicle Technology Assessment, Table 8, at approximately $1,500 per vehicle on average to www.nada.org/statedata/ (last visited June 25, C–64 (Jan. 18, 2017), available at https:// comply with CAFE alone and increase to around 2018) (estimating that these states represented www.arb.ca.gov/msprog/acc/mtr/appendix_c.pdf. $2,100 per vehicle on average to comply with both 28.6% of new motor vehicle registrations in 2016). 546 13 Cal. Code of Regulations 1962.2. CAFE and ZEV).

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00255 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43240 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

B. Preemption Under the Clean Air Act Section 209(b)(1)(A)–(C), 42 U.S.C. conditions), EPA proposes to find that 7543(b)(1)(A–(C) (Emphasis added).550 California does not need its GHG and 1. Background Any one of these three findings operates ZEV standards to meet compelling and (a) Statutory Background: Clean Air Act to forbid a waiver. extraordinary conditions because those Section 209(a) Preemption, Section standards address environmental (1) EPA’s Proposed Action 209(b)(1) California Waiver, and Section problems that are not particular or 209(b)(1)(A)–(C) Prohibitions on Waiver EPA is proposing to withdraw the unique to California, that are not caused EPA’s regulation of new motor January 9, 2013 waiver of preemption by emissions or other factors particular vehicles under Title II generally for California’s Advanced Clean Car or unique to California, and for which preempts state standards in the same (ACC) program, Zero Emissions Vehicle the standards will not provide any subject area. Section 209(a) of the Act (ZEV) mandate, and Greenhouse Gas remedy particular or unique to provides that: (GHG) standards that are applicable to California. new model year (MY) 2021 through Third, under section 209(b)(1)(C) ‘‘No State or any political subdivision 2025. 78 FR 2145 (January 9, (consistency with section 202(a)), EPA thereof shall adopt or attempt to enforce any 2013.) 551 552 EPA proposes to do so on proposes to find that California’s GHG standard relating to the control of emissions from new motor vehicles or new motor multiple grounds. and ZEV standards are inconsistent with vehicle engines subject to this part. No State First, EPA notes that elsewhere in this section 202(a) because they are shall require certification, inspection or any notice NHTSA has proposed to find that technologically infeasible in that they other approval relating to the control of California’s GHG and ZEV standards are provide sufficient lead time to permit emissions from any new motor vehicle or preempted under EPCA. Although EPA the development of necessary new motor vehicle engine as condition has historically declined to consider as technology, giving appropriate precedent to the initial retail sale, titling (if part of the waiver process whether consideration to compliance costs.553 any), or registration of such motor vehicle, California standards are constitutional EPA therefore proposes to make 548 motor vehicle engine, or equipment.’’ or otherwise legal under other Federal findings under sections 209(b)(1)(B) and However, Title II affords special statutes apart from the Clean Air Act, (C), either of which, as discussed above, treatment to California: Subject to EPA believes that this notice presents a independently triggers the statutory certain conditions, it may obtain from unique situation and that it is prohibition that ‘‘no such waiver will be EPA a waiver of section 209(a) appropriate to consider the implications granted.’’ preemption. Specifically, section of NHTSA’s proposed conclusion as In addition, EPA proposes to 209(b)(1) of the Act requires the part of EPA’s reconsideration of the conclude that States may not adopt Administrator, after an opportunity for waiver. In this regard, EPA is proposing California’s GHG standards pursuant to public hearing, to waive application of to conclude that state standards section 177 because the text, context, the prohibitions of section 209(a) to preempted under EPCA cannot be and purpose of section 177 support the California, if California determines that afforded a valid waiver of preemption conclusion that this provision is limited its State standards will be, in the under CAA 209(b). Accordingly, EPA is to providing States the ability, under aggregate, at least as protective of public proposing to conclude that if NHTSA certain circumstances and with certain health and welfare as applicable Federal finalizes a determination that conditions, to adopt and enforce standards.549 A waiver under section California’s GHG and ZEV standards are standards designed to control criteria 209(b)(1) allows California to ‘‘adopt preempted, then it would be necessary pollutants to address NAAQS [and] enforce a[] standard relating to the to withdraw the waiver separate and nonattainment. control of emissions from new motor apart from the analysis under section (2) History of Waiver for California GHG vehicles or new motor vehicle engines.’’ 209(b)(1)(B), (C) that follows. and ZEV Standards, and Associated CAA section 209(a), 42 U.S.C. 7543(a). Second, under section 209(b)(1)(B) Issues of Statutory Interpretation But California’s ability to obtain a (compelling and extraordinary waiver is not unlimited. The statute In December 2005, California for the first time applied to EPA for a provides that ‘‘no such waiver will be 550 As presented in the United States Code, the granted’’ if the Administrator finds any cross-reference in prong (C) is to ‘‘section 7521(a) preemption waiver for GHG standards of the following: ‘‘(A) [California’s] of this title,’’ i.e., CAA section 201(a), 42 U.S.C. for MY 2009 and following. EPA denied 7521(a), which governs EPA’s administration of determination [that its standards in the this request in March 2008, relying on ‘‘Emission standards for new motor vehicles or new the second prong under section aggregate will be at least as protective] motor vehicle engines administration of ‘‘Emissions is arbitrary and capricious, (B) standards for new motor vehicles or new motor 209(b)(1)(B) and finding that California [California] does not need such State vehicle engines.’’ did not need those standards to meet 551 standards to meet compelling and This proposed action does not address compelling and extraordinary whether the statutory interpretations and their conditions. In doing so, it noted that extraordinary conditions, or (C) such policy consequences laid out in the proposal may GHG standards, unlike prior standards State standards and accompanying have implications for past waivers granted to for which California had requested and enforcement procedures are not California for other standards besides its GHG and ZEV standards. EPA proposes to take this action in received waivers, are designed to consistent with section [202(a)].’’ the context of this joint rulemaking with NHTSA, address global air pollution problems— and the California standards identified herein are not air pollution problems specific to 548 Clean Air Act (CAA) section 209(a), 42 U.S.C. the focus of EPA’s proposal. As circumstances 7543(a). require and resources permit, EPA may in future California. 73 FR 12156, March 6, 2008. 549 CAA section 209(b), 42 U.S.C. 7543(b). The actions consider whether this proposal, if finalized, provision does not identify California by name. makes it appropriate or necessary to revisit past 553 Under section 209(b)(1)(C) of the CAA, EPA Rather, it applies on its face to ‘‘any State which grants of other waivers beyond those granted with must deny California’s waiver request if EPA finds has adopted standards (other than crankcase respect to California’s GHG and ZEV program. that California’s standards and accompanying emission standards) for the control of emissions 552 EPA proposes to withdraw the waiver for enforcement procedures are not consistent with from new motor vehicles or new motor vehicle these model years because these are the model years section 202(a). Section 202(a) provides that an engines prior to March 30, 1966.’’ California is the at issue in NHTSA’s proposal. EPA solicits emission standard shall take effect after such period only State that meets this requirement. See S. Rep. comment on whether one or more of the grounds of time as the Administrator finds necessary to No. 90–403 at 632 (1967). This proposal refers supporting the proposed withdrawal of this waiver permit development and application of the requisite interchangeably to ‘‘California’’ and ‘‘CARB’’ (the would also support withdrawing other waivers that technology, giving appropriate consideration to California Air Resources Board). it has previously granted. compliance costs.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00256 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43241

Due to this new circumstance, EPA conditions’’ to mean environmental regulations pursuant to CAA section reconsidered its historic interpretation problems with causes that were specific 209(b). 78 FR 2112. The ACC program and application of section 209(b)(1)(B). to California—given that those is a single coordinated package Although today’s proposal contains standards were designed to address comprising regulations for ZEV and proposed findings under each prong of global air pollution problems as low-emission vehicles (LEV) 209(b)(1), prong (B) was the only one at compared to local or regional air regulations,556 for new passenger cars, issue in the 2008 waiver denial (and pollution problems caused specifically light-duty trucks, medium-duty EPA’s subsequent reversal), and it by certain conditions in California. passenger vehicles, and certain heavy- merits extended discussion at the outset EPA in the 2008 waiver denial also duty vehicles, for MY 2015 through due to its central significance in the applied a second, alternative 2025. Thus, in terms of proportion, the policy and legal context and the history construction of section 209(b)(1)(B). ACC program is comparable to the underlying today’s proposal. Under this alternative construction, EPA combined Federal Tier 3 Motor Vehicle As a general matter, EPA had considered whether the impacts of Emissions Standards and the 2017 and historically interpreted section climate change in California were later MY Light-duty Vehicle GHG 209(b)(1)(B) to require EPA to consider sufficiently different enough from the Standards.557 According to CARB, the whether, to meet compelling and impacts felt in the rest of the country ACC program was intended to address extraordinary conditions in California, such that California could be considered California’s near and long-term smog the state needs to have its own separate to need its GHG standards to meet issues as well as certain specific GHG new motor vehicle program in the compelling and extraordinary emission reduction goals.558 78 FR aggregate.554 Under this historical conditions—interpreting ‘‘compelling 2114. See also 78 FR 2122, 2130–31. approach, EPA considered California’s and extraordinary conditions’’ to mean The ACC program regulations impose need for a separate program as a whole, environmental effects specific to multiple and varying complex rather than California’s need for the California. compliance obligations that have particular aspect of the program for The next year, following a simultaneous, and sometimes which California sought a waiver in any presidential election and change in overlapping, deadlines with each particular instance. (Typically, prior to administration, EPA reconsidered the its ACC program waiver request, 2008 denial at California’s request. On 556 The LEV regulations in question include standards for both GHG and criteria pollutants California would seek a waiver for only reconsideration, EPA reversed course and granted a waiver for California’s (including ozone and PM). Amendments for the particular aspects of its new motor LEV III program included replacement of separate vehicle program.) In the 2008 GHG GHG standards. 