`United States Court of Appeals
`FOR THE DISTRICT OF COLUMBIA CIRCUIT
`
`
`
`Argued May 7, 2020
`
`
`Decided July 14, 2020
`
`No. 19-1231
`
`STATE OF NEW YORK, ET AL.,
`PETITIONERS
`
`v.
`
`ENVIRONMENTAL PROTECTION AGENCY AND ANDREW
`WHEELER, IN HIS OFFICIAL CAPACITY AS ADMINISTRATOR OF
`THE U.S. ENVIRONMENTAL PROTECTION AGENCY,
`RESPONDENTS
`
`ADIRONDACK COUNCIL, ET AL.,
`INTERVENORS
`
`
`On Petition for Review of a Final Action of the
`United States Environmental Protection Agency
`
`
`Steven C. Wu, Deputy Solicitor General, Office of the
`Attorney General for the State of New York, argued the cause
`for petitioners. With him on the briefs were Letitia James,
`Attorney General for the State of New York, Barbara D.
`Underwood, Solicitor General, Morgan A. Costello and
`Claiborne E. Walthall, Assistant Attorneys General, Gurbir S.
`Grewal, Attorney General for the State of New Jersey, Lisa
`Morelli, Deputy Attorney General, and Christopher G. King,
`Senior Counsel, New York City Law Department.
`
`
`
`
`
`
`2
`
`Joshua A. Berman argued the cause for petitioners-
`intervenors Sierra Club, et al. With him on the briefs were Sean
`H. Donahue, Graham G. McCahan, Vickie L. Patton, and
`Liana James.
`
`
`Samara M. Spence, Attorney, U.S. Department of Justice,
`argued the cause for respondent. With her on the brief were
`Jonathan Brightbill, Principal Deputy Assistant Attorney
`General, and Abirami Vijayan and Stephanie L. Hogan,
`Counsel, U.S. Environmental Protection Agency. Sarah A.
`Buckley, Attorney, U.S. Department of Justice, entered an
`appearance.
`
`David M. Flannery, Kathy G. Beckett, Edward L. Kropp,
`Samuel B. Boxerman, Samina M. Bharmal, David M.
`Friedland, Laura K. McAfee, E. Carter Chandler Clements,
`Norman W. Fichthorn, Steven P. Lehotsky, and Michael B.
`Schon were on the brief for respondents-intervenors Midwest
`Ozone Group, et al. Laura M. Goldfarb, Amy M. Smith and
`Peter Tolsdorf entered appearances.
`
`
`Joseph A. Newberg II and Mary Ann Lee were on the brief
`for amicus curiae Commonwealth of Kentucky, Energy and
`Environment Cabinet in support of respondents.
`
`Before: SRINIVASAN, Chief Judge, and GRIFFITH and
`MILLETT, Circuit Judges.
`
`Opinion for the Court filed by Circuit Judge MILLETT.
`
`Concurring opinion filed by Circuit Judge GRIFFITH.
`
`MILLETT, Circuit Judge: Air pollutants do not stay still.
`Nor do they respect state borders. That has created a “complex
`problem”—namely that “air pollution emitted in one State[]
`
`
`
`
`
`3
`
`[can] caus[e] harm in other States.” EPA v. EME Homer City
`Generation, L.P., 572 U.S. 489, 495 (2014).
`
`This case involves a challenge to the Environmental
`Protection Agency’s asserted failure to address cross-border
`pollution under the Clean Air Act’s Good Neighbor Provision,
`42 U.S.C. § 7410(a)(2)(D)(i). The State of New York
`petitioned the EPA to find that power-generating and other
`facilities in nine different States were violating the Good
`Neighbor Provision by producing emissions that contributed
`significantly to New York’s difficulty attaining or maintaining
`compliance with the 2008 and 2015 National Ambient Air
`Quality Standards for ozone.
`
`The EPA denied New York’s petition on the ground that it
`failed to meet the agency’s standard for establishing a violation
`of the Good Neighbor Provision and, in particular, for
`demonstrating that cost-effective controls could be imposed on
`the pollution sources. The State of New York, the State of New
`Jersey, and the City of New York petitioned this court for
`review.
