Committee for a Better Arvin v. United States Environmental Protection Agency

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

No. 24-7270 COMMITTEE FOR A BETTER ARVIN, a nonprofit corporation; Agency No. MEDICAL ADVOCATES FOR EPA–R09–OAR– HEALTHY AIR, a nonprofit 2023–0477 corporation; NATIONAL PARKS CONSERVATION ASSOCIATION, Environmental a nonprofit corporation; SIERRA Protection Agency CLUB, a nonprofit corporation, OPINION Petitioners,

v.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY; MICHAEL REGAN, in his official capacity as Administrator of the U.S. Environmental Protection Agency; MARTHA GUZMAN ACEVES, Regional Administrator of Region 9 of U.S., in her official capacity as Regional Administrator for Region IX of the U.S. Environmental Protection Agency,

Respondents,

----------------------------------------

CALIFORNIA AIR RESOURCES BOARD; SAN JOAQUIN VALLEY UNIFIED AIR POLLUTION CONTROL DISTRICT,

Intervenors.

On Petition for Review of an Order of the Environmental Protection Agency

Argued and Submitted November 19, 2025 San Francisco, California

Filed August 5, 2026

Before: Sidney R. Thomas, Daniel A. Bress, and Salvador Mendoza, Jr., Circuit Judges.

Opinion by Judge Mendoza, Jr.; Dissent by Judge Bress COMM. FOR A BETTER ARVIN V. U.S. EPA 3

SUMMARY*

Clean Air Act

The panel remanded without vacatur in a case challenging a final rule issued by the United States Environmental Protection Agency concerning air quality in California’s San Joaquin Valley. The Clean Air Act delegates authority to the EPA to set national ambient air quality standards (“NAAQS”). A state is primarily responsible for maintaining air quality within its entire geographic area and must submit state implementation plans (“SIPs”) that specify how NAAQS will be met in each air quality control region of the state. The EPA designates regions as “attainment” or “nonattainment” based on whether they meet the NAAQS. Section 172(c)(9) of the Act requires that SIPs for nonattainment areas include contingency measures if the area fails to make reasonable further progress or fails to meet a NAAQS by the applicable attainment date. Petitioners contend that the contingency measures approved by the final rule fall short of the improvement required for the Valley’s air quality. The panel held that the EPA exceeded its authority when it grafted a feasibility exemption onto Section 172(c)(9). Applying Loper Bright Enters. v. Raimondo, 603 U.S. 369 (2024), the panel held that because the text of Section 172(c)(9) is ambiguous, it must use statutory

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.

interpretation to independently interpret the statute to determine its most faithful reading. Congress did not expressly provide that the EPA should consider feasibility in Section 172(c)(9). The panel held that the EPA erred by plucking a feasibility exemption out of thin air and reading it into Section 172(c)(9) when the language of the provision does not provide for one. The panel further held that the provision’s directive to design “specific” contingency measures that shall “take effect” did not implicitly grant the EPA discretion to consider feasibility. In addition, permitting a feasibility exemption for contingency measures undermines both the specific function of Section 172(c)(9) and the overarching purpose of the Act. The panel held that remand without vacatur was the appropriate remedy given the potential disruptive consequences of the Valley lacking contingency measures while the EPA reconsiders the rule. Dissenting, Judge Bress wrote that the majority’s interpretation of the Clean Air Act—that California’s contingency measures do not meet the requirements of Section 172(c)(9) because any consideration of feasibility is completely forbidden when it comes to meeting air quality benchmarks—reflects serious legal error, conflating a supposedly improper feasibility assessment with what are merely the statutory requirements for an emissions reduction measure to qualify as a contingency measure under Section 172(c)(9) of the Clean Air Act. COMM. FOR A BETTER ARVIN V. U.S. EPA 5

COUNSEL

Colin C. O'Brien (argued) and Tyler B. Szeto, Earthjustice, San Francisco, California; Brent J. Newell, Law Office of Brent J. Newell, Petaluma, California; for Petitioners. Sarah Izfar (argued), Senior Trial Counsel, Environment and Natural Resources Division; Robert N. Stander, Deputy Assistant Attorney General; Adam R.F. Gustafson, Acting Assistant Attorney General; United States Department of Justice, Washington, D.C.; Yasmin P. Ortiz and Jefferson Wehling, Attorneys, United States Environmental Protection Agency, Washington, D.C.; for Respondents. Corey M. Moffat (argued), Deputy Attorney General; Jonathan A. Wiener, Acting Supervising Deputy Attorney General; Rob Bonta, California Attorney General; Office of the California Attorney General, San Francisco, California; Isela G. Welz, Assistant District Counsel; Annette A. Ballatore, District Counsel; San Joaquin Valley Unified Air Pollution Control District, Fresno, California; for Intervenors. Brian S. Tomasovic, Assistant Chief Deputy Counsel; Kathryn Roberts, Principal Deputy District Counsel; Barbara Baird, Chief Deputy Counsel; Bayron T. Gilchrist, General Counsel; Office of the General Counsel, South Coast Air Quality Management District, Diamond Bar, California; Kathrine Pittard, District Counsel, Sacramento Metropolitan Air Quality Management District, Sacramento, California; Piero C. Dallarda, Best Best & Krieger LLP, Riverside, California; Ann M. Siprelle, Best Best & Krieger LLP, Sacramento, California; for Amici Curiae South Coast Air Quality Management District, Sacramento Metropolitan Air Quality Management District, Mojave Desert Air

Quality Management District, and California Air Pollution Control Officers Association. Robyn Wille, First Assistant Attorney General; Rylie Slaybaugh, Assistant Attorney General II; Laura T. Mehew, Senior Assistant Attorney General; Kyle Davenport, Senior Assistant Attorney General II; Shannon Stevenson, Solicitor General; Natural Resources and Environment Section, Colorado Office of the Attorney General, Denver, Colorado; for Amicus Curiae Colorado Department of Public Health and Environment.

OPINION

MENDOZA, JR., Circuit Judge:

The San Joaquin Valley’s unique bowl-shaped topography is both a cradle of abundance and home to over four million Californians. It also has “some of the worst air quality in the United States.” Comm. for a Better Arvin v. EPA, 786 F.3d 1169, 1173 (9th Cir. 2015). The Valley has repeatedly failed to meet the Clean Air Act’s air quality standards, a reality reflected by the numerous disputes about emissions regulation in that area that have reached this court. The Committee for a Better Arvin, Medical Advocates for Healthy Air, the National Parks Conservation Association, and the Sierra Club challenge a final rule issued by the United States Environmental Protection Agency, arguing that the contingency measures approved by the rule fall far short of the improvement required for the Valley’s air quality. We conclude that the EPA exceeded its authority COMM. FOR A BETTER ARVIN V. U.S. EPA 7

when it grafted a feasibility exemption onto Section 172(c)(9) of the Clean Air Act, and we remand. I. A. Enacted in 1963, the Clean Air Act (“Act”) was designed to promote air quality improvement through a scheme of “cooperative federalism.” MacClarence v. EPA, 596 F.3d 1123, 1125 (9th Cir. 2010). In the Act, Congress recognizes that “air pollution prevention . . . and air pollution control at its source is the primary responsibility of States and local governments,” but provides for “[f]ederal financial assistance and leadership.” 42 U.S.C. § 7401(a). The Act delegates authority to the EPA to set national ambient air quality standards (“NAAQS”) for certain air pollutants. 42 U.S.C. § 7409; Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 462 (2001). A state is primarily responsible for maintaining air quality within its “entire geographic area,” a responsibility that entails submitting state implementation plans (“SIPs”) that specify how NAAQS will be met in each air quality control region of the state. 42 U.S.C. § 7407(a). Within this framework, the EPA designates regions as “attainment” or “nonattainment” based on whether they meet the NAAQS. 42 U.S.C. § 7407(d). Nonattainment areas are also classified by the severity of nonattainment—fine particulate matter (“PM2.5”) nonattainment areas are initially classified as “moderate” and can be subsequently reclassified as “serious.” 40 C.F.R. § 51.1002.

Section 172(c)(9)1 of the Act also requires that SIPs for nonattainment areas include contingency measures: “specific measures to be undertaken if the area fails to make reasonable further progress,” or fails to meet a NAAQS by the applicable attainment date. 42 U.S.C. § 7502(c)(9). Discussing the Section 172(c)(9) contingency measure requirement, the house report accompanying the 1990 amendments to the Act states that “[s]uch measures are to take effect without further action by the State o[r] Administrator, and are to be adequate to compensate for any emission reduction shortfall.” H.R. Rep. No. 101-490, reprinted in Cong. Rsch. Serv., A Legislative History of the Clean Air Act Amendments of 1990, at 3248 (1993). In this regard, these contingency measures function as a “backup” in case existing measures fall short. Ass’n of Irritated Residents v. EPA, 10 F.4th 937, 947 (9th Cir. 2021). B. The challenged rule approves two SIP submissions designed to meet three contingency measure requirements for the Valley. Petitioners argue that the rule violates the Act by applying a new, lowered progress standard for contingency measures and reading a feasibility exemption into the statute. The EPA maintains that it has applied the best interpretation of the Act’s requirements for contingency measures. The San Joaquin Valley Unified Air Pollution Control District (“District”) is primarily responsible for developing plans to provide for the Valley’s attainment of NAAQS. It works with the California Air Resources Board (“CARB”) to develop those plans. In June 2023, CARB submitted two

1 Section 172(c)(9) of the Act is codified at 42 U.S.C. § 7502(c)(9). COMM. FOR A BETTER ARVIN V. U.S. EPA 9

SIPs to the EPA addressing PM2.5 and precursor pollutants, including nitrogen oxides (“NOx”), in the Valley. 2 They contained three contingency measures designed to collectively meet the contingency measure requirements for the 1997 annual, 2006 24-hour, and 2012 annual PM2.5 standards.3 One of these contingency measures would require stricter wood burning restrictions if triggered, the second would require more “fugitive dust” control measures in certain open areas in the Valley, and the third would require more vehicles in the area to undergo smog checks. Notably, the proposed rule required only “one year’s worth of progress.” This marked a departure from the EPA’s traditional requirement that contingency measures provide for “one year’s worth of reasonable further progress.” The proposed rule also recommended accepting “feasibility analyses” provided by CARB and the District to justify the emissions reductions for precursor pollutant NOx falling short of that lowered “progress” standard. It detailed CARB and the District’s analyses of “contingency measure opportunities,” including CARB’s arguments on technological limitations and time pressure. Accepting the professed infeasibility of additional contingency measures paved the way for the EPA to approve the SIPs, despite the fact that the contingency measures fell significantly short of