74 FR 32744 (July 9, nonmethane organic gas (NMOG) and oxides of waiver denial, EPA determined that this 2009). In granting the waiver, EPA nitrogen (NOX) standards with combined NMOG reverted to its historical interpretation plus NOX standards, which provides automobile interpretation was inappropriate under of section 209(b)(1)(B), under which it manufacturers with additional flexibility in meeting the circumstances. the new stringent standards; an increase of full had construed ‘‘compelling and In its 2008 waiver denial, EPA useful life durability requirements from 120,000 extraordinary conditions’’ to mean miles to 150,000 miles, which guarantees vehicles proceeded under two alternative environmental problems caused by sustain these extremely low emission levels longer; constructions of the statute. Under both conditions specific to California and/or a backstop to assure continued production of super- of these constructions, EPA determined ultra-low-emission vehicles after partial-zero- effects experienced to a unique degree that it was a reasonable interpretation of emission vehicles (PZEVs) as a category are moved or in a unique manner in California, and from the ZEV regulations to the LEV regulations in section 209(b)(1)(B) to require a separate under which it had evaluated 2018; more stringent particulate matter (PM) review of California’s need for standards California’s need for its own, separate standards for light- and medium-duty vehicles, designed to address a global air which will reduce the health effects and premature new motor vehicle program as a whole, pollution problem and its effects, as deaths associated with these emissions; zero fuel rather than California’s need for the evaporative emission standards for PCs and LDTs, distinct from other portions of specific aspects of its separate program and more stringent standards for medium- and California’s new motor vehicle program, for which it was seeking a waiver. In heavy-duty vehicles (MDVs); and, more stringent which up until then had been designed supplemental federal test procedure (SFTP) reverting to this determination, the EPA standards for PC and LDTs, which reflect more to address local or regional air pollution aggressive real world driving and, for the first time, 555 necessarily determined that it makes no problems. Under the first difference whether California seeks a require MDVs to meet SFTP standards. 78 FR 2114. construction, EPA found it relevant that waiver to implement separate standards 557 78 FR 23641, April 22, 2016; 77 FR 62624, elevated GHG concentrations in in response to its own specific, local air October 15, 2012. California were similar to 558 ‘‘The Advanced Clean Cars program . . . will pollution problems, or whether reduce criteria pollutants . . . and . . . help concentrations found elsewhere in the California seeks a waiver to implement achieve attainment of air quality standards; The world, and that local conditions in separate standards designed to address Advanced Clean Cars Program will also reduce California, such as the local topography, a global air pollution problem. greenhouse gases emissions as follows: by 2025, the local climate, and the significant CO2 equivalent emissions will be reduced by 13 Since 2009, EPA has continued to million metric tons (MMT) per year, which is 12 number of motor vehicles in California, adhere to this interpretation and percent from base line levels; the reduction were not the determining factors application of section 209(b)(1)(B) when increases in 2035 to 31 MMT/year, a 27 percent causing the elevated GHG reviewing CARB’s waiver requests, reduction from baseline levels; by 2050, the concentrations found in California and proposed regulation would reduce emissions by regardless of whether the waiver was more than 40 MMT/year, a reduction of 33 percent elsewhere. In sum, EPA found that requested with regard to standards from baseline levels; and viewed cumulatively over California did not need its GHG designed to address traditional, local the life of the regulation (2017–2050), the proposed standards to meet ‘‘compelling and environmental problems, or global Advanced Clean Cars regulation will reduce by extraordinary conditions’’—interpreting more than 850 MMT CO2-equivalent, which will climate issues. In this proposal, the EPA help achieve the State’s climate change goals to ‘‘compelling and extraordinary proposes to determine that this reduce the threat that climate change poses to reversion to the pre-2008 interpretation California’s public health, water resources, 554 See, e.g., 49 FR 18887 (May 3, 1984). was not appropriate. agriculture industry, ecology and economy.’’ 78 FR 555 Criteria pollutants generally present public 2114. CARB Resolution 12–11, at 19, (January 26, health and environmental concern in proportion to On January 9, 2013, EPA granted 2012), available in the docket for the January 2013 their ambient local concentration and California has CARB’s request for a waiver of waiver action, Document No. EPA–HQ–OAR–2012– long had unusually severe problems in this regard. preemption to enforce its ACC program 0562, the docket for the ACC program waiver.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00257 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43242 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

