State of New Jersey v. EPA

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued November 2, 2020 Decided March 5, 2021

No. 08-1065

STATE OF NEW JERSEY,
PETITIONER

v.

ENVIRONMENTAL PROTECTION AGENCY,
RESPONDENT

AIR PERMITTING FORUM, ET AL.,
INTERVENORS

On Petition for Review of a Final Action
of the Environmental Protection Agency

Lisa J. Morelli, Deputy Attorney General, Office of the
Attorney General for the State of New Jersey, argued the cause
for petitioner. With her on the briefs was Gurbir S. Grewal,
Attorney General. Jon C. Martin, Deputy Assistant Attorney
General, entered an appearance.

Laura J. Brown, Attorney, U.S. Department of Justice,
argued the cause for respondent. With her on the brief were
Jeffrey Bossert Clark, Assistant Attorney General, Jonathan D.
Brightbill, Principal Deputy Assistant Attorney General, and
 2
Brian L. Doster, Assistant General Counsel, U.S.
Environmental Protection Agency. Eric G. Hostetler and
Norman L. Rave Jr., Attorneys, U.S. Department of Justice,
and Howard J. Hoffman, Counsel, U.S. Environmental
Protection Agency, entered appearances.

Shannon S. Broome argued the cause for intervenors Air
Permitting Forum, et al. in support of respondent. With her on
the brief were Laura K. McAfee, Charles H. Knauss, and
Alexandra K. Hamilton. David M. Friedland, Stacy R. Linden,
Richard S. Moskowitz, Harry M. Ng, and Michele M. Schoeppe
entered appearances.

Before: SRINIVASAN, Chief Judge, ROGERS and WALKER,
Circuit Judges.

Opinion for the Court by Circuit Judge ROGERS.

Dissenting Opinion by Circuit Judge WALKER.

ROGERS, Circuit Judge: This petition for review concerns
the rule promulgated by the Environmental Protection Agency
(“EPA”) upon remand in response to New York v. EPA, 413
F.3d 3 (D.C. Cir. 2005). There, states, environmental
organizations, and industrial entities challenged the revision of
the Clean Air Act’s new source review (“NSR”) program for
preconstruction permitting of stationary sources of air
pollution. This court held that “EPA acted arbitrarily and
capriciously in determining that sources making changes need
not keep records of their emissions if they see no reasonable
possibility that these changes constitute modifications for NSR
purposes,” id. at 11, and remanded for EPA “either to provide
an acceptable explanation for its ‘reasonable possibility’
standard or to devise an appropriately supported alternative,”
id. at 35–36. The State of New Jersey petitions for review on
 3
the grounds that the rule promulgated by EPA on remand
adopts an arbitrary percent trigger and inadequately accounts
for NSR enforcement. Concluding that challenges to the
State’s Article III standing lack merit, we deny the petition on
the merits because the record confirms that EPA engaged in
reasoned decisionmaking.

I.

Under the Clean Air Act (“Act”), “air pollution prevention
. . . and air pollution control at its source is the primary
responsibility of States and local governments.” 42 U.S.C.
§ 7401(a)(3); see also id. § 7407(a); Util. Air Regulatory Grp.
v. EPA, 573 U.S. 302, 308 (2014). The Act, in fact, establishes
“a joint state and federal program for regulating the nation’s air
quality,” Nat’l Ass’n of Clean Air Agencies v. EPA, 489 F.3d
1221, 1224 (D.C. Cir. 2007) (quoting Envtl. Def. v. EPA, 467
F.3d 1329, 1331 (D.C. Cir. 2006)), directing EPA to formulate
national ambient air quality standards (“NAAQS”) that specify
the maximum permissible concentrations of certain air
pollutants, 42 U.S.C. § 7409(b)(1), and requiring states to
develop EPA approved plans, known as State Implementation
Plans (“SIPs”), describing how they will achieve and maintain
the NAAQS, see id. § 7410; Texas v. EPA, 726 F.3d 180, 183
(D.C. Cir. 2013). “States that fail to comply with these
requirements are subject to various sanctions and the
imposition of a Federal Implementation Plan (‘FIP’).”
Appalachian Power Co. v. EPA, 249 F.3d 1032, 1037 (D.C.
Cir. 2001) (citing 42 U.S.C. § 7509).

The new source review (“NSR”) provisions of the Act
apply to “major” stationary sources and emitting facilities, such
as smelters, power plants, and refineries, that directly emit or
have the potential to emit more than one hundred tons per year
of any air pollutant. See 42 U.S.C. §§ 7602(j), 7479(1). NSR
 4
provisions “require ‘new and modified major stationary
sources’ of air pollution to obtain preconstruction permits and
to install pollution control technology in order to protect and
enhance air quality.” New York, 413 F.3d at 21 (citing 42
U.S.C. §§ 7475, 7502, 7503). The “specific pollution control
requirements depend[] upon the geographic location of the
source.” New York v. EPA, 443 F.3d 880, 883 (D.C. Cir. 2006).
The prevention of significant deterioration (“PSD”) provisions
apply to sources located in areas that meet the NAAQS, see 42
U.S.C. §§ 7470–7479, while more stringent nonattainment
NSR (“NNSR”) provisions apply to sources in areas that do not
meet the NAAQS, see id. §§ 7501–7515; see also New York,
413 F.3d at 13.

All newly constructed major stationary sources must
comply with NSR requirements. For existing sources, these
requirements apply only to “major” modifications that will
result in (1) a “significant emissions increase” and (2) a
“significant net emissions increase” of one or more regulated
pollutants. See, e.g., 40 C.F.R. § 52.21(a)(2)(iv). A
“significant emissions increase” is defined by numeric
significance levels for each regulated pollutant. See, e.g., id.
§ 52.21(b)(40). NSR requirements do not apply to routine
maintenance, repair, and replacement. See, e.g., id.
§ 52.21(b)(2)(iii)(a). States are additionally required to
administer programs to regulate “the modification and
construction of any stationary source . . . as necessary to assure
that national air quality standards are achieved.” 42 U.S.C.
§ 7410(a)(2)(C). Modifications not subject to PSD or NNSR
requirements may nonetheless be subject to these “minor NSR”
requirements.

