Portland Cement Ass'n v. Environmental Protection Agency

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued October 11, 2011 Decided December 9, 2011

No. 10-1358

PORTLAND CEMENT ASSOCIATION,
PETITIONER

v.

ENVIRONMENTAL PROTECTION AGENCY AND LISA PEREZ
JACKSON,
RESPONDENTS

CAPE FEAR RIVER WATCH, ET AL.,
INTERVENORS

Consolidated with 10-1363, 10-1366, 10-1367, 10-1369,
10-1373, 10-1374, 10-1376, 10-1379, 11-1012, 11-1244

No. 10-1359

PORTLAND CEMENT ASSOCIATION,
PETITIONER

v.

ENVIRONMENTAL PROTECTION AGENCY AND LISA PEREZ
JACKSON,
RESPONDENTS
 2

UTILITY SOLID WASTE ACTIVITIES GROUP, ET AL.,
INTERVENORS

Consolidated with 10-1361, 10-1364, 10-1365, 10-1368,
10-1370, 10-1372, 10-1375, 10-1377, 11-1245

On Petitions for Review of a Final Action
of the Environmental Protection Agency
______

Carter G. Phillips argued the causes for Industry
Petitioners on NSPS and NESHAP Issues. With him on the
briefs were Roger R. Martella, Jr., Timothy K. Webster,
Deborah E. Jennings, Richard G. Stoll, Beth S. Ginsberg,
Jason T. Morgan, William M. Bumpers, Debra J. Jezouit,
Ashley C. Parrish, Cynthia A.M. Stroman, and Chet M.
Thompson. Howard L. Gilberg and Roger J. Marzulla entered
appearances.

David D. Doniger argued the cause for Environmental
Petitioners and State Intervenors on NSPS Issues. With him
on the briefs were Meleah Geertsma, Timothy D. Ballo,
Joanne Marie Spalding, and Craig Holt Segall.

Kamala D. Harris, Attorney General, Office of the
Attorney General for the State of California, Thomas G.
Heller, Deputy Attorney General, John Kroger, Attorney
General, Office of the Attorney General for the State of
Oregon, Paul Logan, Assistant Attorney in Charge, Oregon
Department of Justice, Robert M. McKenna, Attorney
General, Office of the Attorney General for the State of
Washington, and Leslie R. Seffern, Assistant Attorney
 3
General, were on the brief for state intervenors California, et
al. in support of Environmental Petitioners. Susan L. Durbin,
Attorney, Office of the Attorney General for the State of
California, entered an appearance.

Julie A. Weis was on the brief for amici curiae Baker
City and Baker County, Oregon, et al. in support of
petitioners.

Russell S. Frye was on the brief for amicus curiae SSM
Coalition in support of petitioners.

T. Monique Peoples and Daniel R. Dertke, Attorneys,
U.S. Department of Justice, argued the causes for respondents
on NSPS and NESHAP Issues. With them on the briefs were
Steven Silverman and Elliott Zenick, Counsel, U.S.
Environmental Protection Agency.

Carter G. Phillips argued the cause for Industry
Intervenors in support of respondents. With him on the briefs
were Roger R. Martella, Jr., Timothy K. Webster, Deborah E.
Jennings, Richard G. Stoll, Beth S. Ginsberg, Jason T.
Morgan, William M. Bumpers, Debra J. Jezouit, Norman W.
Fichthorn, Allison D. Wood, Ashley C. Parrish, Cynthia A.M.
Stroman, and Chet M. Thompson. Douglas H. Green, Aaron
J. Wallisch, and Willian R. Weissman entered appearances.

Timothy D. Ballo, Geoffrey R. Gisler, James S. Pew, Seth
L. Johnson, Avinash Kar, Meleah Geertsma, and John D.
Walke were on the briefs of Environmental Intervenors in
support of respondents. John T. Suttles, Jr. and David D.
Doniger entered appearances.

Before: HENDERSON, TATEL and BROWN, Circuit Judges.
 4
Opinion for the Court filed PER CURIAM.

Concurring opinion filed by Circuit Judge BROWN.