standard. These deadlines began in 2015 pollutant fleet standard was responsible the ACC program can be found in and are scheduled to be phased in for those emission reductions.560 CARB CARB’s waiver request, located in the through 2025. For example, compliance further noted that its ZEV regulation docket for the January 2013 waiver with the GHG requirements began in was intended to focus primarily on zero action, Docket No. EPA–HQ–OAR– 2017 and will be phased-in through emission drive—that is, battery electric 2012–0562. 2025. The implementation schedule and (BEVs), plug-in hybrid electric vehicles 2. Statutory Provisions Applicable to the the interrelationship of regulatory (PHEVs), and hydrogen fuel cell Proposed Action provisions with each of the three vehicles (FCVs)—in order to move standards together demonstrates that advanced, low GHG vehicles from Under section 209(b) of the Clean Air CARB intended that at least the GHG demonstration phase to Act, EPA may reconsider a grant of a and ZEV standards, if not also the LEV commercialization (78 FR 2122, 2130– waiver of preemption and withdraw standards, would be implemented as a 31). Specifically, for 2018 MY through same if the Administrator makes any cohesive program. For example, in its 2025 MY, the ACC program ZEV one of the three findings in section ACC waiver request, CARB stated that requirements mandate use of 209(b)(1)(A), (B) and (C). EPA’s the ‘‘ZEV regulation must be considered technologies such as BEVs, PHEVs and authority to reconsider and withdraw in conjunction with the proposed LEV FCVs, in up to 15% of a manufacturer’s the grant of a waiver for the ACC III amendments. Vehicles produced as a California fleet and in each of the program is implicit in section 209(b) result of the ZEV regulation are part of section 177 States by MY 2025 561 (78 given that the authority to revoke a grant a manufacturer’s light-duty fleet and are FR 2114). Additionally, the ACC of authority is implied in the authority therefore included when calculating program regulations provide various for such a grant. Further support for fleet averages for compliance with the compliance flexibilities allowing for EPA’s authority is based on the LEV III GHG amendments.’’ CARB’s substitution of compliance with one legislative history for section 209(b), Initial Statement of Reasons at 62–63.559 program requirement for another. For and the judicial principle that agencies CARB also noted ‘‘[b]ecause the ZEVs instance, manufacturers may opt to possess inherent authority to reconsider have ultra-low GHG emission levels that over-comply with the GHG fleet their decisions.564 The legislative are far lower than non-ZEV technology, standard in order to offset a portion of history from the 1967 CAA amendments they are a critical component of their ZEV compliance requirement for where Congress enacted the provisions automakers’ LEV III GHG standard MY 2018 through 2021. Further, until now codified in section 209(a) and (b) compliance strategies.’’ Id. CARB MY 2018, sales of BEVs (since MY 2018, provides support for this view. The further explained that ‘‘the ultra-low limited to FCVs) in California count Administrator has ‘‘the right . . . to GHG ZEV technology is a major toward a manufacturer’s credit withdraw the waiver at any time [if] component of compliance with the LEV requirement in section 177 States. This after notice and an opportunity for III GHG fleet standards for the overall is known as the ‘‘travel provision’’ (78 public hearing he finds that the State of light duty fleet.’’ Id. CARB’s request also FR 2120).562 For their part, the GHG California no longer complies with the repeatedly touted the GHG emissions emission regulations include an conditions of the waiver.’’ S. Rep. No. benefits of the ACC program. optional compliance provision that 50–403, at 34 (1967). Additionally, Up until the ACC program waiver allows manufacturers to demonstrate subject to certain limitations, request, CARB had relied on the ZEV compliance with CARB’s GHG administrative agencies possess requirements as a compliance option for standards by complying with applicable inherent authority to reconsider their reducing criteria pollutants. Federal GHG standards. This is known decisions in response to changed Specifically, California first included as the ‘‘deemed to comply’’ circumstances. It is well settled that the ZEV requirement as part of its first provision.563 A complete description of EPA has inherent authority to LEV program, which was then known as reconsider, revise, or repeal past LEV I, that mandated a ZEV sales 560 CARB ACC waiver request at EPA–HQ–OAR– decisions to the extent permitted by law requirement that phased-in starting with 2012–0562–0004. so long as the Agency provides a the 1998 MY through 2003 MY. EPA 561 Under section 177, any State that has state reasoned explanation. This authority issued a waiver of preemption for these implementation plan provisions approved under exists in part because EPA’s part D of Subchapter I of the Act may opt to adopt regulations on January 13, 1993 (58 FR and enforce standards that are identical to interpretations of the statutes it 4166 (January 13, 1993). Since this standards for which EPA has granted a waiver of administers ‘‘are not carved in stone.’’ initial waiver of preemption, California preemption to California under CAA section 209(b). Chevron U.S.A. Inc. v. NRDC, Inc., 467 has made multiple amendments to the EPA’s longstanding interpretation of section 209(b) U.S. 837, 863 (1984). An agency ‘‘must and its relationship with section 177 is that it is not ZEV requirements and EPA has appropriate under section 209(b)(1)(C) to review consider varying interpretations and the subsequently granted waivers for those California regulations, submitted by CARB, through wisdom of its policy on a continuing amendments. In the ACC program the prism of adopted or potentially adopted basis.’’ Id. at 863–64. This is true when, waiver request California also included regulations by section 177 States. as is the case here, review is undertaken a waiver of preemption request for ZEV 562 On March 11, 2013, the Association of Global Automakers and Alliance of Automobile ‘‘in response to . . . a change in amendments that related to 2012 MY Manufacturers filed a petition for reconsideration of administration.’’ National Cable & through 2017 MY and imposed new the January 2013 waiver grant, requesting that EPA Telecommunications Ass’n v. Brand X requirements for 2018 MY through 2025 reconsider the decision to grant a waiver for MYs internet Services, 545 U.S. 967, 981 MY (78 FR 2118–9). Regarding the ACC 2018 through 2025 ZEV standards on technological feasibility grounds. Petitioners also asked for (2005). The EPA must also be cognizant program ZEV requirements, CARB’s consideration of the impact of the travel provision, waiver request noted that there was no which they argue raise technological feasibility https://www.arb.ca.gov/msprog/levprog/leviii/ criteria emissions benefit in terms of issues in section 177 States, as part of the agency’s leviii_dtc_notice05072018.pdf. vehicle (tank-to-wheel—TTW) review under section 209(b)(1)(C). EPA continues to 564 In 2009, EPA reconsidered the 2008 GHG evaluate the petition. waiver denial at CARB’s request and granted it emissions because its LEV III criteria 563 On May 7, 2018, California issued a notice upon reconsideration. 72 FR 32744. The EPA noted seeking comments on ‘‘potential alternatives to a the authority to ‘‘withdraw a waiver in the future 559 Available in the docket for the January 2013 potential clarification’’ of this provision for MY if circumstances make such action appropriate.’’ waiver decision, Docket No. EPA–HQ–OAR–2012– vehicles that would be affected by revisions to the See 74 FR 32780 n.222; see also 32752–53 n.50 0562. Federal GHG standards. The notice is available at (citing 50 S. Rep. No. 403, at 33–34), 32755 n.74.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00258 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43243

where it is changing a prior position and taking into account statutory mandates questions as to CARB’s technological articulate a reasoned basis for the and any applicable court orders.565 projections for ZEV-type technologies, change. FCC v. Fox Television Stations, EPA is proposing to withdraw the which are a compliance option for both Inc., 556 U.S. 502, 515 (2009). This grant of a waiver of preemption for the ZEV mandate and GHG standards. proposal reflects changed circumstances California to enforce the GHG and ZEV As also previously discussed, above, that have arisen since the initial grant of standards of the ACC program for MY CARB’s ZEV regulations include the the 2013 ACC program waiver of 2021–2025. EPA proposes to withdraw travel provision, which previously preemption. They include the agency’s due to separate proposed findings under allowed manufacturers to earn credit for 566 reconsideration of California’s record section 209(b)(1)(B), and (C). ZEVs sold in California (which, despite Under section 209(b)(1)(B), EPA is support for, and EPA’s decision and very slow ZEV sales, far outpaces any proposing to find that California does underlying statutory interpretation on, other State in these sales) to comply not need its ZEV and GHG standards to California’s need for GHG and ZEV with credit requirements in section 177 meet compelling and extraordinary States. Starting with MY 2018, this standards, as well as costs and conditions in California. EPA is provision only applies to FCVs. When technological feasibility considerations proposing to find that CARB does not the travel provision was adopted, it was that differ from California’s assumptions need its own GHG and ZEV standards anticipated that by MY 2018, incentives and which were bases for agency to meet compelling and extraordinary of this type for BEV sales would no conclusions that were made at that time. conditions in California given that longer be necessary—i.e., that When California submits a package of ‘‘compelling and extraordinary consumers would adopt such vehicles standards for EPA review pursuant to conditions’’ mean environmental on their own. Unfortunately, there has CAA section 209, EPA has long problems with causes and effects in been a serious lack of market interpreted the statute as authorizing California whereas GHG emissions penetration, consumer demand levels, EPA to approve certain provisions and present global air pollution problems. and lack of or slow development of defer action on others. EPA believes this Additionally, California does not need necessary infrastructure for any ZEVs— approach of partially approving the ZEV requirements to meet BEV or otherwise—in such States. This submissions is implicit in section 209, ‘‘compelling and extraordinary’’ in turn means that manufacturers’ sales particularly given the fact that EPA’s conditions in California given that it of ZEVs in section 177 States are evaluation of the technological allows manufacturers to generate credits unlikely, contrary to CARB’s projections feasibility of standards is best in section 177 states as a means to in its submissions to support its understood as in effect an evaluation of satisfy those manufacturers’ obligations application for the ACC waiver, to generate sufficient credits to satisfy each standard for each year (i.e., to comply with the mandate that a certain percentage of their vehicles sold those manufacturers’ obligations to standards that are submitted together in California be ZEV (or be credited as comply with the mandate that a certain may vary substantially in their effect such from sales in section 177 States). percentage of their vehicles sold in and some may require longer lead time Under section 209(b)(1)(C), EPA is California be ZEV (or be credited as than others). Furthermore, since proposing to find that CARB’s GHG and such from sales in section 177 States). California always retains the authority ZEV standards are not consistent with In short, EPA is now of the view that as a matter of state law to determine section 202(a) based on changed CARB’s projections and assumptions at whether to implement state standards circumstances since the January 2013 the time of the waiver request were for which a waiver of preemption has waiver. Specifically, the agency is, in overly ambitious and likely will not be been granted, we do not believe this this action, jointly proposing with realized within the provided lead time. approach poses the risk that a partial NHTSA revisions to the Federal GHG Thus, EPA is also proposing to find that approval could force California to and fuel economy standards based on CARB’s ZEV standards for MY 2021 implement a program they would not proposed conclusions that the current through 2025, and the GHG standards have chosen had they anticipated EPA’s (or augural) standards for MY 2021 which rely on the ZEV requirement as decision. EPA believes that because its through 2025 are not feasible. The a compliance option, are technologically authority to grant waivers of preemption proposed findings in this notice call infeasible and therefore, not consistent is best understood as applying on a into question CARB’s projections and with section 209(b)(1)(C). granular level—where the feasibility of assumptions that underlay the As described above, EPA is proposing compliance for a particular year can be technological feasibility findings for its to withdraw the waiver with respect to assessed—rather than being limited to waiver application for the GHG California’s ZEV standards based on approving or disapproving preemption standards and thus the technological findings made pursuant to sections for an entire package of standards findings made by EPA in 2013 in 209(b)(1)(B) and 209(b)(1)(C). EPA is submitted together, it follows that EPA’s connection with the grant of the waiver proposing to withdraw the waiver with authority to withdraw the grant of for the ACC program. respect to California’s GHG standards waiver of preemption should also apply Similarly, with regard to ZEV based on findings made under these on a granular level, i.e., for any model standards, this notice also raises three prongs as well as a separate year for which EPA concludes the finding made under section 209(b)(1)(B). 565 Additionally, because the ZEV and GHG conditions for waiver of preemption no On March 11, 2013, EPA received a petition for reconsideration from the Association of Global standards are closely interrelated, as longer exist or for which it concludes Automakers and Alliance of Automobile demonstrated by the description above that it erred in its prior determination Manufacturers of the decision to grant a waiver for of their complex, overlapping that one of the conditions triggering a MYs 2018 through 2025 ZEV standards. 566 compliance regimes, EPA is proposing denial a waiver was not met. Further, Under this provision, a waiver is not permitted if (A) the protectiveness determination of to withdraw the waiver of preemption because neither the Clean Air Act nor the State is arbitrary and capricious; (B) the State for ZEV standards under the second and the Administrative Procedure Act does not need such State standards to meet third prongs of section 209(b)(1). specify deadlines for reconsideration of compelling and extraordinary conditions; or (C) EPA believes that a finding made agency action, EPA may, issue a new such State standards and accompanying enforcement procedures are not consistent with pursuant to any of the prongs of section final action to change a prior action, section 202(a) of the Act. 209(b)(1) is an independent and

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00259 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 43244 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules

adequate ground to withdraw the Motor and Equip. Mfrs. Ass’n v. EPA, California with the broadest possible waiver. In this regard, EPA notes that 627 F.2d 1095, 1112 (D.C. Cir. 1979) discretion in setting regulations it finds the statute provides that ‘‘No such (MEMA I). Additionally, as discussed in protective of the public health and waiver shall be granted if the section III.B, below, EPA proposes to welfare.’’ 571 Administrator finds that—(B) the State find that there is clear and compelling With respect to the consistency does not need such State standards to evidence that California does not need finding, MEMA I did not articulate a meet compelling and extraordinary its ZEV and GHG standards to meet standard of proof applicable to all conditions; or (C) such State standards compelling and extraordinary proceedings but found that the and accompanying enforcement conditions. Similarly, as discussed in opponents of the waiver were unable to procedures are not consistent with section III.C, below, there is clear and meet their burden of proof even if the section 202(a) of the Act.’’ (Emphasis compelling evidence that both the ZEV standard were a mere preponderance of added.) Consequently, a final waiver and GHG standards are not the evidence. withdrawal decision that relies on more technologically feasible.567 As the agency has consistently than one of these provisions would In MEMA I, 627 F.2d 1095, the U.S. explained, although MEMA I did not present independent and severable Court of Appeals for D.C. Circuit found explicitly consider the standard of proof bases for the decision to withdraw. And, that ‘‘the burden of proving [that for ‘‘standards,’’ as compared to separate and apart from its analysis California’s regulations do not comply ‘‘accompanying enforcement under 209(b)(1)(A)–(C), EPA proposes to with the CAA] is on whoever attacks procedures,’’ nothing in the opinion determine that if NHTSA finalizes its them. California must present its suggests that the court’s analysis would proposed determination that EPCA regulations and findings at the hearing not apply with equal force to such preempts California’s standards, that and thereafter the parties opposing the determinations.572 Moreover, the would provide an independent and waiver request bear the burden of normal standard of proof in civil matters adequate ground to withdraw the waiver persuading the Administrator that the is a preponderance of the evidence. for those standards. EPA proposes to waiver request should be denied.’’ 568 International Harvester Co. v. interpret section 209(b)(1) to only MEMA I dealt with a challenge Ruckelshaus, 478 F.2d 615, 643 (D.C. authorize it to waive CAA preemption brought by Motor and Equipment Cir. 1979). for standards that are not independently Manufacturers Association against The role of the Administrator in preempted by EPCA. EPA’s grant of a waiver of preemption considering California’s application for Additionally, under CAA section 177, for California’s accompanying a preemption waiver is to make a States that have designated enforcement procedures, which in this reasonable evaluation of the information nonattainment areas may opt to adopt instance were vehicle in-use in the record in coming to the waiver and enforce standards that are identical maintenance regulations. The specific decision. The Administrator is required to standards for which EPA has granted challenge to EPA’s action contested to ‘‘consider all evidence that passes the a waiver of preemption to California EPA’s findings that section 209 allowed threshold test of materiality and . . . under CAA section 209(b). For States for a waiver of preemption for CARB’s thereafter assess such material evidence that have adopted the ZEV standards, in-use maintenance regulations. MEMA against a standard of proof to determine the consequence of any final withdrawal I also specifically considered the whether the parties favoring a denial of action would be that they cannot standards of proof for two findings that the waiver have shown that the factual implement these standards. (A State EPA must make in order to grant a circumstances exist in which Congress may not ‘‘make attempt[s] to enforce’’ waiver for an ‘‘accompanying intended a denial of the waiver.’’ 573 California standards for which EPA has enforcement procedure’’ (as opposed to As the court in MEMA I stated, if the not waived preemption. Motor Vehicle standards): (1) Protectiveness in the Administrator ignores evidence Mfrs. Ass’n v. NYS Dep. of Envtl aggregate and (2) consistency with demonstrating that the waiver should Conservation, 17 F.3d 521, 534 (2d Cir. section 202(a) findings. The court not be granted, or if he seeks to 1994)). Where states have adopted instructed that ‘‘the standard of proof overcome that evidence with CARB’s ZEV and GHG standards into must take account of the nature of the unsupported assumptions of his own, their SIPs, under section 177, the risk of error involved in any given he runs the risk of having his waiver provisions of the SIP would continue to decision, and it therefore varies with the decision set aside as ‘‘arbitrary and be enforceable until revised. If this finding involved. We need not decide capricious.’’ 574 Therefore, the proposal is finalized, EPA may how this standard operates in every Administrator’s burden is to act subsequently consider whether to waiver decision.’’ 569 ‘‘reasonably.’’ 575 employ the appropriate provisions of The court upheld the Agency’s The instant action involves a decision the CAA to identify provisions in position that denying a waiver required whether to withdraw a previous grant of section 177 states’ SIPs that may require ‘‘clear and compelling evidence’’ to a waiver of preemption as compared to amendment and to require submission show that proposed enforcement the initial evaluation of and decision of such amendments. procedures undermine the whether to grant a waiver request from EPA is taking comments on all aspects protectiveness of California’s California. Specifically, as discussed in of this proposal. standards.570 The court noted that this Section III, below, EPA is proposing (a) Burden and Standard of Proof in standard of proof ‘‘also accords with the findings for the withdrawal of Waiver Decisions congressional intent to provide preemption for CARB’s ACC program under multiple criteria set out in section Here, the Administrator is proposing 567 EPA is assuming without agreeing that the 209(b)(1). For example, EPA is the withdrawal of a previously granted burden of proof requires clear and compelling proposing to withdraw the waiver based waiver of preemption. As discussed in evidence but believes a preponderance of the section III.A. below, EPA proposes to evidence is the proper burden of proof. Regardless, EPA firmly believes that it has clear and compelling 571 Id. find that there is clear and compelling evidence to support the agency’s statutory findings. 572 74 FR 32748. evidence that California’s protectiveness 568 MEMA I, 627 F.2d at 1122. 573 MEMA I, 627 F.2d at 1122. determination for its ZEV and GHG 569 Id. 574 Id. at 1126. standards was arbitrary and capricious. 570 Id. 575 Id.