In 2002, EPA revised the methodology to be used by
sources to determine whether a modification is “major” and
therefore subject to NSR requirements. See Prevention of
 5
Significant Deterioration and Nonattainment New Source
Review, 67 Fed. Reg. 80,186 (Dec. 31, 2002) (codified at 40
C.F.R. §§ 51, 52) (“2002 Rule”). Two aspects of the 2002 Rule
are relevant to the instant petition. First, the 2002 Rule adopted
the “actual-to-projected-actual” methodology, so a “significant
emissions increase” is projected to occur if the difference
between the source’s baseline actual emissions and its
projected actual emissions equals or exceeds the significance
level for that pollutant. See, e.g., 40 C.F.R.
§ 52.21(a)(2)(iv)(c). To calculate projected actual emissions, a
source “[s]hall consider all relevant information, including but
not limited to, historical operational data, the company’s own
representations, the company’s expected business activity and
the company’s highest projections of business activity, the
company’s filings with the State or Federal regulatory
authorities, and compliance plans under the approved State
Implementation Plan.” See, e.g., id. § 52.21(b)(41)(ii)(a). This
calculation “[s]hall include fugitive emissions to the extent
quantifiable, and emissions associated with startups,
shutdowns, and malfunctions.” See, e.g., id.
§ 52.21(b)(41)(ii)(b). It “[s]hall exclude . . . any increased
utilization due to product demand growth.” See, e.g., id.
§ 52.21(b)(41)(ii)(c). Second, the 2002 Rule mandated
compliance with its recordkeeping and reporting requirements
if a source determined that (1) a modification did not trigger
NSR, and (2) there was still a “reasonable possibility” that the
modification would result in a “significant emissions increase.”
See, e.g., id. § 52.21(r)(6). The 2002 Rule did not define the
term “reasonable possibility.”

The court upheld the 2002 Rule’s “actual-to-projected-
actual” methodology while holding that its “reasonable
possibility” standard was arbitrary and capricious. New York,
413 F.3d at 10–11, 33–36. The 2002 Rule “allow[ed] sources
that take advantage of the ‘reasonable possibility’ standard to
 6
avoid recordkeeping altogether” without providing any
explanation as to how EPA could prove NSR transgressions if
sources deciding no reasonable possibility of a significant
emissions increase exists kept no data. Id. at 35. EPA also
failed to explain how, absent such data, reporting requirements
of the Act’s Title V program and minor NSR programs
provided enforcement authorities with the relevant
information. Id. The court observed that the “intricacies of the
actual-to-projected-actual methodology will aggravate the
enforcement difficulties stemming from the absence of data.”
Id.

In response to the court’s remand, EPA issued a notice of
proposed rulemaking and sought comments on two options for
defining “reasonable possibility,” including its preferred option
of using 50 percent of the applicable significance level for a
regulated pollutant as the trigger for NSR recordkeeping and
reporting requirements. See Prevention of Significant
Deterioration and Nonattainment New Source Review:
Reasonable Possibility in Recordkeeping, 72 Fed. Reg. 10,445,
10,449 (proposed Mar. 8, 2007). Thereafter, upon considering
comments, EPA promulgated the final rule. See Prevention of
Significant Deterioration and Nonattainment New Source
Review: Reasonable Possibility in Recordkeeping, 72 Fed.
Reg. 72,607 (Dec. 21, 2007) (codified at 40 C.F.R. §§ 51.165,
51.166, 52.21) (“Rule”). The Rule provides that:

“(vi) A ‘reasonable possibility’ under paragraph (a)(6)
of this section occurs when the owner or operator
calculates the project to result in either:

(A) A projected actual emissions increase of at
least 50 percent of the amount that is a
‘significant emissions increase,’ as defined
under paragraph (a)(1)(xxvii) of this section
 7
(without reference to the amount that is a
significant net emissions increase), for the
regulated NSR pollutant; or

(B) A projected actual emissions increase that,
added to the amount of emissions excluded
under paragraph (a)(1)(xxviii)(B)(3) [emissions
attributable to demand growth], sums to at least
50 percent of the amount that is a ‘significant
emissions increase,’ as defined under paragraph
(a)(1)(xxvii) of this section (without reference
to the amount that is a significant net emissions
increase), for the regulated NSR pollutant. For
a project for which a reasonable possibility
occurs only within the meaning of paragraph
(a)(6)(vi)(B) of this section, and not also within
the meaning of paragraph (a)(6)(vi)(A) of this
section, then provisions (a)(6)(ii) through (v)
[reporting requirements] do not apply to the
project.” See, e.g., 40 C.F.R. § 52.21(r)(6)(vi).

The State of New Jersey timely filed a petition for review
with this court and a petition for reconsideration with EPA. In
April 2009, EPA agreed to reconsider the Rule, and the court
held the petition for review in abeyance pending EPA’s
reconsideration. EPA took no action on the reconsideration
until November 2019, when it informed petitioner that it was
no longer reconsidering the Rule. The court lifted abeyance of
the petition and therefore proceeds to review of the petition.

II.

Petitioner essentially contends that the Rule fails to correct
the flaws in the 2002 Rule identified in New York, 413 F.3d 3.
See Petitioner Br. 2, 29. Specifically, it maintains that EPA
 8
erred by (1) inadequately accounting for enforcement
difficulties stemming from the predictive nature of calculating
projected emissions; (2) adopting an arbitrary 50 percent
trigger; and (3) failing to explain how authorities could enforce
NSR when sources determined that recordkeeping and
reporting requirements were not triggered. See id. 29–30. EPA
disagrees, responding that it satisfied the court’s remand
direction in New York, 413 F.3d 3, and “reasonably balanced
environmental enforcement concerns with economic and
administrative concerns by establishing a bright-line 50 percent
trigger.” Respondent Br. 16.

As a threshold matter, intervenors challenge petitioner’s
Article III standing. EPA does not raise this objection, but the
court has an “independent obligation” to review petitioner’s
standing before addressing the merits. Summers v. Earth
Island Inst., 555 U.S. 488, 499 (2009); Grocery Mfrs. Ass’n v.
EPA, 693 F.3d 169, 174 (D.C. Cir. 2012). As the party
invoking the court’s jurisdiction, petitioner bears the burden of
establishing its standing, Lujan v. Defs. of Wildlife, 504 U.S.
555, 561 (1992), and clearly met that burden here.

“To establish Article III standing, a petitioner must
demonstrate it has suffered a concrete and particularized injury
that is imminent and not conjectural, that was caused by the
challenged action, and that is likely to be redressed by a
favorable judicial decision.” Texas, 726 F.3d at 198 (citing
Lujan, 504 U.S. at 560–61). On direct review of agency action,
when standing is not self-evident, a petitioner must
“supplement the record to the extent necessary to explain and
substantiate its entitlement to judicial review.” Grocery Mfrs.
Ass’n, 693 F.3d at 174 (quoting Sierra Club v. EPA, 292 F.3d
895, 900 (D.C. Cir. 2002)). “[I]n reviewing the standing
question, the court must be careful not to decide the questions
on the merits for or against the plaintiff, and must therefore
 9
assume that on the merits the plaintiffs would be successful in
their claims.” City of Waukesha v. EPA, 320 F.3d 228, 235
(D.C. Cir. 2003) (citing Warth v. Seldin, 422 U.S. 490, 502
(1975)); see also Air All. Houston v. EPA, 906 F.3d 1049, 1057
(D.C. Cir. 2018). Consequently, the court must assume for
purposes of standing that the Rule is arbitrary and capricious
because it inadequately accounts for NSR enforcement.
Further, petitioner is “entitled to special solicitude” in our
standing analysis because it has “quasi-sovereign interests” in
reducing air pollution and a procedural right to challenge the
Rule under 42 U.S.C. § 7607(b)(1). Massachusetts v. EPA, 549
U.S. 497, 520 (2007); see also North Carolina v. EPA, 587
F.3d 422, 426 (D.C. Cir. 2009).