`
`We grant the petition for review. The EPA offered
`insufficient reasoning for the convoluted and seemingly
`unworkable showing it demanded of New York’s petition. In
`addition, the EPA’s finding that New York did not have an air
`quality problem under the 2008 National Ambient Air Quality
`Standards for ozone relied on two faulty interpretations of the
`Clean Air Act that have since been invalidated. See
`Maryland v. EPA, No. 18-1285, slip op. at 25–34 (D.C. Cir.
`May 19, 2020). For those reasons, we vacate the EPA’s
`decision and remand for further proceedings not inconsistent
`with this opinion.
`
`
`
`
`
`4
`
`I
`
`A
`
`The Clean Air Act, 42 U.S.C. §§ 7401 et seq., directs the
`EPA to establish and periodically revise National Ambient Air
`Quality Standards, or NAAQS, that set the maximum allowable
`concentrations for various air pollutants, including ozone. 42
`U.S.C. §§ 7408(a), 7409. To measure compliance with the
`NAAQS, the EPA, “in coordination with state governments,
`divides the country geographically into ‘air quality control
`regions.’” Natural Res. Defense Council v. EPA, 777 F.3d 456,
`458 (D.C. Cir. 2014) (formatting modified) (quoting 42 U.S.C.
`§ 7407). While some air quality control regions “lie within a
`single state[,] * * * others encompass portions of two or more
`states.” Maryland, slip op. at 6 (quoting Delaware Dep’t of
`Natural Res. & Environmental Control v. EPA, 895 F.3d 90,
`94 (D.C. Cir. 2018)).
`
`Once new air quality standards go into effect, each State
`must develop an implementation plan to ensure the standards
`are met within the State’s air quality control region. See 42
`U.S.C. § 7410(a)(1); see also id. § 7407(b)–(e). In addition,
`those plans must prohibit “any source or * * * emissions
`activity within the State from emitting any air pollutant in
`amounts which will * * * contribute significantly
`to
`nonattainment in, or interfere with maintenance by, any other
`State with respect to” the NAAQS. Id. § 7410(a)(2)(D)(i).
`That subpart is known as the “Good Neighbor Provision.” See
`Wisconsin v. EPA, 938 F.3d 303, 309–319 (D.C. Cir. 2019).
`
`Under Section 110 of the Clean Air Act, the EPA must
`review each State’s implementation plan and ensure its
`compliance with statutory requirements, including the Good
`Neighbor provision. See 42 U.S.C. § 7410(k)(1)–(4). If a State
`fails to timely correct a deficiency in its plan, then the EPA will
`
`
`
`
`
`5
`
`promulgate a federal implementation plan for the relevant
`region(s). Id. § 7410(c)(1).
`
`Section 126(b) of the Clean Air Act, 42 U.S.C. § 7426(b),
`creates an additional mechanism for enforcing the Good
`Neighbor Provision. It authorizes affected States or local
`subdivisions to petition the EPA to make a “finding that any
`major source or group of stationary sources emits or would
`emit any air pollutant in violation of the prohibition of [the
`Good Neighbor Provision.]” Id.1
`
`Under Section 126(b), the EPA must generally respond to
`the petition “[w]ithin 60 days after receipt of [such] petition
`* * * and after public hearing[.]” 42 U.S.C. § 7426(b). The
`agency may, however, grant itself an extension of up to six
`months “upon a determination that such extension is necessary
`to afford the public, and the agency, adequate opportunity to
`Id.
`carry out
`the purposes of
`th[e]
`subsection.”
`
`§ 7607(d)(1)(N), (d)(10).
`
`If an existing pollution source in another jurisdiction is
`found to be in violation of the Good Neighbor Provision, that
`source generally must cease operation within three months. 42
`U.S.C. § 7426(c). But the EPA may allow continued operation
`if the “source complies with such emission limitations and
`compliance schedules * * * as may be provided by the
`Administrator to bring about compliance * * * as expeditiously
`
`
`1 Section 126(b) cross-references Section 110(a)(2)(D)(ii) of the
`Clean Air Act, 42 U.S.C. § 7410(a)(2)(D)(ii). But that is understood
`to be a scrivener’s error. Appalachian Power Co. v. EPA, 249 F.3d
`1032, 1040–1044 (D.C. Cir. 2001). For present purposes, the proper
`cross-reference is
`the Good Neighbor Provision, 42 U.S.C.