2 PM2.5 is fine particulate matter with a diameter of 2.5 micrometers or less. NOx is a precursor pollutant that causes the formation of PM2.5. 3 The EPA classified the San Joaquin Valley as a serious nonattainment area for the 1997 annual PM2.5 standard in 2015, as a serious nonattainment area for the 2006 24-hour PM2.5 standard in 2016, and as a serious nonattainment area for the 2012 annual PM2.5 standard in 2021.

required NOx emissions reductions, even under the lowered one year’s worth of progress standard. Environmental advocacy groups raised numerous concerns about the proposed rule in the subsequent notice- and-comment period. The groups characterized the rule as an attempt by the EPA to “completely abdicate its obligation to require meaningful contingency measures” by abandoning its longstanding interpretation of Section 172(c)(9) as requiring one year’s worth of reasonable further progress. They further criticized the rule’s inclusion of a feasibility demonstration which further lowers the progress required from contingency measures. On this point, they argued that such feasibility demonstrations contradict the “plain language” of the Act, which does not apply any feasibility standard to contingency measures. Despite these concerns, the EPA issued the final rule with the lowered progress standard and the feasibility demonstration intact. The rule acknowledged the EPA’s “long-standing approach” of requiring one year’s worth of reasonable further progress, but it argued that that standard was “unnecessarily conservative.” And it defended further lowering progress requirements for contingency measures in certain circumstances:

[A]s stated in our proposed rule and the EPA’s Draft Revised Contingency Measures Guidance, where a state is unable to identify contingency measures that would provide approximately [one year’s worth] of emission reductions, the state should provide a reasoned justification (referred to herein as an “infeasibility demonstration”) that explains and documents how it has evaluated COMM. FOR A BETTER ARVIN V. U.S. EPA 11

all existing and potential control measures relevant to the appropriate source categories and pollutants in the nonattainment area and has reached reasonable conclusions regarding whether such measures are feasible.

II. The contingency measures create sufficient emission reductions for direct PM2.5 emissions to meet the EPA’s longstanding one year’s worth of reasonable further progress standard. But NOx emission reductions fall short of both one year’s worth of reasonable further progress and the rule’s lowered progress standard. The final rule reasons that the approved contingency measures nevertheless satisfy Section 172(c)(9) because CARB and the District provided “reasoned justifications for not adopting additional contingency measures.” So this challenge turns on whether Section 172(c)(9) allows the EPA to exempt contingency measures from progress standards if the air agency makes a feasibility demonstration.4 We find that it does not. A. Under the general standard set by Section 706 of the Administrative Procedure Act, the court sets aside an agency action if it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with the law.” 5 U.S.C. § 706(2)(A); see also Sierra Club v. EPA, 671 F.3d 955, 961 (9th Cir. 2012) (applying this general standard because the

4 We neither endorse nor reject the EPA’s use of “one year’s worth of progress” instead of the longstanding one year’s worth of reasonable further progress standard because this petition rises and falls on feasibility.

Act does not specify a standard for SIP approvals). “Section 706 makes clear that agency interpretations of statutes . . . are not entitled to deference.” Loper Bright Enters. v. Raimondo, 603 U.S. 369, 392 (2024). Until recently, we would have deferred to the EPA’s reasonable interpretation of ambiguous statutory language, even if it was not the only permissible construction. See Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 (1984), overruled by, Loper Bright, 603 U.S. at 412. But the Supreme Court now instructs courts to use their “full interpretive toolkit” to find the “best meaning” of ambiguous statutes rather than deferring to agency interpretations. Loper Bright, 603 U.S. at 408–09. Because the text of Section 172(c)(9) is ambiguous,5 we must dust off our well-worn tools of statutory interpretation to independently interpret the statute and to determine its most faithful reading. B. The text of Section 172(c)(9) is silent on feasibility. To support the EPA’s contention that feasibility demonstrations are nonetheless allowed by the provision, the challenged rule emphasizes that “the statutory provisions applicable to other nonattainment area plan control measure requirements . . . allow air agencies to exclude certain control measures that are deemed unreasonable or infeasible.” In other words, the EPA’s position is that explicit feasibility allowances elsewhere in the Act suggest that it can consider feasibility in this instance. But the fact that other provisions in the Act

5 “A statute is ambiguous if it gives rise to more than one reasonable interpretation.” Woods v. Carey, 722 F.3d 1177, 1181 (9th Cir. 2013) (citation modified). COMM. FOR A BETTER ARVIN V. U.S. EPA 13

provide for infeasibility exemptions—and that Section 172(c)(9) does not—fatally undermines the EPA’s position. “Congress generally acts intentionally when it uses particular language in one section of a statute but omits it in another.” Dep’t of Homeland Sec. v. MacLean, 574 U.S. 383, 391 (2015). As the Supreme Court observed in an early Clean Air Act case, “[w]here Congress intended the Administrator to be concerned about economic and technological infeasibility, it expressly so provided.” Union Elec. Co. v. EPA, 427 U.S. 246, 257 n.5 (1976). And Congress has provided for feasibility considerations elsewhere in the Act. Section 112 directs the EPA to promulgate and enforce emissions standards to control certain hazardous air pollutants, but it provides that “if it is not feasible in the judgment of the Administrator to prescribe or enforce an emission standard for control of a hazardous air pollutant or pollutants, the Administrator may” promulgate other “design, equipment, work practice, or operational standard[s]” to meet the provision’s goals. 42 U.S.C. § 7412(h)(1) (emphasis added). Then Section 188(e) of the Act permits attainment date extensions for serious PM10 nonattainment areas if the state can show that “the plan for that area includes the most stringent measures that are included in the implementation plan of any State or are achieved in practice in any State, and can feasibly be implemented in the area.” 42 U.S.C. § 7513(e) (emphasis added). It also expressly permits the consideration of “the technological and economic feasibility of various control measures.” Id. (emphasis added). Section 202, which covers motor vehicle emissions, also makes several feasibility allowances. It provides that

regulations cannot go into effect until “it would be feasible to implement” them, requiring collaboration with the National Academy of Sciences to “conduct a comprehensive study and investigation of the technological feasibility of meeting” certain emissions standards required by the provision, and allowing the Administrator to adjust standards related to the “useful life” of vehicles and engines “if he determines that it is feasible for vehicles and engines subject to such standards to meet such standards for a longer useful life.” 42 U.S.C. § 7521(a)(5)(B), (c)(1), (j)(3) (emphases added). As in Department of Homeland Security v. MacLean, the interpretive canon that Congress acts deliberately by omitting language that is included elsewhere in a statute “applies with particular force here.” 574 U.S. at 392. Feasibility is expressly referenced numerous times throughout the Act and the term appears “in close proximity” to Section 172(c)(9). Id. Even elsewhere in Section 172, there is language providing that the Administrator can consider the “availability and feasibility of the pollution control measures” when classifying nonattainment areas and extending attainment dates. See 42 U.S.C. §§ 7502(a)(1)(A), (a)(2)(A) (emphasis added). But there is no such language in Section 172(c)(9). Congress did not expressly provide that the EPA should consider feasibility in Section 172(c)(9). It easily could have done so. Indeed, it has done so elsewhere in the Act. The EPA erred by plucking a feasibility exemption out of thin air and reading it into Section 172(c)(9) when the language of the provision does not provide for one. So the challenged rule is contrary to the Act’s best meaning and must be remanded to the EPA. COMM. FOR A BETTER ARVIN V. U.S. EPA 15

III. The EPA also argues that Congress granted it “discretionary authority to consider feasibility” by stating in Section 172(c)(9) that contingency measures are “specific measures” that will “take effect” without any qualifying language. But the hazy language of Section 172(c)(9) does not automatically give the EPA the right to consider feasibility. The Supreme Court has specifically rejected the presumption “that statutory ambiguities are implicit delegations to agencies.” Loper Bright, 603 U.S. at 399. So we must determine whether the best reading of the provision’s directive to design “specific” contingency measures that shall “take effect” implicitly grants the EPA discretion to consider feasibility. We find that it does not. A. On this point, Union Electric is instructive. There, the Supreme Court considered whether the “operator of a regulated emission source” could “raise the claim that it is economically or technologically infeasible to comply with the plan.” Union Elec., 427 U.S. at 249. The petitioner, an electric utility company, argued that a provision of the Act required consideration of technological and economic infeasibility because it provided that air quality standards be met “as expeditiously as practicable,” and within a “reasonable time.” Union Elec., 427 U.S. at 257–58. The petitioner asserted that “what is ‘practicable’ or ‘reasonable’ cannot be determined without assessing whether what is proposed is possible.” Id. at 258. The Supreme Court soundly rejected the petitioner’s argument, reasoning that the Act set strict compliance requirements that were “expressly designed to force regulated sources to develop pollution control devices that

might at the time appear to be economically or techno[logically] infeasible.” Id. at 257. The Court went as far as concluding that “Congress intended claims of economic and technological infeasibility to be wholly foreign to the Administrator’s consideration of a state implementation plan.” Id. at 256. The Court cited senators’ remarks during the legislative process, including Senator Edmund Muskie’s statement that Congress was responsible for protecting public health, which “may mean that people and industries will be asked to do what seems to be impossible at the present time.” Id. at 258–59 (citation modified). While this petition concerns a different provision of the Act, Union Electric’s rationale nevertheless applies.6 The contested statutory language in Union Electric required evaluation of what time frames would be “reasonable” or “practicable.” Even that was not enough to allow for feasibility considerations that were not explicitly provided for in the statute. In contrast, Section 172(c)(9) is devoid of references to practicability, and the EPA points only to “specific measures to be undertaken” as evidence that Congress intended to provide discretion to consider feasibility. In other words, if the phrases “as expeditiously as practicable,” and “within a reasonable time” do not permit consideration of feasibility, it is difficult to see how “specific