Standing is usually self-evident when the petitioner is an
object of the challenged government action. See Lujan, 504
U.S. at 561–62. Although the Rule itself does not formally
regulate petitioner, it directly implicates petitioner’s ability to
comply with its statutory obligations in administering the NSR
program. The Act entrusts states and local governments with
the primary responsibility for enforcing NSR requirements.
States must include in their SIPs a “program to provide for the
enforcement” of NSR provisions. 42 U.S.C. § 7410(a)(2)(C);
see also id. § 7413(a)(2). Indeed, the preamble acknowledges
that the Rule affects states’ obligations to implement the NSR
scheme. See 72 Fed. Reg. at 72,613–14. Identifying those to
whom the Rule applies, the preamble specifies that the Rule
affects petitioner: “Entities affected by this final rule include
major stationary sources in all industry groups . . . Entities
affected by the rule also include States, local permitting
authorities, and Indian country.” Id. at 72,608.

Petitioner has identified two injuries, either of which
suffices to establish standing to challenge the Rule. Because
redressability follows from the court’s conclusion about
 10
causation, our inquiry focuses on the injury-in-fact and
causation requirements of standing.

First, petitioner maintains that the Rule’s inadequate
recordkeeping and reporting requirements impair its delegated
authority to implement the PSD program for in-state sources.
See Petitioner Br. 27. In support, it points to the declaration of
Danny Wong, Chief of the Bureau of Stationary Sources in the
Division of Air Quality of the New Jersey Department of
Environmental Protection, stating that absent stricter
recordkeeping and reporting requirements, petitioner lacks the
resources necessary to ensure that in-state sources comply with
PSD requirements. See Declaration of Danny Wong in Support
of Petitioner’s Standing (13 Apr. 2020) at ¶¶ 13–17.

Notably, intervenors do not dispute that the exacerbated
administrative costs and burdens imposed by the Rule on
petitioner constitute a concrete and particularized injury. See
Clean Wisconsin v. EPA, 964 F.3d 1145, 1158 (D.C. Cir.
2020). Rather, they argue that petitioner’s alleged injury is
self-inflicted and therefore cannot be fairly traced to the Rule.
They rely on Nat’l Family Planning & Reprod. Health Ass’n,
Inc. v. Gonzales, 468 F.3d 826, 831 (D.C. Cir. 2006), for the
proposition that petitioner “could easily require in-state sources
to keep the more extensive records sought” because the Act
“expressly reserves to states the authority to impose more
stringent requirements on stationary sources than the minimum
federal requirements, including for recordkeeping.”
Intervenors Br. 40 (citing 42 U.S.C. § 7416). EPA, in
addressing the merits of the petition, echoes intervenors’
contention that petitioner’s injury is “self-inflicted” and
additionally invokes Pennsylvania v. New Jersey, 426 U.S.
660, 664 (1976). Respondent Br. 35.
 11
Neither case is apposite as the plaintiffs’ alleged injuries
there stemmed from their own voluntary action or inaction. In
Gonzales, 468 F.3d at 827–28, the plaintiff’s argument that a
statute was unconstitutionally vague rested on an alleged
conflict between the statute and certain regulations. The court
rejected the argument because the plaintiff “ha[d] chosen to
remain in the lurch” despite having “within its grasp an easy
means for alleviating the alleged uncertainty.” Id. at 831.
Plaintiff had failed to inquire as to how the agency proposed to
resolve the alleged conflicts or petition the agency under the
Administrative Procedure Act to adopt a clarifying rule. Id. In
Pennsylvania, 426 U.S. at 663–64, which concerned the
Supreme Court’s original jurisdiction as opposed to Article III
standing, plaintiff-states that gave their residents credits for
taxes paid to other states alleged a loss of revenue due to
defendant-states’ taxes on nonresidents’ income. Because
nothing required plaintiff-states to extend the tax credits, the
Supreme Court concluded that the “injuries to plaintiffs’ fiscs
were self-inflicted, resulting from decisions by their respective
state legislatures.” Id. at 664.

Furthermore, intervenors’ argument is contrary to this
court’s precedent, which has not treated a state’s ability to
change its laws to evade injury as precluding standing to
challenge EPA’s actions under the Act. EPA’s actions injure
states when those actions necessitate changes to state laws and
make “the states’ task of devising an adequate SIP” “more
difficult and onerous.” W. Virginia v. EPA, 362 F.3d 861, 868
(D.C. Cir. 2004); accord Nat’l Ass’n of Clean Air Agencies,
489 F.3d at 1227–28. Regardless of whether in-state sources
“cheat,” by forcing petitioner to expend additional resources or
change its laws in order to verify sources’ calculations and
ensure compliance with PSD requirements, the Rule makes
petitioner’s “task of complying with the Clean Air Act ‘more
difficult and onerous.’” Dissent at 13 (quoting W. Virginia,
 12
362 F.3d at 868). And contrary to intervenors’ suggestion that
petitioner has an “easy means” to alleviate any harm caused by
the Rule, Gonzales, 468 F.3d at 831, the preamble to the Rule
explains that states wishing to impose more stringent
recordkeeping and reporting requirements must still apply for
and obtain EPA’s approval. See 72 Fed. Reg. at 72,614.
Moreover, the Supreme Court’s decision in Clapper v. Amnesty
Int’l USA, 568 U.S. 398, 416 (2013), reinforces the conclusion
that petitioner’s injury is not self-inflicted. It too cited
Pennsylvania, 426 U.S. at 660, and Gonzales, 468 F.3d at 831,
for the proposition that “respondents cannot manufacture
standing merely by inflicting harm on themselves based on
their fears of hypothetical future harm that is not certainly
impending.” Clapper, 568 U.S. at 416. But nothing here
indicates that petitioner assumed authority to implement the
PSD program in order to manufacture standing. To the
contrary, EPA delegated that authority to petitioner. Petitioner
has therefore established standing based on the asserted harm
to its delegated authority to implement the PSD program.