PER CURIAM: Pursuant to the Clean Air Act (“CAA”), the
Environmental Protection Agency enacted twin rules in 2010
setting emissions standards for portland cement facilities—
one under a section called National Emission Standards for
Hazardous Air Pollutants (NESHAP), 42 U.S.C. § 7412(a)(4),
the second under a section called New Source Performance
Standards (NSPS), id. § 7411. Petitioners, Portland Cement
Association and other cement manufacturers (“PCA”), argue
that both rules violate the CAA and are arbitrary and
capricious. A consortium of environmental groups including
the Sierra Club (“Environmental Petitioners”) filed their own
petition, arguing that EPA abused its discretion by declining
to include greenhouse gas emissions standards in its NSPS
rule.

For the reasons set forth below, we agree that EPA acted
arbitrarily when it promulgated the final NESHAP rule and
therefore grant PCA’s petition for review with respect to
EPA’s denial of reconsideration on that issue. We also stay
the NESHAP standards for clinker storage piles pending
reconsideration by EPA. We deny PCA’s petitions with
respect to all other issues relating to NESHAP and NSPS, and
dismiss Environmental Petitioners’ petition for lack of
jurisdiction.

I

Portland cement, a fine gray powder used to make
construction-grade concrete, is produced by combining raw
materials in a kiln and heating the mixture to produce a
substance called “clinker,” which is then cooled and ground
 5
into powder. This kiln firing process causes the airborne
emission of particulate matter (“PM”), as well as a number of
other dangerous chemicals. Once produced, the clinker is
stored in piles which may also continue to emit some
hazardous chemicals.

There are three basic types of portland cement kilns. The
first, called “long wet” or “long dry” process kilns, are the
least efficient. These kilns, which tend to be older, simply
heat raw materials as they pass through a large rotating
cylinder. The second type of kiln, called a “preheater,” is
more modern and efficient. Preheater kilns preheat the raw
materials by first passing them through a tower filled with hot
exhaust gases. Finally, the most modern and efficient kilns,
preheater/precalciner kilns, are equipped with both preheater
towers and a combustion vessel which heats raw materials at a
high temperature before they reach the core of the kiln,
removing moisture and undesirable compounds. Ultimately,
the type of kiln directly affects the amount of uncontrolled
pollutants emitted. For example, long wet and long dry
process kilns emit between eight to ten times the amount of
sulfur dioxide as preheater/precalciner kilns.

Two separate sections of the CAA, 42 U.S.C. § 7401 et
seq., require EPA to promulgate emissions standards for
“stationary sources” of pollution such as cement kilns. The
first, NESHAP, requires EPA to set emissions standards for
both new and existing sources. Id. § 7412(a)(4); (a)(10). The
second, NSPS, requires EPA to set emissions standards for
new and newly-modified sources. Id. § 7411. (A “modified”
source, for the purposes of the CAA, is a source that has
undergone a physical or operational change “which increases
the amount of any air pollutant emitted by such source, or
which results in the emission of any air pollutant not
previously emitted.” Id. § 7411(a)(4)). Thus, although
 6
NESHAP and NSPS overlap as to regulation of new sources,
NESHAP alone governs the regulation of existing sources,
and NSPS alone governs the regulation of modified sources.

Pursuant to CAA Section 112, EPA sets NESHAP
emissions limits in a two-stage process. First, EPA sets what
it calls a “floor.” For new sources, the floor is equal to the
amount of emissions reduction “achieved in practice by the
best controlled similar source.” 42 U.S.C. § 7412(d)(3). For
existing sources, the floor equals the amount of emissions
reduction “achieved [on average] by the best performing 12
percent of the existing sources (for which the Administrator
has emissions information)” in the source category. Id. If the
category contains fewer than 30 sources, the floor is to be set
based on the amount of emissions reduction achieved by the
best performing five sources for which the Administrator has
emissions information. Id. NESHAP emissions standards
“shall not be less stringent than” this floor. Id. Second, EPA
may go “beyond-the-floor” and set a more stringent standard
if, taking cost and other factors into account, it determines
that such a standard would be “achievable.” Id. § 7412(d)(2);
see also Cement Kiln Recycling Coalition v. EPA, 255 F.3d
855, 857–58 (D.C. Cir. 2001) (describing the two-step
regulatory framework and noting that “floors” apply “without
regard to either costs or . . . other factors,” but that EPA may
set limits “beyond-the-floor” if it takes cost and other factors
into account). The promulgated NESHAP standard is known
as the “maximum achievable control technology” or
“MACT.”