`§ 7410(a)(2)(D)(i). See Appalachian Power, 249 F.3d at 1040–1044.
`
`
`
`
`
`6
`
`as practicable, but in no case later than three years after the date
`of such finding.” Id.
`
`B
`
`Over time, the EPA has promulgated increasingly
`stringent ozone standards.2 As relevant here, in 2008, the EPA
`lowered the acceptable ozone level, measured over eight hours,
`from 80 parts per billion to 75 parts per billion. 40 C.F.R.
`§ 50.15. And in 2015, it promulgated an even more restrictive
`ozone standard of 70 parts per billion. Id. § 50.19. Both the
`2008 NAAQS and the more stringent 2015 NAAQS remain in
`effect with differing deadlines for compliance.
`
`Depending on the degree of nonattainment, the Clean Air
`Act provides a deadline by which each air quality control
`region must achieve compliance. See 42 U.S.C. § 7511(a)(1),
`(b)(1). The more severe the noncompliance, the more time the
`region has to remedy the problem. See id. If the region fails to
`meet the compliance deadline, the EPA will reclassify it to a
`higher severity level. See id. § 7511(b)(2). That, in turn,
`automatically extends the deadline for compliance to the
`attainment date for that higher level.
`
`As relevant here, areas in “serious” nonattainment of the
`2008 NAAQS have a statutory attainment deadline of 2021.
`See Determination of Attainment and Reclassification for 2008
`Ozone NAAQS, 84 Fed. Reg. 44,238, 44,244 (Aug. 23, 2019).
`Areas in “moderate” nonattainment of the 2015 NAAQS have
`
`
`2 See, e.g., 44 Fed. Reg. 8202, 8217 (Feb. 8, 1979) (setting the
`primary ozone standard at 120 parts per billion); 62 Fed. Reg.
`38,856, 38,885 (July 18, 1997) (at 80 parts per billion); 73 Fed. Reg.
`16,436, 16,483 (March 27, 2008) (at 75 parts per billion); 80 Fed.
`Reg. 65,292, 65,362 (Oct. 26, 2015) (at 70 parts per billion).
`
`
`
`
`
`7
`
`a 2024 deadline for compliance. Response to Section 126(b)
`Petition from New York, 84 Fed. Reg. 56,058, 56,072 n.48
`(Oct. 18, 2019).
`
`C
`
`The New York-Northern New Jersey-Long Island, New
`York-New Jersey-Connecticut Area (“New York Metropolitan
`Area” or “Area”) is a multistate air quality control region. It is
`currently in “serious” nonattainment of the 2008 ozone
`NAAQS, having twice failed to meet previously applicable
`statutory deadlines for attainment. See Determination of
`Attainment and Reclassification for 2008 Ozone NAAQS, 84
`Fed. Reg. at 44,238, 44,244 (reclassifying seven areas,
`including the New York Metropolitan Area, to serious
`nonattainment); see also
`id. at 44,243
`tbl.2.
` The
`reclassification to serious nonattainment triggered a July 2021
`attainment deadline. Id. at 44,244.
`
`The Area is also in “moderate” nonattainment of the 2015
`NAAQS, with a 2024 deadline for attainment. Additional Air
`Quality Designations for 2015 Ozone NAAQS, 83 Fed. Reg.
`25,776, 25,821 (June 4, 2018); see 84 Fed. Reg. at 56,072 n.48.
`
`In March 2018, New York filed a Section 126(b) petition
`(“Petition”) that asked the EPA to find that approximately 350
`sources of nitrogen oxides in nine States were contributing
`significantly to nonattainment in the New York Metropolitan
`Area under the 2008 and 2015 NAAQS. J.A. 58, 60, 76.3 The
`
`
`3 The Petition also alleged that these out-of-state sources were
`interfering with attainment in Chautauqua County, New York.
`J.A. 60.