Additionally, petitioner has established standing based on
harm to its ability to attain the NAAQS due to unlawful
emissions from upwind states. See Petitioner Br. 27; see also
Declaration of Sharon C. Davis in Support of Petitioner’s
Standing (13 Apr. 2020). A petitioner alleging future injuries
“can establish standing by satisfying either the ‘certainly
impending’ test or the ‘substantial risk’ test.” Attias v.
Carefirst, Inc., 865 F.3d 620, 626–27 (D.C. Cir. 2017) (citing
Susan B. Anthony List v. Driehaus, 573 U.S. 149, 158 (2014),
and other precedent from this circuit); accord Dep’t of
Commerce v. New York, 139 S. Ct. 2551, 2565 (2019). Indeed,
this court has “frequently upheld claims of standing based on
allegations of a ‘substantial risk’ of future injury.” Attias, 865
F.3d at 627. Under that standard, petitioner must show “both
(i) a substantially increased risk of harm and (ii) a substantial
 13
probability of harm with that increase taken into account.”
Food & Water Watch, Inc. v. Vilsack, 808 F.3d 905, 914 (D.C.
Cir. 2015) (internal citations omitted). The declaration of
Sharon C. Davis, Manager of the Bureau of Evaluation and
Planning in the Division of Air Quality of the New Jersey
Department of Environmental Protection, which accompanied
petitioner’s opening brief, explains that cross-state emissions
contribute significantly to petitioner’s ozone levels. See Davis
Decl. ¶¶ 12–14. It cites EPA findings that westerly or
southwesterly air masses moving through the Ohio River
Valley or the Great Lakes area carry ozone to the Northeast and
that more than 70 percent of a state’s ground-level ozone may,
on average, be attributable to precursor emissions from upwind
states. Id. ¶ 14. Such “cross-state emissions from upwind
states have prevented New Jersey from coming into attainment
of the 2008 ozone NAAQS” despite “aggressive measures to
reduce in-state emissions of ozone precursors.” Id. ¶¶ 15–16.
Petitioner has reduced its emissions of ozone precursors by
approximately 77 percent from 1990 to 2017, and between 31
percent and 11 percent annually from 2011 to 2017. Id. ¶ 15.
But high emissions from upwind states jeopardize petitioner’s
ability to come into compliance with the ozone NAAQS by
attainment deadlines in 2021 and 2023 specified by EPA. Id.
¶ 16. In other words, increased cross-state emissions
substantially increase both the risk and probability of harm to
petitioner’s ability to attain the ozone NAAQS.

Intervenors and our dissenting colleague nevertheless
maintain that petitioner’s asserted injury, even if cognizable,
cannot be fairly traced to the Rule by an attenuated chain of
events based on “assumptions regarding the conduct of
unidentified third parties not before the court.” Intervenors Br.
37; see also Dissent at 12. The uncontested factual statements
in the declarations of New Jersey officials notwithstanding,
they conclude that petitioner “offers no facts to substantiate its
 14
theory that third parties will illegally emit air pollution as a
result of the Rule.” Intervenors Br. 39; see also Dissent at 2–
12. Yet reasonably characterized, petitioner’s causal chain
involves three links: (1) the Rule’s inadequate recordkeeping
and reporting requirements make NSR enforcement more
difficult in upwind states; (2) inadequate NSR enforcement
increases the risk of unlawful cross-state emissions; and (3)
such cross-state emissions risk hampering petitioner’s efforts
to attain and maintain the NAAQS. All are supported by the
record.

EPA effectively conceded the first link in the preamble to
the Rule: “If ease of enforcement were our only consideration,
it would point us toward the most inclusive of recordkeeping
and reporting requirements.” 72 Fed. Reg. at 72,610. The
Davis declaration substantiates the third link, connecting cross-
state emissions with petitioner’s ability to attain the NAAQS.
See Davis Decl. ¶¶ 12–16. For the second link, petitioner need
not show that out-of-state sources will definitively emit more
illegal air pollution. “A permissible theory of standing ‘does
not rest on mere speculation about the decisions of third parties;
it relies instead on the predictable effect of Government action
on the decisions of third parties.’” Competitive Enter. Inst. v.
FCC, 970 F.3d 372, 381 (D.C. Cir. 2020) (quoting Dep’t of
Commerce, 139 S. Ct. at 2566). In fact, “an entire line of cases
finds redressability, as well as causation, in . . . circumstances
turning on third-party conduct that is voluntary but reasonably
predictable.” Id. at 384. “In considering the likely reaction of
third parties, we may consider a variety of evidence, including
the agency’s own factfinding; affidavits submitted by the
parties; evidence in the administrative record; arguments
firmly rooted in the basic laws of economics; and conclusions
in other agency orders and rulemakings.” Id. at 382 (internal
citations and quotation marks omitted).
 15
The record demonstrates that an increased risk of cross-
state emissions is the predictable effect of inadequate NSR
enforcement. The Rule’s premise implies a causal connection
between inadequate NSR enforcement and the increased risk of
sources emitting unlawful cross-state emissions. The Rule
itself requires sources to keep records and report their
preconstruction emissions calculations precisely because they
might miscalculate them and erroneously evade NSR
requirements. See 72 Fed. Reg. at 72,610. In the preamble,
EPA repeatedly acknowledges that sources underestimate or
miscalculate their preconstruction emissions calculations. See
id. at 72,610–11. The record also features at least nine
comments discussing the variability and possibility of error in
sources’ preconstruction emissions calculations. See Response
to Comment Document for the PSD and Nonattainment NSR:
Reasonable Possibility in Recordkeeping: Final Rule (Dec.
2007), EPA-HQ-OAR-2001-0004-0848, at pp. 19–21
(“Response to Comment Document”). Contrary to our
dissenting colleague’s characterization, these do not “merely
express non-specific, conclusory fears about the rule’s
enforceability.” Dissent at 4. For instance, the National
Association of Clean Air Agencies’ comment referenced in the
dissent cites reports by the National Academy of Public
Administration and the Government Accountability Office, as
well as congressional testimony of a former EPA Administrator
expressing concerns about allowing polluting sources to “self-
police” their NSR applicability. See Comments of the National
Association of Clean Air Agencies (NACAA) (May 7, 2007),
EPA-HQ-OAR-2001-0004-0822, at p. 4; see also Staying
Healthy: Health Issues Surrounding Proposed Changes in
Clean Air Standards: Hearing on Examining Proposed
Improvements to the NSR Program Before the Sub. Comm. on
Pub. Health of the S. Comm. on Health, 107th Cong. 38–39
(2002) (statement of Carol M. Browner, Partner, The Albright
Group, LLC, and Former EPA Administrator). Moreover,
 16
petitioner cites examples of NSR litigation actions
exemplifying underestimations of emissions calculations by
sources. See Petitioner Br. 34–36.

Our dissenting colleague would require a “more robust
record” of empirical evidence “showing when, where, or how
often out-of-state polluters will make major changes that evade
the permitting process in a way that dirties New Jersey’s air.”
Dissent at 3, 14. Under well-established standing precedent,
however, it suffices to point to “third party conduct that is
voluntary but reasonably predictable.” Competitive Enter.
Inst., 970 F.3d at 384. And under that precedent, petitioner
need not produce “empirical study piled on empirical study
predicting with specificity . . . how many third parties would
injure” it as a “direct result” of the Rule. Dissent at 5. Given
petitioner’s demonstrated likelihood of out-of-state sources
miscalculating their projected emissions, the court may
properly accept that the “predictable effect” of inadequate NSR
enforcement would be increased illegal cross-state emissions.
Competitive Enter. Inst., 970 F.3d at 381 (quoting Dep’t of
Commerce, 139 S. Ct. at 2566). Because the substantial risk of
interference with petitioner’s ability to maintain the NAAQS
fairly traces to the Rule’s inadequate recordkeeping and
reporting requirements, petitioner has met its burden, once
again, to show standing.

III.