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00260 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2 Federal Register / Vol. 83, No. 165 / Friday, August 24, 2018 / Proposed Rules 43245

on considerations such as the nature of confines of state air pollution programs in section 177 states as a means to GHG concentrations as a global air as covered by section 209(b)(1)(B). It satisfy those manufacturers’ obligations pollution problem, rather than a would also be consistent with the GHG to comply with the mandate that a regional or local air pollution problem, waiver denial for EPA to exercise its certain percentage of their vehicles sold whether or not CARB’s particular GHG own judgment in making the requisite in California be ZEV (or be credited as standards actually would reduce GHG findings called for under section such from sales in section 177 States). emissions in California, whether a 209(b)(1)(B) in the instant action. This finding is premised on agency waiver for CARB’s GHG standards is EPA is, thus, soliciting comments on review of the interpretation and permissible if those regulations are the appropriate burden and standard of application of section 209(b)(1)(B) in the preempted by EPCA, and the effect of proof for withdrawing a previously January 2013 ACC waiver request. Thus, technological infeasibility for the 2012 issued waiver, taking into consideration EPA is required to articulate a reasoned Federal GHG standards for MY 2021– that different approaches may apply to basis for the change in its position. FCC 2025. Natural Resources Defense the various criteria of Section 209(b) v. Fox Television Stations, Inc., 556 U.S. Council v. EPA, 655 F.2d 318, 331 (D.C. and that EPA is not merely responsible 502, 515 (2009). Cir. 1981) (‘‘[T]here is substantial room for evaluating a request by California and comments thereon but is proposing (2) Historical Waiver Practices Under for deference to the EPA’s expertise in Section 209(b)(1)(B) projecting the likely course of withdrawal of a grant of preemption. [technological] development.’’) Up until the 2008 GHG waiver denial, 3. Discussion: Analysis Under Section EPA had interpreted section 209(b)(1)(B) (Emphasis added.) EPA believes that 209(b)(1)(B), (C) these are kinds of issues that extend as requiring a consideration of well beyond the boundaries of (a) Proposed Finding Under Section California’s need for a separate motor California’s authority under section 209(b)(1)(B): California Does Not Need vehicle program designed to address 209(b). EPA posits, therefore, that the its Standards To Meet Compelling and local or regional air pollution problems decision to withdraw the waiver would Extraordinary Conditions and not whether the specific standard warrant exercise of the Administrator’s (1) Introduction that is the subject of the waiver request judgment. is necessary to meet such conditions (73 Section 209(b)(1)(B) provides that no FR 12156; March 6, 2008). Additionally, Furthermore, that decision entails waiver of section 209(a) preemption will California typically would seek a waiver matters not only of policy judgment but be granted if the Administrator finds of particular aspects of its new motor of statutory interpretation, chief among that California does not need ‘‘such vehicle program up until the ACC which is the question of what is the standards to meet compelling and program waiver request. In the 2008 appropriate inquiry under section extraordinary conditions.’’ EPA is GHG waiver denial, which was a waiver 209(b)(1) when the Administrator is proposing to withdraw the grant of request for only GHG emissions faced with a request for a preemption waiver of preemption for CARB’s GHG standards, however, EPA determined waiver for standards designed to and ZEV standards for 2021 MY through that its prior interpretation of section address a global environmental 2025 MY based on a finding that 209(b)(1)(B) was not appropriate for problem. EPA has previously expressed California does not need these standards GHG standards because such standards the view that certain waiver requests to meet compelling and extraordinary are designed to address global air might call for the Administrator to conditions, as contemplated under pollution problems in contrast to local exercise judgment in determining section 209(b)(1)(B). As shown below, or regional air pollution problems California’s need for particular EPA is proposing to determine that the specific to and caused by conditions standards, under section 209(b)(1)(B). ACC program GHG and ZEV standards specific to California (73 FR 12156–60). Specifically, in the March 6, 2008 GHG are standards that would not In the 2008 denial, EPA further waiver denial, EPA posited that it was meaningfully address global air explained that its previous reviews of neither required nor appropriate for the pollution problems posed by GHG California’s waiver request under Agency to defer to California on the emissions in contrast to local or regional section 209(b)(1)(B) had usually been statutory interpretation of the Clean Air air pollution problem with causal ties to cursory and undisputed, as the Act, including the issue of the confines conditions in California. As also shown fundamental factors leading to or limits of state authority established below, EPA is proposing to find that California’s air pollution problems— by section 209(b)(1)(B), especially given while potential conditions related to geography, local climate conditions (like that EPA’s evaluation of California’s global climate change in California thermal inversions), significance of the request for a waiver to enforce GHG could be substantial, they are not motor vehicle population—had not standards would relate to the limits of sufficiently different from the potential changed over time and over different California’s authority to regulate GHG conditions in the nation as a whole to local and regional air pollutants. These emissions from new motor vehicles, justify separate state standards under fundamental factors applied similarly instead of particular regulatory CAA section 209(b)(1)(B). Moreover, the for all of California’s air pollution provisions that California was seeking to GHG and ZEV standards would not have problems that are local or regional in enforce. There, EPA construed section a meaningful impact on the potential nature. 209(b)(1)(B) as calling for either a conditions related to global climate In the 2008 denial, EPA noted that consideration of environmental change. EPA is thus proposing to find atmospheric concentrations of GHG are problems with causes that were specific that California does not need GHG substantially uniform across the globe, to California, or in the alternative, standards to meet compelling and based on their long atmospheric life and environmental effects specific to extraordinary conditions, as the resulting mixing in the atmosphere. California in comparison to the rest of contemplated under section Therefore, with regard to atmospheric the nation. EPA further explained that 209(b)(1)(B). Additionally, California GHG concentrations and their this interpretation called for its own does not need the ZEV requirements to environmental effects, the California- judgment because it necessitated a meet ‘‘compelling and extraordinary’’ specific causal factors that EPA had determination of whether elevated conditions in California given that it considered when reviewing previous concentrations of GHGs lie within the allows manufacturers to generate credits waiver applications under section

VerDate Sep<11>2014 23:42 Aug 23, 2018 Jkt 244001 PO 00000 Frm 00261 Fmt 4701 Sfmt 4702 E:\FR\FM\24AUP2.SGM 24AUP2 sradovich on DSK3GMQ082PROD with PROPOSALS2