` New York’s and
`the
`Intervenor Environmental
`Associations’ arguments before this court focus exclusively on the
`
`
`
`
`
`8
`
`Petition pointed to Illinois, Indiana, Kentucky, Maryland,
`Michigan, Ohio, Pennsylvania, Virginia, and West Virginia as
`the sources of infiltrating ozone pollution. J.A. 60. New
`York’s modeling projected that the nine States would
`contribute at least one percent of the 2008 NAAQS (that is, at
`least 0.75 parts per billion) to at least one nonattaining ozone
`monitor in the New York Metropolitan Area. J.A. 60, 69.
`Within those nine States, the Petition focused the need for
`regulation on facilities that emit at least 400 tons of nitrogen
`oxides per year. See J.A. 60, 68, 76.4
`
`Rather than resolving the Petition within the 60-day
`statutory deadline, 42 U.S.C. § 7426(b), the EPA granted itself
`a six-month extension of time. Extension of Deadline, 83 Fed.
`Reg. 21,909, 21,910–21,912 (May 11, 2018); see also 42
`U.S.C. § 7607(d)(1)(N), (d)(10) (authorizing the EPA to grant
`itself an extension under certain circumstances).
`
`When the EPA missed that extended deadline, New York
`filed suit to compel a decision. The United States District
`Court for the Southern District of New York ordered the EPA
`to grant or deny the Petition by September 2019. See New York
`
`
`New York Metropolitan Area. So we do not address the EPA’s
`findings with respect to Chautauqua County.
`
`4 New Jersey, like the nine listed States, was projected to
`contribute at least 0.75 parts per billion to ozone levels in the New
`York Metropolitan Area. J.A. 69. The Petition nevertheless did not
`list New Jersey as a potential violator because New York’s modeling
`indicated that the 400-ton-per-year sources in New Jersey did “not
`significantly contribute to any nonattainment or maintenance
`monitors.” J.A. 71.
`
`
`
`
`
`9
`
`v. Wheeler, No. 19-CV-3287, 2019 WL 3337996, at *2
`(S.D.N.Y. July 25, 2019).
`
`After undertaking notice and comment procedures and
`conducting a public hearing, the EPA issued a final decision
`denying the Petition on September 20, 2019. 84 Fed. Reg. at
`56,093. The decision was published in the Federal Register the
`next month. Id.
`
`In evaluating the Petition, the EPA applied a four-step
`framework derived from prior rulemakings on the interstate
`transport of ozone. 84 Fed. Reg. at 56,058, 56,062–56,063.
`Those steps are: (1) identifying downwind areas that have
`trouble attaining or maintaining the NAAQS; (2) determining
`which upwind States’ emissions are “linked” to downwind air
`quality problems; (3) ascertaining which of those linked States’
`upwind sources “significantly contribute” to nonattainment or
`interfere with maintenance of the NAAQS in a downwind area;
`and (4) implementing emission reductions/budgets within the
`upwind States. Id. at 56,062.
`
`The EPA imposed the burden of satisfying each of those
`steps on New York as the Section 126(b) petitioner. See 84
`Fed. Reg. at 56,069–56,070. The EPA also construed
`Section 126(b) as allowing States to challenge interstate
`transport of pollution only when it impacted downwind
`receptors “within their geographical borders,” even if the
`upwind pollutants impede attainment in the air quality region
`of which the State is a part. Id. at 56,080; see also id. at 56,081
`& n.70.
`
`With respect to Step 1 of the four-part framework, the EPA
`found an air quality problem in the New York Metropolitan
`Area under the 2015 NAAQS. 84 Fed. Reg. at 56,080–56,081.
`But it found no such attainment problem under the 2008
`NAAQS. Id. The EPA reached that conclusion by treating
`
`
`
`
`
`10
`
`2023 as the relevant year for evaluating the existence of an air
`quality problem under the 2008 NAAQS. Id.; see also id. at
`56,074 (“The EPA disagrees that it is inappropriate to rely on
`the 2023 modeling because it does not align with a particular
`attainment date.”). On that basis, the EPA denied the portion
`of the Petition seeking to enforce the 2008 NAAQS. Rather
`than project air quality in 2021—the year by which attainment
`was legally required—the EPA found that “New York has not
`demonstrated that there will be a nonattainment or maintenance
`problem” in 2023. The EPA’s own analysis also projected no
`air quality problems under the 2008 ozone NAAQS by 2023.