6 The dissent spills much ink arguing that Union Electric is distinguishable from this petition for review. This argument misses the point of our analysis. It was never our contention that Union Electric was factually or procedurally identical to this case. Union Electric is an early example of the Supreme Court refusing to “read in” technological and economic infeasibility considerations into the Act when Congress did not expressly include them. COMM. FOR A BETTER ARVIN V. U.S. EPA 17

measures to be undertaken,” a phrase even further disconnected from feasibility, would allow it. B. Department of Treasury, IRS v. FLRA is also instructive. 494 U.S. 922 (1990). In that case, the Supreme Court rejected the government’s argument that the term “applicable laws” in one provision of the Civil Service Reform Act of 1978 was equivalent to the phrase “any law, rule, or regulation,” which is used elsewhere in the statute. Id. at 931–33. The majority reasoned that a “statute that in one section refers to ‘law, rule or regulation,’ and in another section to only ‘laws’ cannot, unless we abandon all pretense at precise communication, be deemed to mean the same thing in both places.” Id. at 932. If we hold that “specific measures to be undertaken” means “feasibility,” a term that is used throughout the Act and which holds a particular meaning, then we will be choosing to “abandon all pretense at precise communication,” as the Supreme Court warned against in Department of Treasury. Reading a feasibility standard into a provision whose text plainly does not provide for one would undermine the clarity of the Act’s language and usurp Congress’s choices in favor of an agency’s preference. As discussed supra, Congress is perfectly capable of writing feasibility into statutory language; it has done so repeatedly throughout the Act. So we find that Congress did not implicitly grant the EPA discretion to consider feasibility with the language of Section 172(c)(9). IV. Turning to another tool in our statutory interpretation toolbox, “[i]t is a fundamental canon of statutory

construction that the words of a statute must be read in their context and with a view to their place in the overall statutory scheme.” West Virginia v. EPA, 597 U.S. 697, 721 (2022) (quoting Davis v. Mich. Dep’t of Treasury, 489 U.S. 803, 809 (1989)). “[W]e must interpret statutes as a whole, giving effect to each word and making every effort not to interpret a provision in a manner that renders other provisions of the same statute inconsistent, meaningless or superfluous.” Boise Cascade Corp. v. EPA, 942 F.2d 1427, 1432 (9th Cir. 1991). “By requiring contingency measures, the Act closes any potential gap in progress should a nonattainment area miss a milestone.” Ass’n of Irritated Residents, 10 F.4th at 942. The feasibility demonstrations approved by this rule effectively allow states to eliminate that statutory imperative on demand. Accepting such demonstrations is particularly harmful to the purpose of contingency measures because the EPA’s guidance anticipates that these demonstrations will allow for a substantial lowering of progress standards. The EPA has stated that “[w]ith sufficient justification, [contingency measures] representing a small amount of reductions may be reasonable.” Lowering or largely eliminating progress standards for Section 172(c)(9) contingency measures undermines their role in the Act’s overall statutory scheme. To maintain contingency measures’ function as a “plan B,” 7 they should

7 The dissent suggests that our use of the phrase “plan B” indicates a larger misunderstanding of contingency measures’ purpose. But whether we say “plan B” or some other term, contingency measures’ function as a backup should not be discounted. As we have explained, “the reason the statute requires contingency measures is to have a backup that can be put in place immediately in case already-implemented measures in a plan fail to achieve reasonable further progress. The agency may not avoid COMM. FOR A BETTER ARVIN V. U.S. EPA 19

be designed to compel a nonattainment area to meet a consistent standard of progress. 8 Reading feasibility demonstrations into Section 172(c)(9), which does not provide for them, undermines that function. V. We end our analysis with statutory purpose. It is a “familiar canon of statutory construction that remedial legislation should be construed broadly to effectuate its purposes.” Tcherepnin v. Knight, 389 U.S. 332, 336 (1967). “The overriding purpose of the Clean Air Act is to force the states to do their job in regulating air pollution effectively so as to achieve baseline air quality standards, the NAAQS.” Exxon Mobil Corp. v. EPA, 217 F.3d 1246, 1255 (9th Cir. 2000). As Congressman Bob Carr stated during the legislative debate of the 1990 amendments, “[u]ltimately, the purpose of the Clean Air Act is to improve the public’s health through a cleaner environment.” House Debate on the Clean Air Act Amendments of 1990 Conference Report, reprinted in Cong. Rsch. Serv., A Legislative History of the Clean Air Act Amendments of 1990, at 1342 (1993). Section 172 of the Act furthers those goals of effective regulation and public health by setting general requirements for nonattainment areas. See 42 U.S.C. § 7502. Those requirements include nonattainment plans that “require reasonable further progress,” and contingency measures that pick up the slack if those nonattainment plans fail to meet

the need for robust contingency measures by assuming that they will not be needed.” Ass’n of Irritated Residents v. EPA, 10 F.4th 937, 947 (9th Cir. 2021) (citation modified). 8 Again, we take no position as to whether that consistent standard should be one year’s worth of reasonable further progress or one year’s worth of progress.

the reasonable-further-progress standard. Id. at §§ (c)(2), (9). Contingency measures are required by the Act with the purpose of bringing nonattainment areas in line with PM2.5 NAAQS. Permitting a feasibility exemption for these contingency measures undermines both the specific function of Section 172(c)(9) and the overarching purpose of the Act. VI. This case is about a limited question of statutory interpretation: does the best meaning of Section 172(c)(9) include a feasibility exemption, even if the text does not expressly provide for one? The answer, based on our established canons of statutory interpretation, is no. The dissent extensively discusses policy concerns. It warns that this decision will lead to “regulatory freefall” and a “treacherous trail.” First, we will note that it is unlikely that this limited statutory interpretation decision will bring about an administrative law apocalypse. It is worth reiterating that the EPA, not the petitioners, is proposing a new path. Our decision not to allow a novel feasibility analysis maintains the status quo. The dissent suggests that we should discuss the “actual feasibility analyses,” and stresses that the EPA’s analysis is “173 single-spaced pages.” We do not discuss the depth of the feasibility analyses because that has no bearing on the best meaning of the statutory provision at issue. The thoroughness of the EPA’s individual feasibility analyses does not ex post facto justify reading a feasibility exemption into Section 172(c)(9). To consider that would be to improperly broaden the scope of our analysis. This is a statutory interpretation question. No more, no less. COMM. FOR A BETTER ARVIN V. U.S. EPA 21

Much of the dissent’s argument boils down to the idea that we should make Section 172(c)(9) mean what the EPA and air quality districts say it should mean—and if we don’t, catastrophe will ensue. But we will not warp our statutory interpretation to reach a predetermined conclusion. However sincerely held the dissent’s concerns may be, we cannot substitute policy goals for neutral statutory interpretation. It is the prerogative of Congress, not this court, to amend legislation based on policy concerns. In the post-Chevron era, we are tasked with employing our “full interpretive toolkit” to find the “best meaning” of ambiguous statutes, without deferring to agency interpretations. Loper Bright, 603 U.S. at 408–09. We have done so. VII. Having concluded that the challenged rule violates the Act, we must now determine the appropriate remedy. Petitioners request that this court “remand[] the EPA’s approval” of the contingency measures so that “California will need to revisit its decision[-]making and adopt meaningful control measures in compliance with the Clean Air Act.” They also seek an award of litigation costs, including attorney’s fees. They do not request vacatur. “The traditional remedy for erroneous administrative decisions is vacatur, but we will leave invalid agency action in place when equity demands that we do so.” Migrant Clinicians Network v. EPA, 88 F.4th 830, 847 (9th Cir. 2023) (citation modified). The Ninth Circuit employs a two-factor balancing test to determine whether an agency action should stay in effect on remand—weighing the severity of the agency’s errors against “the disruptive consequences of an interim change that may itself be changed.” Cal. Cmtys. Against Toxics v. EPA, 688 F.3d 989, 992 (9th Cir. 2012)

(quoting Allied–Signal, Inc. v. U.S. Nuclear Regul. Comm’n, 988 F.2d 146, 150–51 (D.C. Cir. 1993)); Migrant Clinicians Network, 88 F.4th at 848. Evaluating agency error includes considering whether “vacating a faulty rule could result in possible environmental harm.” Pollinator Stewardship Council v. EPA, 806 F.3d 520, 532 (9th Cir. 2015). Petitioners have not requested vacatur, and we find that remand alone is appropriate given the potential disruptive consequences of the Valley lacking contingency measures while the EPA reconsiders this rule. So we remand without vacatur. As for costs, the Act provides that “[i]n any judicial proceeding under this section, the court may award costs of litigation (including reasonable attorney and expert witness fees) whenever it determines that such award is appropriate.” 42 U.S.C. § 7607(f). We find that such an award is appropriate in this case. REMANDED. COMM. FOR A BETTER ARVIN V. U.S. EPA 23

BRESS, Circuit Judge, dissenting: The majority opinion holds that Section 172(c)(9) of the Clean Air Act, 42 U.S.C. § 7502(c)(9), requires States to adopt technologically and economically infeasible contingency measures for meeting air quality benchmarks. This alarming ruling is mistaken, sending States into a regulatory freefall in which they must either adopt infeasible emission control measures or else face severe, escalating sanctions. California and the federal government do not always see eye to eye, but they both agree the Clean Air Act was never meant to operate in this Kafkaesque way. In this case, California and the EPA exhaustively explained why further contingency measures beyond the ones California has adopted do not meet the requirements of Section 172(c)(9) because they could not take effect automatically and yield emissions reductions promptly— precisely the analysis that Section 172(c)(9) requires. The majority takes issue with none of these extensive findings. Nor does it point to other possible contingency measures that California and the EPA failed to consider. Instead, California’s contingency measures fail because any consideration of feasibility is completely forbidden when it comes to meeting these air quality benchmarks. That interpretation of the Clean Air Act reflects serious legal error, conflating a supposedly improper feasibility assessment with what are merely the statutory requirements for an emissions reduction measure to qualify as a contingency measure under Section 172(c)(9). Because the majority rests on a fundamental misapprehension of the statutory scheme and puts States in an untenable stranglehold, I must respectfully dissent.