Under NSPS, however, EPA is required to set standards
for emissions that “reflect[] the degree of emission limitation
achievable through the application of the best system of
emission reduction.” 42 U.S.C. § 7411(a)(1). In contrast to
NESHAP’s two-stage process, under which EPA is prohibited
 7
from considering cost, achievability, or countervailing
considerations at step one, NSPS requires EPA take into
account the “cost of achieving” emissions reductions, as well
as health, environmental, and energy considerations. Id. §
7411(a)(1).

In June 2008, EPA initiated two rulemaking procedures
to revise emissions standards for the portland cement
industry: one under NESHAP and one under NSPS.
Following a comment period, these rules were finalized in
September 2010. In the NESHAP rule, EPA set standards for
new sources and existing sources for emissions of PM,
mercury, hydrochloric acid, and hydrocarbons. EPA did not
go “beyond-the-floor,” so these standards are instead equal to
the respective floors. Because the rulemaking took place
entirely at the first NESHAP step, EPA did not—because it
could not at that step—take into account cost or other
considerations.

In the NSPS rule, EPA, for the first time, set standards
for both new and modified sources for emissions of nitrogen
oxide and sulfur dioxide. In addition, EPA revised its existing
NSPS emissions standard for PM, setting a limit of 0.01
pounds of PM emitted per ton of clinker produced. EPA
concluded that this revised PM standard was achievable if
kilns installed a particular type of pollution control
technology: fabric filters with membrane bags.

EPA declined to include emissions standards for carbon
dioxide or other greenhouse gases in its final NSPS rule.
Explaining its decision to omit such standards, EPA noted that
because it had proposed no specific emissions standard for
greenhouse gases in the proposed regulations, “promulgating
such a standard without providing opportunity to comment on
it would . . . violate the norms of notice and comment
 8
rulemaking.” 75 Fed. Reg. 54,970, 54,996 (Sept. 9, 2010).
Moreover, although EPA’s “preliminary evaluation” indicated
“it may be appropriate for the Agency to set a standard of
performance for greenhouse gases,” EPA determined that it
did “not yet have adequate information about greenhouse gas
emissions to set a standard.” Id. at 54,996–97. EPA then
identified specific types of pertinent information it was
lacking, such as information about greenhouse gas emissions
from cement plants and site-specific factors that could affect
the performance of emissions controls. Id. at 54,997. EPA
concluded by stating that it was “working towards a proposal
for greenhouse gas standards,” which it would promulgate
after receiving additional information from cement facilities.
Id.

PCA sought administrative reconsideration of both the
NESHAP and NSPS rules. EPA denied PCA’s petitions for
reconsideration on all but two issues. First, EPA granted
PCA’s petition for reconsideration of emissions regulations
for outdoor clinker storage piles, holding that “petitioners are
correct that the Agency did not give sufficient notice of what
[clinker storage pile] standards might be.” 76 Fed. Reg.
28,318, 28,325 (May 17, 2011). Second, EPA granted PCA’s
petition for reconsideration of the NSPS PM emissions
standards for modified sources. Although PCA asked EPA to
stay both standards pending reconsideration, EPA declined to
do so.