`Id. at 56,080–56,081.
`
`The EPA agreed with New York, though, that the New
`York Metropolitan Area would likely be in nonattainment of
`the 2015 NAAQS in 2023. See 84 Fed. Reg. at 56,080–56,081.
`
`At Step 2, the EPA “assum[ed], without deciding” that the
`emissions in the nine States identified in the Petition were
`“linked” to air quality problems in the New York Metropolitan
`Area. 84 Fed. Reg. at 56,082.
`
`At Step 3, the EPA denied the Petition in full based on
`New York’s failure to carry its assigned burden of establishing
`significant contributions from upwind sources under either the
`2008 or 2015 NAAQS. In particular, the EPA decided that the
`Petition’s “assessment of whether the sources” could be
`“further controlled through implementation of cost-effective
`controls [was] insufficient[.]” 84 Fed. Reg. at 56,059; see also
`id. at 56,088–56,089.
`
`The EPA reasoned that New York could have met its
`evidentiary burden of demonstrating the availability of cost-
`effective controls by producing “one or more of the following”
`analyses:
`
`
`
`
`
`11
`
`(i) Verifying that the named sources whose emissions
`are those from the most recent emissions inventory
`continue to emit [nitrogen oxides] at the same rate or
`continue to operate; (ii) describing or quantifying
`potentially available emissions reductions from the
`(i.e.,
`named
`sources
`the
`control
`technologies/techniques and the costs of those control
`technologies/techniques);
`(iii) describing
`the
`downwind air quality impacts of controlling the
`named sources
`relative
`to other sources; or
`(iv) providing information on the relative cost of the
`available emissions reductions and whether they are
`less expensive than other reductions from other
`sources.
`
`84 Fed. Reg. at 56,088–56,089.
`
`The EPA then added that it could “[]not determine whether
`it would be appropriate to regulate any of the hundreds of”
`named sources unless they were all “compared to one another
`or * * * compared to other, unnamed sources in the same
`upwind states or in other states.” 84 Fed. Reg. at 56,090. In
`the EPA’s view, this comparison must involve
`
`identifying the current operating status of each named
`facility, the magnitude of emissions from each
`emitting unit within each named facility, the existing
`controls on each of these emissions units, additional
`control options on each emissions unit, the cost of
`each potential control option,
`the emissions
`reductions potential resulting from the installation of
`controls, and potential air quality
`impacts of
`emissions reductions.
`
`Id.
`
`
`
`
`
`12
`
`In addition to finding that New York failed to carry its
`burden with respect to Step 3, the EPA concluded that a prior
`Cross-State Air Pollution Rule Update, which we shall refer to
`as the 2008 Update Rule, had fully addressed any Good
`Neighbor Provision violations arising in the nine named States.
`84 Fed. Reg. at 56,089 (“[T]he EPA has now determined * * *
`that the emissions reductions required under the * * * [2008]
`Update [Rule] fully address the good neighbor requirements
`with respect to the 2008 ozone NAAQS for all the States named
`in the [P]etition.”); see Cross-State Air Pollution Rule Update
`for the 2008 Ozone NAAQS, 81 Fed. Reg. 74,504 (Oct. 26,
`2016).
`
`The State of New York, the State of New Jersey, and the
`City of New York petitioned this court for review. Three
`environmental organizations intervened in support of the
`petitioning States.5 Several parties (collectively, “Industry
`Intervenors”) separately intervened in support of the EPA.6 On
`December 20, 2019, this court granted expedited review.
`
`II
`
`This court has jurisdiction under Section 307(b)(1) of the
`Clean Air Act. See 42 U.S.C. § 7607(b)(1); see also Sierra
`Club v. EPA, 955 F.3d 56, 61 (D.C. Cir. 2020).
`
`
`5 The intervening environmental organizations are Adirondack
`Counsel, Environmental Defense Fund, and Sierra Club.
`
`6 The Industry Intervenors are: Midwest Ozone Group, the Air
`Stewardship Coalition, GenOn Holdings, LLC, the National
`Association of Manufacturers, and the Chamber of Commerce of the
`United States of America.