I A Before turning to the facts of this case, I begin with an explanation of how Section 172(c)(9) contingency measures fit into the broader Clean Air Act apparatus, which is important in understanding where the majority goes wrong. To protect public health, the Clean Air Act instructs the EPA to establish national ambient air quality standards (NAAQS) for various pollutants. See EPA v. EME Homer City Generation, L.P., 572 U.S. 489, 498 (2014); see 42 U.S.C. § 7409(a); 40 C.F.R. § 50.7(a). If a region meets NAAQS for a given pollutant, the EPA classifies that region as “attainment.” 42 U.S.C. § 7407(d)(1)(A)(ii). If a region does not meet NAAQS, the EPA classifies it as “nonattainment” and sets a deadline by which the region must reach compliance. Id. §§ 7407(d)(1)(A)(i), 7502(a). Under the Act, “[t]here are several degrees of nonattainment, ranging from marginal to extreme, and each classification imposes increasingly stringent requirements to reduce emissions and promote progress toward attainment.” Ass’n of Irritated Residents v. EPA, 10 F.4th 937, 942 (9th Cir. 2021) (citations omitted). For fine particulate matter, the pollutant at issue here, nonattainment areas are initially classified as “moderate” and are reclassified as “serious” if the EPA finds that the area failed to attain the NAAQS by the moderate-area attainment date. 40 C.F.R. § 51.1002(a), (b)(2). When a region is designated “nonattainment” for a given pollutant, the state must create a State Implementation Plan (SIP), which the EPA must approve. See Bahr v. Regan, 6 F.4th 1059, 1065 (9th Cir. 2021); 42 U.S.C. §§ 7410, 7502(b). “Each SIP must include enforceable emission COMM. FOR A BETTER ARVIN V. U.S. EPA 25

limitations and other control measures necessary to attain the NAAQS, as well as timetables for compliance.” Bayview Hunters Point Cmty. Advocs. v. Metro. Transp. Comm’n, 366 F.3d 692, 695 (9th Cir. 2004), as amended on denial of reh’g and reh’g en banc (June 2, 2004). The SIP must “provide for the implementation of all reasonably available control measures as expeditiously as practicable,” and these provisions must “require reasonable further progress” in emissions reduction. 42 U.S.C. § 7502(c)(1), (2). The Act defines “reasonable further progress” as “such annual incremental reductions in emissions of the relevant air pollutant as are required . . . for the purpose of ensuring attainment of the applicable national ambient air quality standard by the applicable date.” Id. § 7501(1); accord 40 C.F.R. § 51.1000. In other words, reasonable further progress is the yearly amount of emission reductions necessary to stay on track to meet the NAAQS goal. See Irritated Residents, 10 F.4th at 941; Nat. Res. Def. Council, Inc. v. EPA, 638 F.3d 1183, 1191, 1194 (9th Cir. 2011) (explaining that reasonable further progress requires “generally linear progress”). A SIP does not always work out as planned, and Congress specified what should happen in that event. Specifically, this case is about the Clean Air Act’s requirement in Section 172(c)(9), 42 U.S.C. § 7502(c)(9), that a SIP include “contingency measures.” Contingency measures are additional emissions control measures that must take effect automatically if the primary control measures end up falling short of the SIP’s objectives. See Irritated Residents, 10 F.4th at 942; Bahr v. EPA, 836 F.3d 1218, 1235 (9th Cir. 2016). In full, Section 172(c)(9) provides that a SIP

shall provide for the implementation of specific measures to be undertaken if the area fails to make reasonable further progress, or to attain the national primary ambient air quality standard by the attainment date applicable under this part. Such measures shall be included in the plan revision as contingency measures to take effect in any such case without further action by the State or the [EPA] Administrator.

42 U.S.C. § 7502(c)(9). Under this provision, if a triggering event occurs—that is, if the State fails “to make reasonable further progress” or attain the NAAQS—the SIP’s contingency measures are activated and must “take effect,” “without further action by the State or the [EPA] Administrator.” Id. The majority describes contingency measures as a “plan B” to the State’s primary control measures, Majority Op. at 18–19, but that label is somewhat imprecise. When contingency measures take effect, they do not replace the State’s primary control measures, which remain in place and continue to provide emissions reductions. See Irritated Residents, 10 F.4th at 946 (explaining that contingency measures are additional measures beyond those “the State is already implementing in its plan”); Bahr, 836 F.3d at 1236 (same); 78 Fed. Reg. 53113, 53115 (Aug. 28, 2013) (noting that reductions from contingency measures are “in addition to those that were already scheduled to occur”). A State is not pivoting from its “plan A” to its “plan B” when it is required to activate its contingency measures. Rather, as the EPA has explained, contingency measures are “additional measures beyond or in addition to the required ‘core’ control COMM. FOR A BETTER ARVIN V. U.S. EPA 27

measures,” which are designed “to provide interim public health and welfare protection” when the State has started to fall behind. 59 Fed. Reg. 41998, 42015 (Aug. 16, 1994). The “interim” relief provided by contingency measures gives a State time to develop its actual “plan B,” which is an amended SIP. See Irritated Residents, 10 F.4th at 941–42 (noting that “[i]f the State does not meet a milestone . . . it must submit a plan revision”). The Clean Air Act requires that a SIP’s primary control measures provide reasonable further progress, 42 U.S.C. § 7502(c)(2), and if the EPA determines that a previously approved SIP is failing to comply with that requirement, the State must amend its SIP to rectify that deficiency, id. §§ 7410(k)(5); 7502(d); see also id. § 7513a(c)(3) (the EPA must require a plan revision within nine months if the relevant pollutant is fine particulate matter). So the purpose of Section 172(c)(9) contingency measures is not to provide a comprehensive “plan B”—that is what the SIP amendment process is for—but rather “to continue progress in reducing emissions while a state revises its SIP to meet the missed [reasonable further progress] requirement or to correct ongoing nonattainment.” 88 Fed. Reg. 87988, 87993 (Dec. 20, 2023). In other words, contingency measures are designed to be a splint, not the surgery. As I will explain, this misunderstanding of the purpose of contingency measures is one of the key flaws underlying the majority’s strained interpretation of Section 172(c)(9). B Two decisions from our court have clarified the role of contingency measures and made it more difficult for States to comply with Section 172(c)(9). First, in Bahr v. EPA, 836 F.3d 1218, 1235 (9th Cir. 2016), we held that contingency

measures cannot be measures the State has already implemented. There, the EPA approved Arizona’s proposed contingency measures even though all of them were already in effect when Arizona submitted its plan for approval. 836 F.3d at 1227–29. The EPA argued that “permitting early implementation of contingency measures is consistent with the overall policy of the [Clean Air Act] to reduce particulate emissions and protect public health.” Id. at 1236. We rejected this argument, holding that the EPA’s approval violated Section 172(c)(9), which is “clear that ‘contingency measures’ are control measures that will be implemented in the future.” Id. at 1235. “Control measures that have already been implemented are not measures ‘to be undertaken’ or ‘to take effect’ in the future, and the statute cannot reasonably be so interpreted.” Id. at 1236. Second, and five years later, we considered an attempt by California to work around our decision in Bahr. See Ass’n of Irritated Residents v. EPA, 10 F.4th 937, 942–43 (9th Cir. 2021). In response to Bahr, California proposed a single contingency measure for the San Joaquin Valley region which it had not already implemented, providing only meager reductions in emissions. Id. at 942. But California attributed a significantly higher reduction in emissions to the contingency measure, arguing that it could “combine emissions reductions achieved through contingency measures” with “surplus emissions reductions” from measures that were already in effect. Id. at 946 (quotation omitted). We rejected that maneuver as “a transparent effort to circumvent Bahr,” holding that a State could not justify meager contingency measures by counting “surplus” emissions reductions from measures that were already in effect. Id. at 942–43, 945–47. COMM. FOR A BETTER ARVIN V. U.S. EPA 29

In the wake of these two decisions, the EPA updated its guidance regarding how it would evaluate contingency measures. See 88 Fed. Reg. 17571, 17571 (Mar. 23, 2023) (notice of updated guidance). 1 The EPA explained that it was issuing updated guidance because “recent court decisions . . . have invalidated key aspects of EPA’s historical approach to implementing the [contingency measure] requirement.” 2023 Guidance at 2. The EPA had “received feedback” from States that Bahr and Irritated Residents had “increased the difficulty that they face in identifying measures sufficient to meet the [contingency measure] requirement.” 2023 Guidance at 2. In particular, States were now finding it difficult to meet the EPA’s recommended targets for emissions reductions from contingency measures, since they could no longer count surplus reductions from already-implemented measures. The EPA made two notable changes to address these concerns. First, the agency lowered its recommended target. Unlike other provisions in the Clean Air Act, Section 172(c)(9) does not set an explicit target for the quantity of emissions reductions that contingency measures should achieve. See 42 U.S.C. § 7502(c)(9); 57 Fed. Reg. 13498, 13511 (Apr. 16, 1992) (noting, shortly after the passage of Section 172(c)(9), that the provision does “not specify . . . the magnitude of emissions reductions that must be provided by these measures”). However, the EPA has historically recommended that contingency measures “should be

1 The EPA published the guidance document on its website. See U.S. Env’t Prot. Agency, DRAFT: Guidance on the Preparation of State Implementation Plan Provisions that Address the Nonattainment Area Contingency Measure Requirements for Ozone and Particulate Matter (Mar. 16, 2023), https://perma.cc/R432-6APJ [hereinafter 2023 Guidance].

approximately equal to the emissions reductions necessary to demonstrate [reasonable further progress] for one year.” See 57 Fed. Reg. at 13543–44. In its 2023 guidance, the EPA lowered its recommendation from one year’s worth of “reasonable further progress” to one year’s worth of “progress.” 2023 Guidance at 3, 21–23; see also 88 Fed. Reg. at 87994 (noting that under the EPA’s new guidance, “the recommended level of emissions reductions that contingency measures should achieve would represent one year’s worth of ‘progress’ as opposed to one year’s worth of [reasonable further progress]”). This new “progress” metric is calculated differently, such that it provides a more modest target than “reasonable further progress.” See 88 Fed. Reg. at 87994 (describing the new formula).2 The parties debate whether the EPA’s shift from “reasonable further progress” to “progress” is valid. But I agree with the majority opinion that this case does not turn on that issue and so we need not resolve it, as I discuss further below. See also Majority Op. at 11 & n.4. Second, and more relevant to this case, the EPA also gave States new guidance on what they should do if they could not identify contingency measures that would achieve the EPA’s recommended target. Since 2016, the EPA has allowed States to provide a “reasoned justification” for why they are unable to meet the EPA’s target. See 81 Fed. Reg.