PCA subsequently filed the instant petitions for review of
both the rules themselves and EPA’s denials of
reconsideration. Environmental Petitioners filed their own
petition challenging EPA’s decision not to include greenhouse
gas emissions standards in its final NSPS rule. PCA
intervened on behalf of EPA on this issue.
 9
II

While EPA was establishing the NESHAP standards at
issue in this case, it was simultaneously developing a
definition of commercial and industrial solid waste
incinerators (“CISWI”). This definition would create a
separate category of pollutant sources subject to emissions
standards distinct from NESHAP. This rulemaking process,
which EPA described as “relevant” to NESHAP, would
impact the NESHAP rulemaking because some cement kilns
“combust secondary materials [like solid waste] as alternative
fuels.” Such kilns would be subject to standards under the
CISWI rules rather than under the NESHAP rules, since the
two regimes are mutually exclusive. 74 Fed. Reg. 21,136,
21,138 (May 6, 2009); see also 42 U.S.C. § 7429(h)(2)
(requiring exclusivity); Natural Res. Def. Council v. EPA, 489
F.3d 1250, 1260–61 (D.C. Cir. 2007) (same). EPA proposed
the CISWI definition ten months after the close of the
NESHAP comment period but three months before the final
NESHAP rule was issued. The CISWI definition was enacted
six months after the NESHAP rule became final.

PCA argues that EPA improperly ignored this ongoing
CISWI process when it set the NESHAP standards. EPA
realized the CISWI definition could potentially impact the
NESHAP rule, since under the proposed definition, EPA
could reclassify close to a third of all cement kilns out of
NESHAP and into CISWI.1 (In fact, PCA notes that some of
the best performing sources central to the setting of the
NESHAP floor are excised from the NESHAP universe

1
The final CISWI definition ultimately reclassified fewer sources.
PCA and EPA disagree as to the number: EPA’s most recent
estimate is that closer to 16 percent were reclassified. 76 Fed. Reg.
28,318, 28,322 (May 17, 2011).
 10
altogether under the new CISWI rule.) But EPA was
unconcerned that its NESHAP floor-setting calculations might
include sources that actually would not be subject to the
NESHAP standard once the rules were completed. Instead of
treating the two rules as truly interdependent efforts and
acknowledging their close correlation, EPA let each run its
own course regardless of the collateral impact. PCA argues
that it both violated the CAA and was arbitrary and capricious
for EPA to have set the NESHAP standard on the premise that
all kilns would be subject to NESHAP while at the same time
modifying the dataset to change that premise. We agree it is
arbitrary and capricious.2

Before reaching the merits, we must decide whether we
have jurisdiction. PCA cannot challenge the rule directly.
Before an objection can be raised in this Court, it must be
“raised with reasonable specificity during the period for
public comment.” 42 U.S.C. § 7607(d)(7)(B). PCA did not
comment on this issue in the NESHAP rulemaking, so even if
we agreed with PCA on the merits, we could not vacate the
NESHAP rule. However, “[i]f the person raising an objection
can demonstrate . . . that it was impracticable to raise [an]
objection within [the comment period] or if the grounds for

2
EPA did not violate the provision of the CAA stating that “no
solid waste incineration unit subject to performance standards under
[CISWI] shall be subject to standards under [NESHAP],” 42 U.S.C.
§ 7429(h)(2), because no unit has actually been subjected to both
standards. The provisions requiring EPA to set NESHAP standards
based on emissions reductions achieved by similar sources within
the same NESHAP category, id. § 7412(d)(1), (d)(3), on the other
hand, come closer to being implicated here. But at the time EPA
set the NESHAP standards, all of the sources it examined were
within the same category. While we find EPA’s ostrich-like
approach to its recategorization efforts was arbitrary, it did not
violate the text of the CAA.
 11
[the] objection arose after the period for public comment . . . ,
the Administrator shall convene a proceeding for
reconsideration of the rule.” Id. If reconsideration is denied,
review of the Administrator’s refusal is available “in the
United States court of appeals for the appropriate circuit.” Id.
Although it is a very close question, we are satisfied PCA
could not have reasonably anticipated the extent to which
EPA would base the final NESHAP standard on data from
kilns it would soon reclassify into a different—and mutually
exclusive—regulatory regime. Because EPA refused to
reconsider the rule, we have jurisdiction to review that
refusal.