Satisfied of our jurisdiction, we turn to the merits of the
petition. The Act authorizes the court to “reverse any [EPA]
action found to be . . . arbitrary, capricious, an abuse of
discretion, or otherwise not in accordance with law.” 42 U.S.C.
§ 7607(d)(9)(a). “A rule is arbitrary and capricious if the
agency: (1) ‘has relied on factors which Congress has not
intended it to consider,’ (2) ‘entirely failed to consider an
 17
important aspect of the problem,’ (3) ‘offered an explanation
for its decision that runs counter to the evidence before the
agency,’ or (4) ‘is so implausible that it could not be ascribed
to a difference in view or the product of agency expertise.’”
United States Sugar Corp. v. EPA, 830 F.3d 579, 606 (D.C.
Cir. 2016) (quoting Motor Vehicle Mfrs. Ass’n of U.S., Inc. v.
State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983)). “Our
review under the ‘arbitrary and capricious’ standard is narrow
and does not permit us to substitute our policy judgment for
that of the Agency.” Maryland v. EPA, 958 F.3d 1185, 1210
(D.C. Cir. 2020) (quoting Bluewater Network v. EPA, 370 F.3d
1, 11 (D.C. Cir. 2004)). Applying this deferential standard,
the court concludes that petitioner’s challenges to the Rule fail.

Petitioner contends that EPA inadequately considered
concerns stemming from the predictive and subjective nature
of projected emissions calculations. Specifically, it maintains
that the “[i]nclusion of demand growth and startup, shutdown,
and malfunction elements as well as netting analyses make
predicting emissions under the actual-to-projected-actual
methodology susceptible to manipulation.” Petitioner Br. 34.
Petitioner focuses on the Rule’s requirement that sources
maintain records but not report their preconstruction emissions
calculations and post-construction emissions when their
projected emissions increase equals or exceeds the 50 percent
trigger due to demand growth emissions. Id. at 37. Absent
independent verification, sources have compelling incentives,
petitioner asserts, to apply “unsubstantiated or overly
optimistic pre-construction analyses” and evade NSR
requirements. Id. at 35. In its view, EPA actions following the
Rule’s promulgation magnify these enforcement concerns. Id.
at 39–40.

The record, however, demonstrates that EPA adequately
considered the enforcement problems referenced by petitioner.
 18
To address concerns about demand growth emissions raised by
commenters, including petitioner, EPA modified the proposed
rule to trigger recordkeeping requirements when the projected
emissions increase added to demand growth emissions equaled
or exceeded the 50 percent trigger. See 72 Fed. Reg. at 72,609–
10. In the preamble to the Rule, EPA reasoned that this “pre-
change recordkeeping requirement establishes an adequate
paper trail to allow enforcement authorities to evaluate the
source’s claims concerning what amount of an emissions
increase is related to the project and what amount is attributable
to demand growth.” Id. at 72,611. In EPA’s opinion, this
approach “balances ease of enforcement with avoidance of
requirements that would be unnecessary or unduly burdensome
on reviewing authorities or the regulated community.” Id.
Petitioner points to no contrary evidence undermining EPA’s
approach to demand growth emissions.

Petitioner’s concerns about netting analyses are likewise
unavailing because the Rule excludes netting analyses from
projected emissions calculations. See, e.g., 40 C.F.R.
§ 51.165(a)(6)(vi)(A). Further, EPA offered a reasoned
conclusion about why projections of fugitive, startup, or
malfunction emissions are not “likely to be significant causes
of variability or error that would lead to underestimates of
emissions increases from existing units.” 72 Fed. Reg. at
72,612. These types of emissions, EPA pointed out, are
“included in the project’s baseline actual emissions, and [there
was] no reason to expect greater amounts of these types of
emissions in the post-project projections.” Id. Nothing in the
record contradicts or casts doubt on this reasoning.
Enforcement problems stemming from EPA’s actions
following the Rule’s promulgation are beyond the current
record for judicial review. 42 U.S.C. § 7607(d)(7)(A).
 19
Petitioner next challenges EPA’s choice of the 50 percent
trigger. Relying on Atlas Copco, Inc. v. EPA, 642 F.2d 458,
465 (D.C. Cir. 1979), it maintains that the Rule remains
“completely subjective” and “impermissibly vague”: “If a
source determines, based on whatever criteria it decides to use,
that a ‘reasonable possibility’ of crossing the 50% threshold
does not exist, there is no independent check on that
determination.” Petitioner Br. 42. The Rule, in fact, gives clear
guidance that recordkeeping and reporting requirements apply
if the projected emissions increase equals or exceeds 50 percent
of a regulated pollutant’s numeric significance levels. See 72
Fed. Reg. at 72,616–17. NSR regulations codifying the 2002
Rule specify the methodology and criteria to be used to
calculate such projected emissions. See, e.g., 40 C.F.R.
§ 52.21(b)(41). In New York, 413 F.3d at 10, 31–32, the court
upheld the use of projected emissions as well as the criteria for
their calculation. The Rule, therefore, does not amount to a
“blanket requirement compelling compliance in the absence of
an indication of the factors considered controlling.” Atlas
Copco, 642 F.2d at 465.

Furthermore, EPA offered a rational basis for adopting the
50 percent trigger. An agency “is not required to identify the
optimal threshold with pinpoint precision,” but required only
“to identify the standard and explain its relationship to the
underlying regulatory concerns.” WorldCom, Inc. v. FCC, 238
F.3d 449, 461–62 (D.C. Cir. 2001). In New York, 413 F.3d at
34, the court recognized that “less burdensome requirements
may well be appropriate for sources with little likelihood of
triggering NSR.” The preamble to the Rule explains that, in
finalizing a bright-line test, EPA “strove for a balance between
ease of enforcement and avoidance of requirements that would
be unnecessary or unduly burdensome on reviewing authorities
or the regulated community.” 72 Fed. Reg. at 72,610. In
addition to its preferred 50 percent trigger, EPA solicited
 20
comments on other possible values such as 25, 33, 66 or 75
percent. Id. at 72,611; see also id. at 10,449. Nineteen of the
thirty-eight commenters supported a trigger of 50 percent or
higher. See Response to Comment Document at p. 7. As noted,
EPA’s final rule accounted for variability in projections due to
demand growth emissions and thereby addressed the principal
objection of commenters, including petitioner, to the 50 percent
trigger. See 72 Fed. Reg. at 72,611. EPA explained that
commenters’ other objections to the 50 percent trigger were
“general” and “did not give specific examples of projects for
which sources would project emissions increases of less than
50 percent of the significant level but would nevertheless be
likely to cause emissions increases above the significant level.”
Id. Petitioner does not identify any contradictory record
evidence or objections that EPA failed to consider. In these
circumstances, EPA could reasonably determine that the 50
percent trigger “will capture projects that have a higher
probability of variability and/or error in projected emissions.”
Id.