2 “Progress” is calculated by (1) determining the average annual reductions between the base year emissions inventory and the projected attainment year emissions inventory; (2) determining what percentage of the base year emissions inventory this amount represents; and (3) applying that percentage to the projected attainment year emissions inventory. See 88 Fed. Reg. at 87994. COMM. FOR A BETTER ARVIN V. U.S. EPA 31

58010, 58067 (Aug. 24, 2016) (“In the rare event that an area is unable to identify contingency measures to account for approximately 1 year’s worth of emissions reductions, the state should provide a reasoned justification why the smaller amount of emissions reductions is appropriate.”). In 2023, the EPA built on this guidance and clarified that one type of “reasoned justification” is if “the area lacks sufficient feasible measures to achieve one year’s worth of reductions.” 88 Fed. Reg. at 87993–94; see also 2023 Guidance at 32. The EPA’s 2023 guidance extensively details what this feasibility analysis should look like. States must follow a four-step process: (1) identify all potential control measures that provide a non-trivial reduction in emissions and that are not already included in the SIP, (2) assess whether each identified measure is “feasible as a contingency measure” and adopt all that are, (3) explain why each rejected measure was deemed infeasible, and (4) document the process with “supporting information and analysis” in a reasoned- justification addendum for the EPA’s review. 2023 Guidance at 32–40. The EPA also gave States substantial guidance for how to assess both technological and economic feasibility. Measures might be technologically infeasible if they “would render the source completely inoperable, make the source clearly unsafe to operate, or create significant negative non- air quality related environmental impacts.” 2023 Guidance at 36. However, the EPA reminded States that the question was whether these measures were feasible as contingency measures, so the analysis must be tied to the specific requirements in Section 172(c)(9). 2023 Guidance at 31, 36.

For example, States must consider any “factors that might prevent a particular measure from taking effect within 60 days of triggering.” 2023 Guidance at 36. This is because Section 172(c)(9) requires that contingency measures “take effect” automatically when a triggering event occurs, “without further action by the State or the Administrator.” 42 U.S.C. § 7502(c)(9); see Irritated Residents, 10 F.4th at 942 (noting that contingency measures “take effect automatically ‘upon a failure by the State to meet the applicable milestone’” (quoting 42 U.S.C. § 7502(c)(9))); Bahr, 836 F.3d at 1235 (same). In practice, the EPA has formalized this requirement by telling States that it “expect[s] all actions needed to effect full implementation of the [contingency] measures to occur within 60 days after the EPA notifies the state of a [triggering event].” 88 Fed. Reg. at 87993. So if a proposed measure is not able to take effect automatically when triggered (i.e., within 60 days), it is not feasible as a contingency measure, since it cannot comply with this requirement in Section 172(c)(9). 2023 Guidance at 36–37. Along similar lines, contingency measures once implemented must be able to provide emissions reductions quickly. The EPA “expect[s] the additional emissions reductions from the contingency measures to be achieved within a year of the triggering event,” or two years in limited circumstances. 88 Fed. Reg. at 87993, 87994. So, when considering technological feasibility, States must also consider “the time frame for achieving reductions.” 2023 Guidance at 36. If a “measure requires longer than 2 years to achieve reductions,” it is likely not feasible as a contingency measure under the EPA’s guidance. 2023 Guidance at 36. COMM. FOR A BETTER ARVIN V. U.S. EPA 33

The EPA’s 2023 Guidance envisioned a more limited role for considerations of economic feasibility. The agency observed that “by definition, triggering [contingency measures] means the area has failed to attain or meet [reasonable further progress] through implementing the set of measures within the plan,” meaning States will likely “need to consider more costly measures.” 2023 Guidance at 37. However, the EPA noted that there was likely “some upper limit relevant to costs,” such “that at some point costs could become so exorbitant that it would render the otherwise possible [contingency measure] infeasible.” 2023 Guidance at 38. In particular, the EPA directed States to consider how much it would cost to implement the measure as quickly as its guidance requires. For both economic and technological feasibility, the EPA emphasized that just because a State had deemed a measure “infeasible” for purposes of other provisions in the Clean Air Act, that did not mean the measure was infeasible for purposes of Section 172(c)(9). As the EPA advised, States “should not simply repeat” prior infeasibility showings and should instead follow the EPA’s specific guidance for assessing the feasibility of contingency measures. 2023 Guidance at 31. C With this legal background in mind, I turn to the facts of this case, which is the latest episode in a long-running controversy regarding California’s proposed contingency measures for PM2.5 in the San Joaquin Valley. PM2.5 is fine particulate matter with a diameter of 2.5 micrometers or less. A precursor to PM2.5 is nitrogen oxides, or NOx. California has made significant strides in working to manage the air quality of the San Joaquin Valley, an important area

of the State that—due to its extensive agricultural operations, topography, and population—struggles to meet air quality standards. See 87 Fed. Reg. 31443, 31453 (May 24, 2022) (“EPA acknowledges California may have one of, if not the, most stringent emissions control strategy in the country.”). But as California tells us, and no one disputes, as the State has implemented stricter emissions controls over the years, fewer options are available to be held in reserve as Section 172(c)(9) contingency measures, particularly after our decisions in Bahr and Irritated Residents. On multiple occasions in the past, the EPA has disapproved of California’s proposed contingency measures for PM2.5. See, e.g., 76 Fed. Reg. 69896, 69897 (Nov. 9, 2011) (disapproving measures “for failing to provide sufficient emissions reductions”); 80 Fed. Reg. 49190, 49191 (Aug. 17, 2015) (proposing disapproval for reliance on non-SIP measures to provide contingency reductions); 86 Fed. Reg. 38652, 38669 (July 22, 2021) (proposing disapproval for “several deficiencies”). After the 2021 disapproval, Petitioners sued, seeking to compel the EPA to directly promulgate contingency measures for the San Joaquin Valley under its federal implementation authority. See 42 U.S.C. § 7410(c)(1) (giving the EPA the authority to promulgate a “Federal implementation plan” after “disapprov[ing] a State implementation plan submission in whole or in part”); Comite Progreso De Lamont v. EPA, No. 3:21-cv- 8733-WHA (N.D. Cal. Nov. 10, 2021), ECF No. 1 (Compl.). That suit resulted in a stipulated consent decree, pursuant to which the EPA agreed to prepare its own contingency measure plan for the region. See id., ECF No. 42 (Consent Decree). COMM. FOR A BETTER ARVIN V. U.S. EPA 35

Meanwhile, California continued working on its own revised contingency measures. On June 8, 2023, California submitted a revised SIP for approval, which included three proposed contingency measures. 88 Fed. Reg. at 87992, 87995. If triggered, the contingency measures would (1) further restrict residential wood burning in the Central Valley, (2) extend fugitive dust control requirements to more residential and commercial lots, and (3) subject more cars to smog check requirements. Id. at 87995–99. Whether the target is “reasonable further progress” or the new lower “progress” standard, these three contingency measures were projected to achieve the EPA’s recommended quantity of reductions for direct PM2.5 emissions, but not for NOx emissions. Id. at 88004, tbl. 3. In line with the EPA’s 2023 updated guidance, California submitted a detailed infeasibility demonstration with its 2023 contingency measures plan, explaining why it was unable to meet the NOx target. In this “reasoned justification” addendum, California considered a wide range of potential contingency measures and analyzed whether they were technologically and economically feasible. For example, California considered accelerating electrification requirements for residential water heaters and furnaces, but it decided that such a measure was infeasible as a contingency measure because it would yield only negligible emissions reductions within two years after being triggered. 88 Fed. Reg. at 88007. California also investigated whether to lower NOx limits for oilfield steam generators, but it similarly concluded that affected units would likely need to install selective catalytic reduction, which could not feasibly occur within the two-year contingency-measure timeframe. 88 Fed. Reg. at 88008. Ultimately, California concluded that there were limited solutions that could be implemented

immediately and that would achieve emissions reductions quickly enough to qualify as contingency measures under Section 172(c)(9). See 88 Fed. Reg. at 88001. A few months later, the EPA unveiled its own proposed contingency measures plan for the San Joaquin Valley. 88 Fed. Reg. 53431, 53431 (Aug. 8, 2023). The EPA proposed just two measures, which were effectively identical to the wood-burning and dust-control measures proposed by California. Compare 88 Fed. Reg. at 53445, with 89 Fed. Reg. 80749, 80752 (Oct. 4, 2023). The EPA acknowledged that the reductions these two measures would achieve were less than what it “normally recommended,” but it explained that, like California, the EPA was “unable to identify and adopt any additional feasible contingency measures that would achieve” the recommended target. 88 Fed. Reg. at 53445–46. The EPA also prepared its own “reasoned justification” analysis explaining in great detail, over the course of 173 single-spaced pages, why additional contingency measures, which the EPA rejected, were infeasible. For example, the EPA considered whether California could fashion new engine standards but concluded it would take too long to implement following a triggering event, meaning that these measures would not qualify as contingency measures. On December 20, 2023, the EPA proposed approving California’s proffered contingency measures. 88 Fed. Reg. at 87988. The EPA found that the measures satisfied Section 172(c)(9) because they were prospective and conditional, included triggering mechanisms, and were structured to be implemented in a timely manner without significant further action by California or the EPA. 88 Fed. Reg. at 87995–99. The EPA also found that California had provided a reasoned justification for not meeting the NOx target, by COMM. FOR A BETTER ARVIN V. U.S. EPA 37