In its proposed rule, EPA acknowledged the CISWI
rulemaking was ongoing and noted some unknown number of
kilns might ultimately be classified as CISWI sources.
Because EPA did not yet know what shape the CISWI rule
would take, however, EPA said it would continue to assume
no kilns were CISWI sources “until the solid waste definition
. . . is promulgated.” 74 Fed. Reg. 21,136, 21,138 (May 6,
2009). These statements left open a couple of possibilities.
First, the re-sorting of some kilns into CISWI was, while
likely, not inevitable. Id. (“EPA therefore cannot reliably
determine at this time if the secondary materials combusted
by cement kilns are to be classified as solid wastes.”); see also
Office of Air Quality Planning and Standards, Development
of the MACT Floors for the Proposed NESHAP for Portland
Cement 4 (Apr. 15, 2009) (“Pending the outcome of other
rulemakings, there is a possibility that some of the kilns
currently in the Portland Cement NESHAP source category
will at some point become subject to the [CISWI] regulation,
and thus no longer subject to this regulation.”) (emphases
added). Second, EPA’s conditional language—“until the
solid waste definition . . . is promulgated”—suggested that,
 12
should any kilns ultimately fall under the CISWI definition,
EPA would adjust the NESHAP rule accordingly.

While we certainly require some degree of foresight on
the part of commenters, we do not require telepathy. We
should be especially reluctant to require advocates for
affected industries and groups to anticipate every
contingency. To hold otherwise would encourage strategic
vagueness on the part of agencies and overly defensive,
excessive commentary on the part of interested parties
seeking to preserve all possible options for appeal. Neither
response well serves the administrative process. Whatever
warning EPA offered regarding CISWI was too vague and
noncommittal to trigger a response from PCA. Indeed, as far
as EPA did hint at its next steps, it suggested it would
reevaluate the NESHAP standards after the CISWI definition
was promulgated.

Having determined that PCA is not jurisdictionally barred
from petitioning EPA for reconsideration and that it may
therefore seek review in this Court of EPA’s denial, we
proceed to the merits of its objection. In none of EPA’s
proposals, final rules, or briefs in this Court has EPA
attempted to defend the principle that, in the face of a final
and promulgated CISWI definition, data from CISWI kilns
could now be considered in setting NESHAP standards. And
rightly so: it would certainly be arbitrary, as well as a
violation of the CAA itself, for EPA to set one standard based
on data already placed in another source category in light of
the mutual exclusivity of the standards themselves. See North
Carolina v. EPA, 531 F.3d 896, 930 (D.C. Cir. 2008) (noting
that it is “entirely arbitrary” to base standards on “irrelevant
factors”); 42 U.S.C. § 7412(d)(1), (d)(3) (requiring EPA to set
NESHAP standards based on emissions reductions achieved
by similar sources within the same NESHAP category).
 13

EPA instead defends the otherwise indefensible by
claiming the unique circumstances of this parallel rulemaking
left no other choice. As EPA said in the final NESHAP rule,
because it “cannot prejudge the outcome of the recently
proposed [CISWI] rulemaking” and because it could only
“bas[e] all determinations as to source classification on the
emissions information now available,” it simply had to
include “all portland cement kilns as . . . subject to regulation
under [NESHAP].” 75 Fed. Reg. 54,970, 54,972 (Sept. 9,
2010). EPA made the same argument here: a substantial
number of sources subject to being reclassified as CISWI
sources were included in the NESHAP calculation because
EPA had not yet decided the precise parameters of its
definition.

Basing its decision on a premise the agency itself has
already planned to disrupt is arbitrary and capricious.
Reasoned decisionmaking requires an agency to “examine the
relevant data and articulate a satisfactory explanation for its
action[s].” Motor Vehicles Mfrs. Ass’n v. State Farm Mut.
Auto. Ins. Co., 463 U.S. 29, 43 (1983). The impending
definition of an undeniably related source category is clearly a
“relevant factor[]” or an “important aspect of the problem”
that must be considered. Id. Indeed, EPA stated as much in
its proposed rule, describing the CISWI rulemaking as
“relevant” to the NESHAP proceeding. 74 Fed. Reg. 21,136,
21,138 (May 6, 2009).