Petitioner further challenges EPA’s explanation that
enforcement authorities may rely on other records — such as
Title V records, minor NSR records, state and national
emissions inventory records, and business records — to
evaluate preconstruction NSR compliance when the Rule’s
recordkeeping and reporting requirements are not triggered. It
principally maintains that such records lack the type of project-
specific, preconstruction information needed to evaluate NSR
compliance. It also maintains that EPA failed to explain how
enforcement authorities may draw on these records collectively
to trace emissions increases to specific modifications. Even
assuming for the purposes of argument that non-NSR specific
records are poor substitutes, petitioner fails to show how EPA
acted arbitrarily or capriciously. It cites no authority to support
the proposition that EPA had an obligation to show that its
 21
“reasonable possibility” standard achieves “perfect NSR
compliance.” New York, 413 F.3d at 44 (Williams J.,
concurring). Rather, EPA’s obligation was “to analyze the
trade-off between compliance improvement and the burdens of
data collection and reporting” and “articulate a reasoned
judgment as to why any proposed additional burden would not
be justifiable in terms of the likely enhancement of
compliance.” Id. By adequately considering NSR enforcement
concerns raised during this rulemaking and offering a reasoned
explanation for its 50 percent trigger, EPA satisfied this
obligation. On this record, petitioner otherwise fails to show
that EPA’s action was arbitrary or capricious.

Accordingly, we deny the petition for review.
 WALKER, Circuit Judge, dissenting: New Jersey claims
the EPA promulgated an arbitrary and capricious rule. For the
reasons explained in the Court’s thoughtful opinion, New
Jersey’s argument lacks merit. But rather than reaching the
merits, I would, with respect, dismiss New Jersey’s petition for
lack of Article III standing.1 New Jersey has not shown the
EPA caused it to “suffer[] an injury in fact . . . that is fairly
traceable” to the EPA’s rule.2

* * *

A command by the Clean Air Act requires large-scale
emitters of air pollutants to get permits when they make certain
changes to their facilities.3 The permit process is triggered by
changes that are “major.”4 New Jersey argues the EPA should
impose stricter record-keeping and reporting requirements on
out-of-state polluters who aren’t making changes that are
expected to be “major” or have a “reasonable possibility”5 of
requiring a permit. New Jersey’s theory of standing goes like
this:

1. In the future, some out-of-state companies will desire
a change that has a “reasonable possibility” of
triggering the permit process.

2. Maybe one of those companies would not receive a
permit if it applied for one.

3. So maybe that company will avoid triggering the
permitting process by underestimating its expected
emissions.

1
U.S. CONST. art. III, § 2.
2
Spokeo, Inc. v. Robins, 136 S. Ct. 1540, 1547 (2016).
3
See 42 U.S.C. § 7475.
4
See 40 C.F.R. § 52.21(a)(2)(iii).
5
Id. § 52.21(r)(6).
 2

4. Maybe the company’s change ends up being major.

5. Maybe the EPA will not detect that deception (or
unintentional miscalculation) because the company
didn’t have to keep any records.

6. Maybe the pollution caused by the undetected major
change will reach New Jersey.

In other words, pollution might reach New Jersey from a
company that might have hidden a major change’s pollution
from the EPA, which might have denied a permit for the major
change that might have gone undetected.

For the sake of argument, I will assume that if that
happens, New Jersey will be injured in two ways. First, its
citizens will breathe dirtier air. Second, it will be harder for
New Jersey to meet its obligations under the Clean Air Act.6

In addition, I will assume that most of steps one through
six will happen. For example, with regard to step six, the
record shows that about 70% of New Jersey’s ozone is made
up of air pollutants from other states.7 So if sources close to
New Jersey make major changes that would not be permitted,
and if they avoid detection because of the rule, those illegal
emissions will end up harming New Jersey.

However, I cannot assume, and New Jersey has not shown,
that the third step above — the EPA’s rule will cause injurious
underreporting of undetected major changes — is a

6
See 42 U.S.C. §§ 7409-7410.
7
Davis Decl. at ¶ 14.
 3
“certain[ty],”8 a “predictable effect,”9 or a “substantial risk.”10
Even though courts owe states “special solicitude”11 in EPA
emissions cases, this solicitude doesn’t cover unknown injuries
inflicted by unknown companies at some unknown time in the
possibly distant future.12

Perhaps New Jersey could have established in the record
that the “likely reaction” of sources to the rule will be to
underestimate expected emissions to avoid detection.13 But
that’s not this record. New Jersey has directed us to no
evidence in the record showing when, where, or how often out-
of-state polluters will make major changes that evade the
permitting process in a way that dirties New Jersey’s air.

Instead, New Jersey has provided us with only a couple
declarations, about a dozen comments, the rule’s preamble, and
the purpose of the rule. But the declarations are long on
conclusory statements and data about how much pollution from
upwind states reaches New Jersey, and short on how much of
8
Clapper v. Amnesty International USA, 568 U.S. 398, 410 (2013).
9
Competitive Enterprise Institute v. FCC, 970 F.3d 372, 381 (D.C.
Cir. 2020) (quoting Department of Commerce v. New York, 139 S.
Ct. 2551, 2566 (2019)).
10
Food & Water Watch, Inc., v. Vilsack, 808 F.3d 905, 914 (D.C.
Cir. 2015) (quoting Susan B. Anthony List v. Driehaus, 573 U.S. 149,
158 (2014)); cf. Public Citizen, Inc. v. National Highway Traffic
Safety Administration, 489 F.3d 1279, 1298 (D.C. Cir. 2007)
(“[W]ere all purely speculative increased risks deemed injurious, the
entire requirement of actual or imminent injury would be rendered
moot, because all hypothesized, nonimminent injuries could be
dressed up as increased risk of future injury.”) (cleaned up).
11
Massachusetts v. EPA, 549 U.S. 497, 520 (2007).
12
Clapper, 568 U.S. at 409 (an “injury must be certainly
impending”) (cleaned up); id. (a “possible future injury” is “not
sufficient”) (cleaned up).
13
Competitive Enterprise Institute, 970 F.3d at 382.
 4
that pollution comes from major changes that will go
undetected absent a stricter rule.14 Those declarations include
no firsthand knowledge or empirical evidence of major changes
that will be detected only under a stricter rule.

New Jersey’s other sources help it even less. The rule’s
preamble includes no data relevant to Article III standing. And
the relevant comments merely express non-specific,
conclusory fears about the rule’s enforceability.15 Again, they
don’t contain empirical data or firsthand accounts alleging
when, where, or whether out-of-state polluters will make major
changes that evade the permitting process.

As for the rule’s purpose of preventing malfeasance and
miscalculations, that’s not enough either. Some rules can be
prophylactic. Others are senseless. A rule’s mere existence
does not mean it is actually solving a problem that is injuring
anyone.

In contrast, in Department of Commerce v. New York —
where plaintiffs had standing because a federal policy’s
predictable effect on third parties would be an increase in the
third parties’ illegal refusal to submit census forms, thereby
injuring the plaintiffs — the district court record was

14
See Davis Decl. at ¶ 20 (“Sources in upwind states may be emitting
air pollutants at levels that would necessitate [a permit] . . . .”); see
also Wong Decl. at ¶ 6 (“New Jersey is harmed by the possibility of
unlawful emissions from sources in states upwind of New
Jersey . . . .”).
15
See Comments of the National Association of Clean Air Agencies
(May 7, 2007), at 2 (The rule “is likely to result in
diminished . . . compliance.”).
 5
voluminous.16 It included empirical study piled on empirical
study predicting with specificity — down to the tenth of a
percentage point — how many third parties would injure the
plaintiffs as a direct result of the agency’s decision to ask
census respondents whether they were citizens.17 And the
district court’s (179-page) opinion made factual findings based
on extensive evidence unlike anything in the administrative
record here.