demonstrating that additional contingency measures would be infeasible. 88 Fed. Reg. at 88005–88009. On October 4, 2024, the EPA finalized its approval of California’s plan, thereby mooting the (nearly identical) federal plan it had prepared. 89 Fed. Reg. at 80749. II The question for us is whether the EPA’s approval of California’s contingency measure plan violated Section 172(c)(9) of the Clean Air Act. In my view, the analysis here should be straightforward. As the EPA found—and no one disputes—California’s three proposed contingency measures are additional measures that can take effect automatically in the future, as soon as a triggering event occurs. In other words, they comply with our decisions in Bahr and Irritated Residents and with Section 172(c)(9)’s requirements. Nevertheless, the majority holds that the EPA should have rejected the contingency plan because the agency improperly “grafted a feasibility exemption onto Section 172(c)(9).” Majority Op. at 6–7. According to the majority opinion, the Clean Air Act prohibited California from considering whether a given contingency measure is feasible. That is mistaken. What the majority deems an improper feasibility analysis is just California and the EPA ensuring that the proposed contingency measures comply with the specific requirements of Section 172(c)(9). In reality, it is the majority, not California or the EPA, that is reading requirements into the statute that do not exist. In the process, the majority sends us down a treacherous trail, requiring States to adopt infeasible contingency measures— which no State can reasonably do.

A The crux of Petitioners’ argument is that because Section 172(c)(9) does not contain the word “feasibility,” States and the EPA are prohibited from considering it. The majority buys this argument. In the majority’s telling, “Section 172(c)(9) is silent on feasibility,” and therefore “the fact that other provisions in the Act provide for infeasibility exemptions . . . fatally undermines the EPA’s position.” Majority Op. at 12–13. But what do we really mean when we are talking about “feasibility” in this context? The majority does not spend any time discussing the actual feasibility analyses that it deems unlawful, deeming them to have “no bearing on the best meaning of the statutory provision at issue.” Majority Op. at 20. But this refusal to consider California and the EPA’s actual analyses betrays the majority’s elevation of labels over substance. Because when one examines the actual analyses at issue, it becomes clear that “feasibility” here is really just shorthand for “compliant with the specific requirements of Section 172(c)(9).” The EPA’s feasibility analysis is 173 single-spaced pages. The agency considered eight different categories of measures: fuel combustion, waste disposal, cleaning and surface coating, petroleum production and marketing, industrial processes, solvent evaporation, mobile sources, and miscellaneous processes. In each of those categories, it identified “potential control measures” which were “not already included in the plan,” as required by Bahr and Irritated Residents. Then, for each proposed measure, it considered whether that measure was “feasible as a contingency measure.” California’s feasibility analysis followed that same process. 88 Fed. Reg. at 88000–01. COMM. FOR A BETTER ARVIN V. U.S. EPA 39

For example, both California and the EPA considered a measure that would have mandated the electrification of residential water heaters and furnaces. 88 Fed. Reg. at 88007. However, they found that significant “lead time [was] necessary for manufacturers to design and produce electric units,” and that implementing the measure would require “collaboration with energy and building code regulators” to modify existing building standards. 88 Fed. Reg. at 88007. So California decided not to adopt this measure because it was infeasible as a contingency measure, i.e., it could not “take effect” automatically, “without further action by the State or the Administrator.” 42 U.S.C. § 7502(c)(9); see 88 Fed. Reg. at 88007. California and the EPA also considered requiring particulate matter controls on underfired commercial charbroilers. 88 Fed. Reg. at 88007–08. But California had concerns about whether these unproven devices were technologically feasible. And even if they were, the new equipment would require “specialized staff at restaurants” to operate them, and they also raised “fire safety certification concerns.” 88 Fed. Reg. at 88007. The EPA’s own feasibility analysis echoed these concerns. The EPA also noted that it had reached out to the two jurisdictions that have enacted a charbroiler requirement like this—the San Francisco Bay Area and New York City—and “each agency indicated that they have no known installations.” So California and the EPA rejected this measure because they were concerned whether it could “take effect” at all. 42 U.S.C. § 7502(c)(9); see 88 Fed. Reg. at 88007. Contrast these two rejected measures with the “smog check” measure that California did adopt, which would require more vehicles to be subject to mandatory smog

checks.3 That measure was found to be feasible because the California legislature had already amended the State’s Health & Safety Code to authorize CARB to unilaterally narrow smog check exemptions if it “finds that providing an exception for these vehicles will prohibit the state from meeting the requirements of . . . the federal Clean Air Act.” Cal. Health & Safety Code § 44011(a)(4)(B)(ii). Because the legislative work was finished and CARB’s authority was already on the books, no “further action by the State” was necessary for this measure to take effect. 42 U.S.C. § 7502(c)(9); see 88 Fed. Reg. at 87984–86. You can say these concerns are grounded in “feasibility” if you want—California and the EPA use that term. But that is just a helpful label for what is, in substance, a rigorous analysis of whether various proposed contingency measures comply with the specific requirements of Section 172(c)(9). In other words, California and the EPA are not considering whether these measures are “feasible” in the abstract— they’re considering whether the measures are feasible as contingency measures, given Section 172(c)(9)’s requirements. For this reason, the majority’s reliance on Department of Treasury, IRS v. FLRA, 494 U.S. 922 (1990), is misplaced. See Majority Op. at 17. There, the Supreme Court rejected the argument that two distinct terms in a statute “mean the same thing.” Id. at 932. Here, though, the EPA is not

3 Specifically, the California Air Resources Board (CARB) agreed to, “within 30 days of the effective date of an EPA determination that an applicable triggering event has occurred for San Joaquin Valley for the PM2.5 NAAQS . . . narrow the newer vehicle exemption from eight or fewer model years old to seven or fewer model years old throughout the San Joaquin Valley.” 88 Fed. Reg. at 87998. COMM. FOR A BETTER ARVIN V. U.S. EPA 41

arguing that “take effect” in Section 172(c)(9) is equivalent to “feasibility” as it is used elsewhere in the Act. To the contrary, a “feasibility” analysis in this context looks very different from what it looks like in other contexts—as the EPA emphasized in its 2023 guidance. 2023 Guidance at 31. A State’s primary control measures, for example, don’t need to take effect immediately. See 42 U.S.C. § 7502(c)(1) (requiring implementation “as expeditiously as practicable”). So even if a proposed measure is not feasible as a contingency measure, it may be perfectly feasible as a primary control measure. The EPA made this point in its own feasibility analysis, emphasizing that it was “only evaluating these potential controls with respect to contingency measures requirements,” and it did “not intend for this evaluation to apply to other nonattainment control measures requirements.” The EPA even “encourage[d] [California] to review the feasibility of” several measures it had ruled out as contingency measures “when considering control strategies to meet PM2.5 attainment planning requirements in the future.” In other words, these control measures were promising—they just couldn’t serve as contingency measures. All of this to say: the majority is wrong that the “text of Section 172(c)(9) is silent on feasibility.” Majority Op. at 12. The provision does not use the word “feasibility.” But the majority’s analysis—which turns entirely on this point— elevates the non-selection of one particular word over substance. This is far too simplistic for such an important matter. Section 172(c)(9) is crystal clear that contingency measures must be able to “take effect” when triggered “without further action by the State or the Administrator.” 42 U.S.C. § 7502(c)(9); see Bahr, 836 F.3d at 1236

(“[U]nless [measures] are ‘to be undertaken’ in the event of a contingency, they do not fit the definition of ‘contingency measures’ provided in § 7502(c)(9).”). When California and the EPA conducted their feasibility analyses, they were just making sure that the measures they proposed complied with Section 172(c)(9)’s specific requirements. This does not mean that California or the EPA can just invoke “feasibility” and have their analyses rubber-stamped. As described above, the EPA has given States detailed guidance on how to conduct infeasibility demonstrations in accordance with Section 172(c)(9). And in this case, the agency did not accept all of California’s feasibility justifications. See 88 Fed. Reg. at 87994, 88005–09. For example, California considered “equity considerations for low-income and environmental justice communities” as one factor in ruling out the electrified-water-heaters contingency measure, but the EPA said it did not necessarily endorse that specific consideration. Id. at 88007. If Petitioners believe a State’s infeasibility demonstration is insufficiently supported or not grounded in the specific requirements of Section 172(c)(9), they can bring an arbitrary and capriciousness challenge based on those specific shortcomings. Or if Petitioners believe California or the EPA failed to consider a specific contingency measure that would have been feasible, they could raise that issue, too. The majority says that feasibility determinations “effectively allow [S]tates to eliminate” Section 172(c)(9)’s “statutory imperative on demand.” Majority Op. at 18. But nothing could be further from the truth. As I have noted, if any aspect of the feasibility determination is infirm or unsupported, it may be challenged as arbitrary and capricious. But again, neither Petitioners nor the majority opinion purport to take issue with any specific aspect of COMM. FOR A BETTER ARVIN V. U.S. EPA 43

California or the EPA’s analysis. 4 Instead, the majority holds that anything that falls under the (undefined) label of “feasibility” is unlawful. That sweeping ruling threatens to prohibit States from considering whether proposed contingency measures comply with Section 172(c)(9)’s basic requirements, and it cannot be right. Properly understood, Section 172(c)(9) does not just allow California and the EPA to consider feasibility. It effectively requires them to consider it, within the confines of Section 172(c)(9)’s requirements—which is what California and the EPA did. This is not policy-driven interpretation, as the majority claims. Majority Op. at 21. It is simply a matter of reading what Congress has written, considered in context. Imagine California had proposed the electrified-water- heaters contingency measure, even though it requires “more than two years” of lead time for manufacturers, as well as revisions to California’s existing building regulations. 88 Fed. Reg. at 88006. Such a measure would not comply with the EPA’s guidance, which requires that contingency measures take effect within 60 days of a triggering event and achieve reductions within a year. See 88 Fed. Reg. at 87994. And such a measure would violate Section 172(c)(9)’s requirement that contingency measures “take effect . . . without further action by the State.” In other words, if California failed to consider feasibility, the EPA could—and