Since agencies “have an obligation to deal with newly
acquired evidence in some reasonable fashion,” Catawba
Cnty. v. EPA, 571 F.3d 20, 45 (D.C. Cir. 2009), or to
“reexamine” their approaches “if a significant factual
predicate” changes, Bechtel v. FCC, 957 F.2d 873, 881 (D.C.
Cir. 1992), an agency must have a similar obligation to
 14
acknowledge and account for a changed regulatory posture
the agency creates—especially when the change impacts a
contemporaneous and closely related rulemaking. See Office
of Commc’n of United Church of Christ v. FCC, 707 F.2d
1413, 1441–42 (D.C. Cir. 1983) (finding it “seriously
disturbing” and “almost beyond belief” that an agency would
take rulemaking action undercutting another “concurrent”
rulemaking process); see also Gen. Chem. Corp. v. United
States, 817 F.2d 844, 846 (D.C. Cir. 1987) (finding agency
action arbitrary and capricious because it was “internally
inconsistent and inadequately explained”). All EPA did to
satisfy this obligation with regard to the NESHAP rule was
decide that it would do nothing “so long as no final definition
of solid waste changed [the] status [of cement kilns] prior to
promulgation of the NESHAP.” EPA Br. 25 (emphasis
added). This is not a “satisfactory explanation,” State Farm,
463 U.S. at 43, or a “hard look at the salient problems,”
Panhandle E. Pipe Line Co. v. FERC, 890 F.2d 435, 439
(D.C. Cir. 1989). It is nothing more than a determination that
EPA would not address the problem unless it happened to
appear at an inconvenient time—an eventuality over which
EPA had full control. The refrain that EPA must promulgate
rules based on the information it currently possesses simply
cannot excuse its reliance on that information when its own
process is about to render it irrelevant.

EPA makes two arguments in response, neither of which
addresses this basic principle. First, EPA insists that it would
be absurd to require revised calculations every time the
content of a source category changes, i.e., when a source
closes or a new source is built. EPA Br. 26. But no such
absurdity is involved here. No actions by the regulated
community changed the dataset relevant to EPA’s
calculations; EPA’s definition did that. And EPA undermined
the premise of its calculations at the same time it was
 15
enshrining those calculations in a final rule—without
accounting for the impact of the change. It is not absurd to
require that an agency’s right hand take account of what its
left hand is doing. In fact, this is nothing new: this Court has
required EPA to recalculate standards because of changes to
category definitions when “the populations of units subject to
[exclusive] rules will change substantially.” Natural Res.
Def. Council, 489 F.3d at 1261.

Second, EPA asserts it could not delay finalizing the
NESHAP rule until after it promulgated a definition of solid
waste. EPA insists such a delay would have been harmful to
air quality and health. EPA Br. 27. Perhaps. But reasoned
decisionmaking is not a dispensable part of the administrative
machine that can be blithely discarded even in pursuit of a
laudable regulatory goal. “The importance of reasoned
decisionmaking in an agency action cannot be over-
emphasized. When an agency . . . is vested with discretion to
impose restrictions on an entity’s freedom to conduct its
business, the agency must exercise that discretion in a well-
reasoned, consistent, and evenhanded manner.” Greyhound
Corp. v. ICC, 668 F.2d 1354, 1359 (D.C. Cir. 1981). EPA
also notes that it was facing a NESHAP deadline pursuant to a
settlement agreement with PCA, but of course, it could have
begun the CISWI process much sooner. After all, EPA had
been working on the NESHAP rule for ten years, and so it
should have come as no surprise that the CISWI definition
would play a critical role in setting that standard. It takes a
certain amount of chutzpah for EPA to claim it had no time to
be careful—after ten years of work on NESHAP—when it
waited to propose a CISWI definition until after the NESHAP
comment period had closed. It takes even more chutzpah to
repeat that claim after the district court has already called the
argument “silly” in a closely analogous context. Am.
Petroleum Inst. v. Johnson, 541 F. Supp. 2d 165, 185 (D.D.C.
 16
2008) (“The fact that the proposed rule had been on the shelf
for ten years is no excuse for failing to consider a directly
relevant” intervening legal change “decided before the final
rule was promulgated.”).