`
`
`
`
`
`13
`
`We may set aside the EPA’s decision under Section 126 if
`it is “arbitrary, capricious, an abuse of discretion, or otherwise
`not in accordance with law.” 42 U.S.C. § 7607(d)(1)(N),
`(d)(9); see also Maryland, slip op. at 18 (“[W]e apply the same
`standard of review under the Clean Air Act as we do under the
`Administrative Procedure Act.”) (quoting Allied Local & Reg’l
`Mfrs. Caucus v. EPA, 215 F.3d 61, 68 (D.C. Cir. 2000)).
`
`III
`
`The EPA’s reasons for rejecting New York’s Petition were
`arbitrary and capricious in two respects.
`
`First, the EPA failed to provide a reasoned explanation for
`why, under Step 3 of its framework, the Petition failed to show
`that the named sources contributed significantly to downwind
`nonattainment. The EPA’s test, at best, was a moving target
`and, at worst, demanded likely unattainable standards of proof.
`
`Second, binding circuit precedent flatly rejects the two
`grounds on which the EPA relied in deciding, under Step 1, that
`the New York Metropolitan Area did not have a cognizable air
`quality problem under the 2008 NAAQS.
`
`A
`
`The EPA denied the Petition in full, as to compliance with
`both the 2008 and 2015 NAAQS, under Step 3. Specifically,
`the EPA pointed to perceived inadequacies in New York’s
`evidence that cost-effective emission reductions could be
`imposed at the sources of the offending contamination. 84 Fed.
`Reg. at 56,059, 56,088–56,089.
`
`The central problem is that the standard by which the EPA
`deemed New York’s cost-effectiveness showing
`to be
`insufficient is impossible to discern because the explanation
`
`
`
`
`
`14
`
`kept shifting. And if the standard truly means what the EPA’s
`decision at times says, it would be nigh impossible to meet.
`
`First, the EPA’s decision denying the Petition said that
`New York could have carried its burden by undertaking “one
`or more of” four possible analyses. 84 Fed. Reg. at 56,088–
`56,089. By way of reminder, those were: (i) “[v]erifying that
`the named sources * * * continue to emit [nitrogen oxides] at
`the same rate or continue to operate”; (ii) “describing or
`quantifying potentially available emissions reductions from the
`named sources”; (iii) “describing the downwind air quality
`impacts of controlling the named sources relative to other
`sources”; and (iv) evaluating “the relative cost of the available
`emissions reductions and whether they are less expensive than
`other reductions from other sources.” Id. at 56,088–56,089.
`
`Taking the EPA at its word, the Petition’s satisfactory
`demonstration of any “one” of those prongs should have
`sufficed. 84 Fed. Reg. at 56,088.
`
`Yet the EPA denied the Petition without any reasoned
`explanation as to how New York failed to satisfy the first of the
`four analyses, in particular with respect to sources that are
`electric generating units, or “EGUs.” J.A. 76 (“Appendix B
`includes average emission rates by EGU facility for the 2014
`to 2016 period (these data are unavailable for non-EGUs)[.]”);
`see also J.A. 90–92 (listing EGU emission rates). The agency’s
`decision never offered a coherent explanation for why it
`nonetheless rejected the Petition in full at Step 3. It simply
`went on to discuss more potential hurdles for New York’s
`Petition to clear. 84 Fed. Reg. at 56,089–56,090.
`
`At oral argument, the agency could not say whether New
`York had satisfied the first of the four listed analyses. Counsel
`simply said it was “questionable” whether the first analytical
`option had been met. Oral Arg. Tr. 41:22–24. But the EPA
`
`
`
`
`
`15
`
`cannot sensibly reject a petition on the ground that it has not
`yet figured out if the information provided is sufficient.
`
`Perhaps recognizing the problem, the EPA sidestepped the
`issue by claiming that the four analyses proposed by the agency
`are not “a specific test,” and instead simply “lay[] out the
`categories of things [the agency] is looking for.” Oral Arg. Tr.
`42:9–11; see also id. at 44:5–11. So, we are told, when the
`EPA said “one or more” in its decision, it actually meant more
`than one but maybe not all. Id. at 44:10–12.