I will quote those factual findings at length in order to
illustrate their specificity and thoroughness:

191. The Census Bureau’s conclusions are spelled out
in three memoranda. First, the Census Bureau’s
December 22 Memo summarized evidence that a
citizenship question would cause a then-estimated
5.1% decline in self-response rates among
noncitizens. See December 22 Memo, at AR 11639-
40. It noted that “this evidence is consistent with
citizenship questions being more sensitive for
household with noncitizens,” id. at AR 11640, a fact
that is not in dispute, see PX-297 at RFA 70.

192. Second, the Census Bureau’s January 19 Memo
similarly concluded that addition of a citizenship
question would reduce self-response
rates. See January 19 Memo, at AR 1280. The Memo
summarized “[t]hree distinct analyses” that “support
the conclusion of an adverse impact on self-response”
caused by the addition of a citizenship question. Id.

16
139 S. Ct. 2551; see New York v. Department of Commerce, 351
F. Supp. 3d 502, 622-25 (S.D.N.Y.), affirmed in part, reversed in
part, 139 S. Ct. 2551 (2019).
17
351 F. Supp. 3d at 578-81.
 6
First, data show that, on the ACS survey, Hispanic
households are disproportionately less likely to
respond to the citizenship question, whether
responding by mail or online. Id. Second, a
comparison of self-response rates for the 2000
census’s long-form census questionnaire (which
included a citizenship question) and its short-form
census questionnaire (which did not) revealed that
noncitizen households were 3.3% less likely than all-
citizen households to respond to the long-form
questionnaire. Id. A similar comparison of 2010
census self-response rates to 2010 ACS self-response
rates (the latter of which included a citizenship
question) produced a similar result: Noncitizen
households were 5.1% less likely than all-citizen
households to respond to the survey containing a
citizenship question. See id. Based on these
comparisons, the Memo noted, it was a “reasonable
inference that a question on citizenship would lead to
some decline in overall self-response” and “a larger
decline in self-response for noncitizen
households.” Id. at AR 1281. Finally, the Memo
analyzed the “breakoff rates” (the rate at which a
respondent stops responding to the survey when he or
she comes to a particular question) on the 2016 ACS
internet survey. Those rates indicated that Hispanics
were disproportionately likely to “breakoff” in their
responses when they came to the citizenship
question. See id.

193. Third, a comprehensive study by Census Bureau
staff published on August 6, 2018 and referred to at
trial as the Brown Memo (so named for its lead
author) consolidated the existing data on the impact of
a citizenship question. The Brown Memo also
 7
concluded that a citizenship question would
disproportionately reduce noncitizens’ self-response
rates. See Brown Memo at 1, 54. The Brown Memo
presented data illustrating that Hispanics and
noncitizens are disproportionately unlikely to respond
to a citizenship question. See id. at 7-9. The data also
showed that those subpopulations became even less
likely to respond to a citizenship question during the
middle of this decade. See id. at 9-10 (“[T]hat
sensitivity has increased in recent years.”).

194. Whereas the January 19 Memo had predicted
that addition of the citizenship question would cause
a 5.1% differential decline in noncitizen household
self-response rates, see January 19 Memo, at AR
1280, the Brown Memo updated that figure to 5.8%
on the basis of more recent data, see Brown Memo at
39. Notably, it emphasized that the 5.8% estimate was
still “conservative.” Id.; see also Tr. 900-01. It was
conservative, the Memo explained, because the
analysis supporting the estimate relied on ACS data,
and the effect of a citizenship question on the ACS
may have been muted by its presence among the large
number of questions. See Brown Memo at 39; see
also Tr. 87, 89, 901-02. A citizenship question on the
shorter 2020 census questionnaire “will be more
visible” and thus likely to produce a more pronounced
effect. Brown Memo at 39. And changes in the
macroenvironment since the ACS data was collected,
including a higher “level of concern about using
citizenship data for enforcement purposes,” could also
exacerbate the effects of adding a citizenship
question. Id.
 8
195. Separate and apart from its effects on self-
response rates among noncitizen households, the
Brown Memo supports the conclusion that adding a
citizenship question to the 2020 census will
disproportionately depress self-response rates among
Hispanic households (some, but not all, of which are
also noncitizen households). The Brown Memo
showed that Hispanics were more than twice as likely
as non-Hispanic whites to skip the citizenship
question on the ACS and that the differential in such
item nonresponse rates increased between 2013 and
2016. Id. at 8-10. Other ACS questions did not
produce the same differential effects. See id. And the
Memo found that the citizenship-question breakoff
rate for Hispanics on the ACS was eight times higher
than the breakoff rate for non-Hispanic
whites. See id. at 10; accord January 19 Memo, at
AR 1281.

196. As the Census Bureau has observed, this
differential breakoff effect is growing. The breakoff
rate among Hispanics for the 2017 ACS citizenship
question (which was not available in time to be
incorporated into the Brown Memo’s analysis)
was twelve times higher than the breakoff rate for
non-Hispanic whites. See AR 12757-62; Tr. 916.
Moreover, the breakoff rate for Hispanics, but not for
non-Hispanic whites, increased between 2016 and
2017 — suggesting that the effects of a citizenship
question on Hispanic self-responses have been
“increas[ing].” AR 12757-62; Tr. 916-17. The
Census Bureau believes that “Hispanics are more
sensitive to survey questions about citizenship than
they were a few years ago”; non-Hispanic whites “are
not.” Census Bureau 30(b)(6) Dep. 366-69.
 9

197. Defendants’ expert, Dr. Abowd, credibly
testified to the soundness of the Census Bureau’s
analyses and conclusion that adding a citizenship
question to the 2020 census would result in a
differential decline in self-response rates among
noncitizen households. With regard to methodology,
Dr. Abowd testified not only that the Brown Memo
was “methodologically appropriate,” but also that it
“constitutes the best analysis that the Census Bureau
can do of the consequences of adding the citizenship
question to the 2020 census” given the available data.
Tr. 897. With regard to conclusions, Dr. Abowd
testified that both he and the Census Bureau agreed
that adding a citizenship question to the 2020 census
would lead to a lower self-response rate among
noncitizen households. See id. at 881-82. Finally,
Dr. Abowd agreed that “[t]he bulk of the evidence
suggests that the citizenship question is likely to be
responsible for the decline in self-response,” and that
5.8% was a “conservative estimate” of the likely
differential decline in self-response rates among
noncitizen households if a citizenship question were
added to the 2020 census questionnaire. Id. at 1352,
900-02.