4 In a sentence in their reply brief and at oral argument (when asked), Petitioners mentioned a few measures they think California should have adopted as contingency measures, like the charbroiler measure discussed above. But Petitioners’ argument is not that the EPA’s specific reasons for rejecting these measures were arbitrary and capricious. Their argument, which the majority adopts in full, is that the EPA violated Section 172(c)(9) merely by considering feasibility at all.

should—reject such a measure as noncompliant with Section 172(c)(9). Or consider an even more extreme example, in which California proposes a contingency measure based on a promising futuristic NOx-capture technology that is in the nascent stages of development. Surely the EPA could reject that measure based on concerns about its technological feasibility on an appropriate timeline. The Clean Air Act does not require California and the EPA to ignore such an obvious flaw in a proposed contingency measure. The primary reason the majority concludes otherwise is that other provisions in the Clean Air Act use the word “feasibility,” which the majority takes to mean that considerations of feasibility are prohibited anywhere that word is not used. See Majority Op. at 13–14 (collecting examples) (citing Dep’t of Homeland Sec. v. MacLean, 574 U.S. 383, 391 (2015) (“Congress generally acts intentionally when it uses particular language in one section of a statute but omits it in another.”)). But “feasibility” is not a magic word in the Clean Air Act. Even Petitioners do not contend that Congress always has to use the word “feasibility” to signal that States can consider it. For example, Section 189 requires that States adopt “reasonably available control measures” (RACM) in Moderate nonattainment areas and the “best available control measures” (BACM) in Serious nonattainment areas. 42 U.S.C. § 7513a(a)(1)(C), (b)(1)(B). The EPA has interpreted “available” to permit consideration of feasibility for both RACM and BACM. See 57 Fed. Reg. at 13540–41; 59 Fed. Reg. at 42008–12; see also Latino Issues F. v. EPA, 558 F.3d 936, 946–47 (9th Cir. 2009) (finding that the EPA properly approved BACMs proposed for the San Joaquin COMM. FOR A BETTER ARVIN V. U.S. EPA 45

Valley because, among other reasons, the agency conducted “thorough” and “detailed analyses” “of the technological and economic feasibility of the control measures”). Petitioners do not contend that the EPA’s interpretation of these provisions is inaccurate or that the word “available” is insufficient to allow consideration of feasibility. The majority opinion does not address these provisions one way or the other. So when Congress refers to “available” control measures, that means States can consider feasibility, even to the point of ruling out measures for being too expensive. See 59 Fed. Reg. at 42010 (BACM should be “determined on a case-by-case basis, taking into account energy, environmental, and economic impacts and other costs”); Latino Issues F., 558 F.3d at 947 (approving State’s selection of BACM based on a “Cost Effectiveness Analysis”). But when Congress requires that contingency measures “take effect,” States cannot consider whether proposed measures could actually be implemented? And when Congress requires that those measures take effect “without further action by the State or the Administrator,” States cannot consider whether additional actions would be required to implement the measure? Ultimately, the majority’s inference from silence is weak support for the strong medicine it prescribes—particularly when that “silence” is not really silence at all. Section 172(c)(9) is clear that contingency measures must be able to spring into effect automatically when a triggering event occurs. See Irritated Residents, 10 F.4th at 942; Bahr, 836 F.3d at 1235; Regan, 6 F.4th at 1084. And that means States must be able to rule out measures that cannot do so, regardless of whether that inquiry is labeled “feasibility.”

B Puzzlingly, the majority also devotes an entire section of its opinion to Union Electric Co. v. EPA, 427 U.S. 246 (1976), a fifty-year-old case that none of the experienced parties in this case rely on. 5 According to the majority, Union Electric held that “the phrases ‘as expeditiously as practicable’ and ‘within a reasonable time’ do not permit consideration of feasibility.” Majority Op. at 16–17. From this the majority seems to infer that Union Electric directs a global rule of statutory construction, by which feasibility can never be considered unless Congress uses the word “feasibility.” As I will explain, Union Electric does not stand for that proposition, and that is not the only reason why Petitioners presumably chose not to place any weight on it. Union Electric was about whether a State could adopt control measures that were more stringent than federal law required, even if the EPA was concerned that the plan was infeasible. 427 U.S. at 261. After Missouri adopted an aggressive SIP with “emission limitations . . . effective at once,” a utility company with coal-fired plants subject to the new regulations petitioned for review of the EPA’s approval. Id. at 252–53. The company argued that it was not feasible for it to comply with Missouri’s plan and that the EPA should have disapproved the plan on that basis. Id. At that time, the Clean Air Act provided that the EPA “shall approve” a proposed plan as long as eight specified

5 The case is cited once by Petitioners in their opening brief, for the background proposition that Congress adopted the Clean Air Act as “a drastic remedy to what was perceived as a serious and otherwise uncheckable problem of air pollution.” Id. at 256. The case is also cited once by Petitioners in their reply brief, in a parenthetical noting that another case cited it. Neither party discussed the case in substance. COMM. FOR A BETTER ARVIN V. U.S. EPA 47

criteria were met.6 One of those criteria was that the SIP must “provide[] for the attainment of such primary standard as expeditiously as practicable but . . . in no case later than three years from the date of approval of such plan.” Id. at 250 n.1 (quoting 84 Stat. at 1680, § 110(a)(2)(A)(i)). The parties disagreed about whether that provision allowed the EPA to reject a plan because it was infeasible, in the sense that it provided for attainment faster than the EPA thought was “practicable.” The Court agreed with the EPA that it could not reject a SIP on that basis, holding that

if a State makes the legislative determination that it desires a particular air quality by a certain date and that it is willing to force technology to attain it—or lose a certain industry if attainment is not possible—such a determination is fully consistent with the structure and purpose of the Amendments, and § 110(a)(2)(B) provides no basis for the

6 Since Union Electric was decided in 1976, “[t]he Act has been significantly amended twice, in 1977 and 1990.” Resisting Env’t Destruction on Indigenous Lands v. EPA, 704 F.3d 743, 748 (9th Cir. 2012). The provision at issue in Union Electric was overhauled in 1990, and its successor provision is not at issue in this case. Compare Clean Air Amendments of 1970, Pub. L. No. 91-604, § 4(a), 84 Stat. 1676, 1680–81, with Clean Air Act Amendments of 1990, Pub. L. No. 101- 549, § 101(b)–(c), 104 Stat. 2399, 2404–09 (codified at 42 U.S.C. § 7410(a)(2), (k)). This is another reason to be cautious about extrapolating Union Electric’s holdings to this case. Cf. Gross v. FBL Fin. Servs., Inc., 557 U.S. 167, 174 (2009) (“When conducting statutory interpretation, we ‘must be careful not to apply rules applicable under one statute to a different statute without careful and critical examination.’” (quoting Federal Express Corp. v. Holowecki, 552 U.S. 389, 393 (2008))).

EPA Administrator to object to the determination on the ground of infeasibility.

Union Electric, 427 U.S. at 265. This holding was based on the particulars of the statutory scheme, which allowed States to “adopt emission standards stricter than the national standards.” Id. at 263–64; see also id. at 265 (grounding its holding in the fact that “the States may submit implementation plans more stringent than federal law requires and that the Administrator must approve such plans if they meet the minimum requirements” for a SIP). The Court’s route to its holding—based on agency deference and legislative history, id. at 256, 258–59—is not one it would likely take today. Cf. Loper Bright Enters. v. Raimondo, 603 U.S. 369, 400–01 (2024); Louisiana v. Callais, 146 S. Ct. 1131, 1146 (2026) (giving less weight to a decision that “was decided at a time when this Court often paid insufficient attention to the language of statutory provisions” and “followed this pattern” by “jump[ing] right to the [legislative history]” (quotation omitted)). But more importantly, the Court’s decision in Union Electric was not, as the majority suggests, a referendum on whether the word “practicable” in the statute permitted considerations of feasibility. To the contrary, the Court equated “practicable” with “feasible.” Id. at 260. And it went on to hold that, when it came to whether the statute’s “minimum standard” was met—i.e., that the State achieves attainment “as expeditiously as practicable”—the EPA was allowed to consider feasibility. Id. at 264 & n.13. As the Court explained:

Economic and technological factors may be relevant in determining whether the COMM. FOR A BETTER ARVIN V. U.S. EPA 49

minimum conditions are met. Thus, the Administrator may consider whether it is economically or technologically possible for the state plan to require more rapid progress than it does. If he determines that it is, he may reject the plan as not meeting the requirement that primary standards be achieved “as expeditiously as practicable” or as failing to provide for attaining secondary standards within “a reasonable time.”