`
`We are at a loss. Nowhere does the decision explain which
`of these four analyses are necessary or sufficient. Instead, the
`EPA faulted New York for failing to provide “this or any such
`similar analyses[.]” 84 Fed. Reg. at 56,089. But the decision
`never explains what “this” analysis is or why the Petition did
`not meet it. Nor did it shed light on what “similar analysis”
`would suffice. The EPA’s decision just left the court and future
`Section 126(b) petitioners to guess at the agency’s meaning.
`The reasoned agency decisionmaking that the Clean Air Act
`demands, 42 U.S.C. § 7607(d)(1)(N), (d)(9), does not allow the
`EPA to keep moving the finish line.
`
`Second, the EPA’s decision sent contradictory messages
`about whether, or to what extent, New York had to produce a
`global comparative analysis of potential emission reductions at
`listed and unnamed sources within each of the nine States.
`
`The EPA lists “describing the downwind air quality
`impacts of controlling the named sources relative to other
`sources” as one of the four analyses that would have allowed
`New York to meet its burden of proof. 84 Fed. Reg. at 56,089
`(emphasis added). The EPA then explained that it “cannot
`determine whether it would be appropriate to regulate any of
`the hundreds of” named sources unless those sources are first
`“compared to one another or * * * compared to other, unnamed
`
`
`
`
`
`16
`
`sources in the same upwind states or in other states.” Id. at
`56,090. This suggests that a comprehensive comparative
`analysis of all sources—named and unnamed—within each
`designated State is strictly required.
`
`In response to commenters’ concerns that such a universal
`source comparison requirement was unworkable, the EPA
`stated that such “[a]pportioning” of “responsibility for
`emissions reductions across many sources in many states is a
`key outcome of applying the four-step interstate transport
`framework * * * under step 3[.]” 84 Fed. Reg. at 56,089–
`56,090 (emphasis added). The EPA then elaborated that the
`critical “source comparison necessarily
`involves”
`the
`petitioning State
`
`identifying the current operating status of each named
`facility, the magnitude of emissions from each
`emitting unit within each named facility, the existing
`controls on each of these emissions units, additional
`control options on each emissions unit, the cost of
`each potential control option,
`the emissions
`reductions potential resulting from the installation of
`controls, and potential air quality
`impacts of
`emissions reductions.
`
`Id. at 56,090.
`
`that, without such detailed
`The EPA concluded
`about
`individual
`sources’
`comparative
`information
`technological and operational capabilities, the agency “cannot
`determine whether the sources named in the [Petition] have
`available or cost-effective emissions reductions either as
`compared to one another or as compared to other, unnamed
`sources in the same upwind states or in other states.” 84 Fed.
`Reg. at 56,090 (emphasis added). Without that broad swath of
`comparative data, the decision said, the “EPA cannot determine
`
`
`
`
`
`17
`
`whether it would be appropriate to regulate any” of the sources
`identified in New York’s Petition. Id.
`
`But, despite comments flagging the concern, 84 Fed. Reg.
`at 56,089, the EPA left entirely unexplained how States are
`supposed to obtain the required detailed and technically
`particularized internal information from some unknown
`number of unnamed and unidentified sources. On top of the
`crushing breadth of the demand for information from unnamed
`sources across each State, the EPA directed that the analysis
`must “necessarily”
`identify each
`individual
`source’s
`“magnitude of emissions from each emitting unit within each
`named facility,” as well as “the existing controls [and]
`additional control options” for each unit, and “the emissions
`reductions potential resulting from the installation of controls”
`on each unit. Id. at 56,090.
`
`Those analyses—especially determining the emission
`reductions that would result from installing a particular control
`technology on each emitting unit—would require detailed and
`intricate inside knowledge of each facility’s equipment and
`operations. Such information is frequently not publicly
`available, especially for non-EGUs. See Oral Arg. Tr. 34:13–
`17, 59:5–9. Nor did the EPA explain why sources charged with
`polluting would hand such information out at the asking.
`
`At oral argument, the EPA backed away from the plain
`language of its decision, insisting that “[i]t is not EPA’s
`position that a petitioning state would have to do a comparative
` Oral Arg. Tr. 46:6–8.7
`analysis.”