198. Dr. Abowd testified that considerations beyond
those mentioned in the Brown Memo further
supported the view that the 5.8% estimate was
“conservative.” See id. at 944. For instance, he
referred to the Census Bureau’s Census Barriers,
Attitudes, and Motivators Survey (“CBAMS”). See
id.; PX-662. The CBAMS found that, in 2018, only
67% of people said they were likely to respond to the
2020 census, as compared to the 86% who had said in
 10
2008 they were likely to respond to the 2010
census. See PX-662, at 12. It noted that “[t]he
citizenship question may be a major barrier” in part
because people believed that the census’s “purpose is
to find undocumented immigrants.” Id. at 43. Dr.
Abowd testified that the increase in sensitivity to a
citizenship question reflected in the CBAMS study
“would not be captured in the 5.8 percentage point
estimate that is based on data only up through 2016.”
Tr. 944-45; see also id. at 902, 916-17.

199. Testimony from at least three of Plaintiffs’
expert witnesses bolsters the Census Bureau’s and Dr.
Abowd’s conclusions about self-response rates. First,
Dr. D. Sunshine Hillygus credibly and reliably
testified that “noncitizens and Hispanics are
differentially concerned about the confidentiality of a
citizenship question” and, thus, “would be less likely
to participate” in a survey that includes such a
question. See id. at 50-51; see also id. at 57-58. She
noted that this concern has increased in the last few
years. Id. at 51-53. Notably, Dr. Hillygus testified
that a citizenship question would be likely to affect the
response rates of all Hispanics, “regardless of their
own immigration or citizenship status.” Id. at 51-52,
1404; see also PX-152; PX-662; PX-663. That
testimony is supported by evidence showing that
Hispanics who are citizens are disproportionately
hesitant to engage with the government by seeking
food stamps or health care out of fear that a family
member could be deported. See Tr. 52-54, 57, 85-86.

200. Second, Dr. Matthew Barreto, “an expert in
survey methodology, public opinion polling, and
racial and ethnic politics,” credibly testified that “the
 11
addition of a citizenship question . . . in today’s macro
environment would result in reduced participation in
Latino and immigrant communities in 2020.” Id. at
589, 620-21. He based this conclusion on a review of
existing social science literature and on the results of
a public opinion survey that he designed and
conducted. See id. at 620, 643-44. On the basis of
that evidence, Dr. Barreto credibly concluded that
Hispanic households would be substantially less
willing to participate in the census if there were a
citizenship question, regardless of whether they were
given assurances that their responses would be kept
confidential. Tr. 682-85; see also PX-670.

201. Third, Dr. Jennifer L. Van Hook’s expert
analysis of 2017 ACS data demonstrates that
nonresponse to the ACS citizenship question has
continued to increase among Hispanics relative to
other subgroups since 2013. See Docket No. 489-3
(“Van Hook Decl.”), ¶¶ 69-71. By contrast, there has
not been a significant increase in nonresponse rates
for the citizenship question for other racial
groups. See id. ¶ 70. On an absolute basis,
nonresponse rates for the citizenship question for
Hispanics have also increased since 2013. See id. ¶¶
72-73.18

If our record looked like that record, then we could say
“the predictable effect of [the EPA’s rule] on the decisions of
[polluters]” is an increase in illegal pollution significant
enough to implicate Article III standing.19 This is not to say

18
Id.
19
Competitive Enterprise Institute, 970 F.3d at 381 (quoting
Department of Commerce, 139 S. Ct. at 2566).
 12
standing requires as much evidence as in Department of
Commerce v. New York. But the difference between there and
here is the difference between a ton of apples and an ounce of
an orange.

For that reason, this case is much more like Clapper20 (no
standing) than Massachusetts v. EPA21 (standing). Just as the
alleged injury in Clapper was “too speculative to satisfy the
well-established requirement that threatened injury must be
‘certainly impending,’” New Jersey has not alleged “actual
knowledge” that, because of the challenged rule, a particular
source upwind has emitted or “certainly” will emit more
pollution than the EPA’s emissions standards allow.22 Nor has
it demonstrated a “substantial risk” of that pollution.23 It
instead relies on an “attenuated chain of possibilities”24 and “on
speculation about the unfettered choices made by independent
actors not before the court.”25

In contrast, in Massachusetts v. EPA, there was no
question about the exact sources of pollution. All cars were
injuring the state in a specific way (by emitting carbon that
reduced the state’s coastline).26 But here, because some large-
scale polluters accurately disclose their pollution without
stricter monitoring and record-keeping requirements, New
Jersey doesn’t know which (if any) out-of-state polluter is
injuring it.

20
568 U.S. at 422.
21
549 U.S. at 525.
22
Clapper, 568 U.S. at 401 (quoting Whitmore v. Arkansas, 495 U.S.
149, 158 (1990)); id. at 410-11.
23
Food & Water Watch, Inc., 808 F.3d at 914.
24
Clapper, 568 U.S. at 410.
25
Id. at 414 n.5 (cleaned up).
26
549 U.S. at 521-23.
 13
* * *

New Jersey also claims it’s injured by the EPA’s lax
record-keeping requirements for in-state polluters. According
to New Jersey, the EPA’s rule makes it harder for the state to
comply with its permitting obligations under the Clean Air
Act.27 And here again, I agree with New Jersey that if the
EPA’s rule makes its task of complying with the Clean Air Act
“more difficult and onerous,” then it has standing.28

But as with out-of-state polluters, that “if” depends on
whether in-state polluters will cheat. And New Jersey’s
prediction of in-state cheating is no less speculative than its
prediction of out-of-state cheating.29

I do not read West Virginia v. EPA to alter that analysis.
There, the EPA’s policy limited the choices available to the
petitioner states.30 Absent the EPA’s policy, those states would
have been free to adopt regulatory regimes more desirable to
them than what was possible under the EPA’s policy.31 Here,
in contrast, the EPA’s rule does not make New Jersey’s
compliance with the Clean Air Act any more difficult unless
we accept New Jersey’s speculation about polluters cheating
— which brings us back to where we started.32

To be sure, if New Jersey believes what it says — that the
EPA’s rule goes too easy on polluters — then New Jersey may
well decide down the road to impose additional requirements
on in-state polluters. But absent a record containing more
27
See 42 U.S.C. § 7410(a)(2).
28
West Virginia v. EPA, 362 F.3d 861, 868 (D.C. Cir. 2004).
29
See infra pp. 1-12.
30
362 F.3d at 868.
31
Id.
32
See infra pp. 1-12.
 14
specific evidence than what we have here, New Jersey’s
decision will be based on the state’s best guess about the effect
of those additional requirements — the type of speculation that
is often enough for legislators, but not enough for courts
applying the elements of Article III standing.

* * *

Absent a more robust record, New Jersey’s predictions of
pollution from illegal major changes traceable to the EPA’s
monitoring and recordkeeping requirements are “conjectural”
and “hypothetical.”33 The record does not show the challenged
rule’s “predictable effect” will be major changes that cause
illegal pollution.34 I respectfully disagree with the Court’s
conclusion to the contrary.

33
Lujan v. Defenders of Wildlife, 504 U.S. 555, 560 (1992).
34
Competitive Enterprise Institute, 970 F.3d at 381 (quoting
Department of Commerce, 139 S. Ct. at 2566).