Id. at 264–65 n.13. Indeed, the Court specifically clarified that “we do not hold that claims of infeasibility are never of relevance in the formulation of an implementation plan or that sources unable to comply with emission limitations must inevitably be shut down.” Id. at 266. So, Union Electric did not say that the phrases “as expeditiously as practicable” and “within a reasonable time” were, on their own, insufficient textual bases to permit considerations of feasibility. Contra Majority Op. at 16–17. To the contrary, the Court held that those phrases could allow the EPA to consider feasibility. See Union Electric, 427 U.S. at 264–65 n.13, 266. The Court’s concern was instead about who could consider feasibility and when under the particular provisions at issue. Namely, the EPA could consider feasibility when assessing whether a State met minimum standards, but the State could decide whether to disregard feasibility when electing to exceed those standards. Union Electric concerned a State exceeding minimum standards, which is not the context here. Union Electric certainly did not establish any kind of hard-and-fast rule about the particular language Congress has to use if it wants to permit considerations of feasibility, whether under

the Clean Air Act or otherwise. And if Union Electric did stand for that groundbreaking rule of statutory interpretation, Petitioners presumably would have trumpeted the case from the top of every almond tree in the San Joaquin Valley. But even if we were deciding this case in the 1970s, Union Electric would not be the case of choice. That honor would likely go to Bunker Hill Co. v. EPA, 572 F.2d 1286 (9th Cir. 1977), a case decided after Union Electric and one that the parties do discuss. Bunker Hill was a “minimum conditions” case (in Union Electric’s parlance), in which the EPA rejected Idaho’s plan due to a concern that it was not achieving attainment “as expeditiously as practicable.” Id. at 1291; see also id. at 1294 n.12 (citing Union Electric, 427 U.S. at 264–65 n.13). The EPA and Idaho disagreed on what was feasible, i.e., whether Idaho’s plan provided for attainment in a sufficiently expedited manner. See id. at 1290–91. Based on its own feasibility analysis, Idaho determined that a 72% reduction in emissions was the “maximum percentage feasible under currently available technology” for a certain regulated factory. Id. at 1290. The EPA disagreed, finding that an 82% reduction was feasible, and directing Idaho to follow that requirement instead. Id. at 1291. We sided with Idaho, reasoning that “the EPA cannot require a level of control technology that is technologically and economically infeasible.” Id. at 1293. The EPA was “not bound by Idaho’s determination of technological feasibility,” but under the Administrative Procedure Act, the agency’s contrary finding of feasibility had to be “‘based on a consideration of the relevant factors’ and not ‘a clear error of judgment.’” Id. at 1294 (quoting Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 415–16 (1971)). COMM. FOR A BETTER ARVIN V. U.S. EPA 51

In the feasibility context, that meant the “proposed technology must not be ‘at a level that is purely theoretical or experimental.’” Id. at 1294 (quoting Essex Chem. Corp. v. Ruckelshaus, 486 F.2d 427, 433–34 (D.C. Cir. 1973)). Because the record indicated that the measures the EPA was requiring were not technologically feasible, we remanded for the EPA to adjust its finding or supplement its analysis. Id. at 1294, 1304–05; see also id. at 1306 (explaining that “the Administrator could mandate a particular constant emission control when it was technologically and economically feasible”). Bunker Hill is also an old case employing somewhat outdated methods to interpret a statutory provision that is no longer in effect. Petitioners point out that “Bunker Hill did not address provisions applicable to particulate matter nonattainment areas or contingency measures.” But the same is true of Union Electric. To the extent the majority thinks we should be looking at cases like Union Electric to shed light on the question here, Bunker Hill is much more on point. California is not arguing that it wants to adopt infeasible measures in a potential effort to force industry out of business, like in Union Electric. See Union Electric, 427 U.S. at 265 (allowing a state to “make[] the legislative determination that it desires a particular air quality by a certain date and that it is willing to force technology to attain it or lose a certain industry if attainment is not possible”). Instead, this case is about whether the EPA should have required California to adopt infeasible measures. And for that question, as between Union Electric and Bunker Hill, we should look to Bunker Hill’s holding that the EPA “cannot require a level of control technology that is technologically and economically infeasible.” Id. at 1293.

C Finally, the majority maintains that its holding is consistent with the broader statutory scheme and purpose. It holds that California and the EPA’s position would “allow for a substantial lowering of progress standards” and undermine “contingency measures’ function as a diss Once again, the majority is incorrect, and its error again flows from a misunderstanding of the role of contingency measures in the Clean Air Act. As I explained at the outset, contingency measures are not a State’s true “plan B.” Rather, contingency measures are designed to be a “quick fix”—measures that can fill in immediately and stop the bleeding when a State is starting to fall behind. See Regan, 6 F.4th at 1084 (“A contingency measure’s sole purpose is to be implemented in the event the defined condition occurs.”). When contingency measures take effect, they do not replace a State’s primary control measures. See Irritated Residents, 10 F.4th at 946; Bahr, 836 F.3d at 1236. Nor do they relieve a State of its obligation to fix its primary control measures via the SIP amendment process. See Irritated Residents, 10 F.4th at 941–42; 42 U.S.C. §§ 7410(k)(5), 7502(d), 7513a(c)(3). Contingency measures are like first responders. They might not have all the resources of a hospital, but they can arrive quickly and keep the situation from getting worse. To serve that purpose, it is most important that contingency measures are nimble and reliable. That is why Section 172(c)(9) explicitly requires that contingency measures must be able to “take effect . . . without further action by the State or the Administrator.” A big emissions reduction is great, if feasible. But given the role of contingency measures, the magnitude of reductions they can COMM. FOR A BETTER ARVIN V. U.S. EPA 53

achieve is of secondary importance to the speed and reliability of their implementation. If they cannot achieve these latter requirements, they are not contingency measures at all, and the qualifying emissions reductions they produce will be: zero. The majority gets that prioritization backwards. Under its view, the EPA must doggedly enforce its recommended target for emissions reductions, even to the point of forcing States to adopt plainly infeasible contingency measures. That would have to include, for example, the proposed charbroiler contingency measure, which both California and the EPA rejected as infeasible. Recall that two air quality districts—the Bay Area and New York City—already purport to require the sort of control devices California was considering mandating in the San Joaquin Valley. But both districts admitted to the EPA that they could not identify a single instance where the devices had actually been installed. Perhaps in theory, those devices could provide substantial reductions in emissions. And one day, maybe they will. But this charbroiler requirement would not do as a contingency measure. California could say that it will require those devices, on a dime, when a triggering event occurs. But when implementation turns out to be slow or exceedingly costly, it may well be looking at no additional emissions reductions, at least in the near term. Such a contingency measure will have utterly failed to serve its purpose. D What happens if, as will surely come to pass as a result of the majority opinion, a State cannot realistically adopt infeasible contingency measures? As California points out, one option may be for it to hold emission reduction measures in reserve, not implementing

them in the original SIP in case they are needed later as contingency measures. It is hard to see how that will serve the broader goals of the Clean Air Act. Another option is that the State will simply be unable to comply with Section 172(c)(9), leading to escalating sanctions. The Clean Air Act imposes mandatory sanctions starting eighteen months after the EPA disapproves a SIP in whole or in part. 42 U.S.C. § 7509(a); 40 C.F.R. § 52.31(d); see also Ass’n of Irritated Residents v. EPA, 686 F.3d 668, 671 (9th Cir. 2012) (noting that “EPA disapproval of a submitted plan” starts a “sanctions clock”). Offset sanctions come first, setting a punitive 2-to-1 emission offset ratio for any new or modified major stationary source. 42 U.S.C. § 7509(b)(2); 40 C.F.R. § 52.31(d)(1). That means companies must find and permanently retire two tons of existing pollution for every one ton they intend to create, making it very difficult to economically justify new businesses and facility expansions. See Com. of Va. v. Browner, 80 F.3d 869, 874, 882 (4th Cir. 1996) (explaining how offset sanctions operate). If the deficiency remains uncorrected for another six months, the region loses federal highway funds. 42 U.S.C. § 7509(b)(1). These sanctions start out area-specific, but after two years they can apply statewide. 42 U.S.C. § 7410(m); 40 C.F.R. § 52.30(c). Amicus briefs filed in this case underscore the magnitude of the majority’s erroneous decision. For it is not just the San Joaquin Valley that relies on feasibility considerations in drafting contingency measure plans. Three other air quality districts representing large and economically critical portions of California—Los Angeles, Sacramento, San Bernardino, and Riverside—tell us that, in accordance with the EPA’s guidance, they “have already prepared or submitted plans . . . based on demonstrations that additional COMM. FOR A BETTER ARVIN V. U.S. EPA 55

measures would be infeasible.” South Coast Air Quality Management District Amicus Br. 3. The South Coast Air Quality Management District, which represents Los Angeles and surrounding counties, gives examples of the sorts of infeasible measures that it would need to adopt to meet the EPA’s (lower) target of 7.92 tons per day of NOx—e.g., eliminating all NOx emissions from manufacturing and industrial sources (7.59 tons per day), eliminating half of emissions from residential fuel consumption (14.85 tons per day), or eliminating 90% of emissions from passenger automobiles (9.14 tons per day). South Coast Air Quality Management District Amicus Br. 7–8. As these amici tell us, the rule the majority adopts today will “be economically devastating” and “create significant risks to public health.” Id. at 8. An amicus brief from the Colorado Department of Public Health and Environment similarly alerts us that in Colorado, “the low-hanging fruit . . . have all been picked,” and that disallowing considerations of feasibility would leave it “in a no-win situation, forced to either adopt infeasible rules or face sanctions.” Colorado Amicus Br. 3. Section 172(c)(9) does not compel this bind. Contingency measures play an important, but focused, role in the Clean Air Act scheme. By disregarding that role, the majority defeats the very ends it purports to serve. The Clean Air Act is a statute that promotes “cooperative federalism.” MacClarence v. EPA, 596 F.3d 1123, 1125 (9th Cir. 2010); see Ohio v. EPA, 603 U.S. 279, 286 (2024) (“The Clean Air Act envisions States and the federal government working together to improve air quality.”). Not contemplated in that model is that courts would uncooperatively introduce new, severe limitations that the statute does not contain. In this case, California and the EPA did what that system is supposed to encourage: they worked

through the available options, rejected measures that could not take effect automatically, and adopted measures that could. Without finding any aspect of the EPA’s analysis unsupported, the majority’s decision to nevertheless override California and the EPA’s work will leave the San Joaquin Valley no better than it was before. And I am not optimistic that the adoption of infeasible contingency measures or the imposition of sanctions will leave the Valley better in the long run, either. The Clean Air Act did not require California to single-mindedly pursue cleaner air at the expense of every other possible consideration, including economic ones. See 42 U.S.C. § 7401(c) (directing that a “primary goal” of the Clean Air Act is to “promote reasonable Federal, State, and local governmental actions . . . for pollution prevention”) (emphasis added). This case did not need to be difficult. Section 172(c)(9) is a short provision with just a few specific requirements. No one disputes that those requirements are met. In fact, ironically, rigorous compliance with those requirements is what spelled California and the EPA’s demise. Because the statute requires what the majority forbids, I respectfully dissent.