` Rather, the EPA
`
`
`7 But see 84 Fed. Reg. at 56,088–56,089 (EPA directing New
`York to conduct “one or more” of four possible analyses, including
`“describing the downwind air quality impacts of controlling the
`named sources relative to other sources[,]” and analyzing “the
`relative cost of the available emissions reductions and whether they
`
`
`
`
`
`18
`
`characterized such a comparative analysis as simply “one way”
`for a petitioning State to show cost-effectiveness. Id. at 46:13.
`The EPA also insisted that the statement in its decision that
`States can demonstrate cost-effectiveness by “describing the
`downwind air quality impacts of controlling the named sources
`relative to other sources[,]” 84 Fed. Reg. at 56,089 (emphasis
`added), did not suggest a comparative analysis. See Oral Arg.
`Tr. 46:14–23.
`
`This is all quite mystifying. If New York did not have to
`undertake the comparative analysis flagged in two of the four
`proposed analyses and discussed over two pages of the Federal
`Register, and if the EPA cannot definitively say whether New
`York has satisfied “one or more” of the preferred analyses, 84
`Fed. Reg. at 56,088, then we are left with no coherent
`explanation of what was missing from New York’s Petition.
`The required analysis seems to be a constantly moving target,
`with the words of explanation from the agency variously
`meaning and not meaning what they say.
`
`At bottom, the EPA’s Delphic explanation of New York’s
`purported failure to carry its burden of proof—and of even
`what
`that burden
`is—falls
`far
`short of
`reasoned
`decisionmaking. See Environmental Defense Fund v. EPA,
`922 F.3d 446, 454 (D.C. Cir. 2019) (“An agency acts arbitrarily
`
`
`are less expensive than other reductions from other sources”)
`(emphasis added); id. at 56,090 (EPA stating that New York has
`provided insufficient information to allow it to “determine whether
`the sources named in the New York [P]etition have available or cost-
`effective emissions reductions either as compared to one another or
`as compared to other, unnamed sources in the same upwind states or
`in other states”) (emphasis added).
`
`
`
`
`
`19
`
`and capriciously when it offers * * * unreasoned justifications
`for a decision.”).
`
`Third, in addition to dismissing New York’s cost-
`effectiveness analysis as insufficient to support a Step 3
`finding, the decision claims “that the emissions reductions
`required under the * * * [2008] Update [Rule] fully address the
`good neighbor requirements with respect to the 2008 ozone
`NAAQS for all the States named in the [P]etition.” 84 Fed.
`Reg. at 56,089. The EPA added that the electric generating unit
`control technologies identified by New York and by certain
`commenters had already been accounted for in the 2008 Update
`Rule’s trading scheme. Id. at 56,092.
`
`The EPA has abandoned the first contention—that the
`2008 Update Rule fully satisfies
`the Good Neighbor
`requirements under the 2008 NAAQS. And for good reason.
`This court has held that the 2008 Update Rule did not satisfy
`any States’ Good Neighbor obligations. See New York v. EPA,
`781 F. App’x 4, 6–7 (D.C. Cir. 2019) (vacating the EPA’s
`Determination Regarding Good Neighbor Obligations for the
`2008 Ozone National Ambient Air Quality Standard, 83 Fed.
`Reg. 65,878 (Dec. 21, 2018)); see also Wisconsin, 938 F.3d at
`309, 313–318 (holding that the 2008 Update Rule violated the
`Clean Air Act by allowing upwind States to continue
`contributing to downwind air quality problems “beyond the
`statutory deadlines by which downwind States must
`demonstrate their attainment”).
`
`But the EPA continues to press its second rationale—that
`the 2008 Update Rule’s emissions trading scheme fully
`addressed any Good Neighbor Provision obligations associated
`with electric generating units. To be sure, this court’s decision
`in Wisconsin upheld the reasonableness of the Update’s cap-
`
`
`
`
`
`20
`
`its specific emissions budgets.
`and-trade scheme and
`Wisconsin, 938 F.3d at 329–335.
`
`But the EPA is incorrect to argue that Wisconsin also held
`that the 2008 Update Rule comprehensively addressed all Good
`Neighbor Provision obligations associated with electric
`generating units. To the contrary, Wisconsin described the
`2008 Update Rule as only a “first, partial step to addressing a
`given upwind State’s significant contribution.” 938 F.3d