Simply put, there was no CISWI definition when the
NESHAP rule was finalized because, even though EPA knew
it would be critical to the NESHAP process, EPA did not even
propose it until after the comment period for the NESHAP
rules had closed. Far from justifying EPA’s conduct, the
unique circumstances of this parallel rulemaking are what
doom it: when an agency is simultaneously in control of both
defining the universe of relevant data and of applying that
data to a given rulemaking, it cannot allow itself to do the
latter without having already done the former. If an agency
can say its failure to decide what data are relevant justifies its
decision to just consider all data, arbitrary and capricious
review would be pointless. EPA has put the cart before the
horse, and there is no justification, least of all an agency’s
own timing choices, for such a cavalier and unscientific
attitude.

EPA points out that the final CISWI definition—
promulgated a mere six months after NESHAP—resulted in
about 23 kilns being reclassified, and that removing these
kilns from the NESHAP calculations does little to relax the
ultimate standards. In fact, one emissions standard would
even become more stringent after removing the CISWI kilns
from the data set. We have no reason to doubt that
conclusion; perhaps PCA would be better off had they not
brought this issue to our attention. But we are not interested
in whether the rule becomes more or less stringent upon
reconsideration. Our province is simply to ensure that
agencies do not act arbitrarily or capriciously, 5 U.S.C. §
 17
706(2)(A), and the magnitude or direction of the effect of an
agency’s arbitrariness does not excuse it.

PCA also argues EPA violated the CAA when it
premised the NESHAP standards on bare emissions data
rather than on data that specifically isolated the effect of
technology by controlling for variations in input quality. We
considered and rejected this very argument quite recently. In
Sierra Club v. EPA, 479 F.3d 875, 883 (D.C. Cir. 2007), we
declined to read the “the Clean Air Act’s command that it
assess the emission ‘control’ or ‘limitation’ ‘achieved’ [as]
refer[ring] to the deliberate steps kiln operators take to reduce
emissions rather than to the ‘happenstance’ of being located
near cleaner clay.” Instead, we held that EPA must do exactly
what it did here. Id. PCA’s attempt to dismiss this holding as
dicta is unavailing since the question of how to account for
raw material quality—in that case, “clay type”—was squarely
presented. Id. at 882–83.

We have carefully considered PCA’s other objections to
the NESHAP rule and are unconvinced by them. PCA’s
argument that EPA’s pollutant-by-pollutant approach to
setting NESHAP floors violates the CAA is barred because it
was not raised within sixty days of EPA’s first use of that
approach, Medical Waste Inst. v. EPA, 645 F.3d 420, 427
(D.C. Cir. 2011), and their argument that EPA impermissibly
reset NESHAP floors rather than revise existing floors is
based on a flawed reading of the CAA. Though EPA must
review and revise standards “no less often than every eight
years,” 42 U.S.C. § 7412(d)(6), nothing prohibits EPA from
reassessing its standards more often.

PCA also argues that the adoption of a continuously-
monitored standard (“CEMS”) rather than a sampling
standard for particulate matter emissions was not a “logical
 18
outgrowth” of the proposed rule, Small Refiner Lead Phase-
Down Task Force v. EPA, 705 F.2d 506, 543 (D.C. Cir.
1983), but this is not true. EPA sought comment on a CEMS
requirement in its first proposal, and PCA even commented
on it. 74 Fed. Reg. 21,136, 21,157 (May 6, 2009). Moreover,
any individual hardship resulting from the CEMS requirement
is mitigated by the fact that a kiln may employ “alternative
monitoring” if it demonstrates the “technical or economic
infeasibility” of installing CEMS. 40 C.F.R. § 63.8(f)(4)(ii).
Similarly, the proposed NESHAP rule provided adequate
notice that EPA was considering modifying emissions
standards for startup and shutdown periods, and PCA
commented on that as well. 74 Fed. Reg. 21,136, 21,162
(May 6, 2009). PCA’s final claims of arbitrariness also fail
since EPA adequately explained its reasons for, among other
things, not setting separate hydrocarbon standards for raw
material dryers. Nothing in the CAA or our caselaw requires
EPA to collect additional data before making that decision.
See Sierra Club v. EPA, 167 F.3d 548, 662 (D.C. Cir. 1999)
(noting that EPA “typically has wide latitude in determining
the extent of data-gathering necessary to solve a problem”).

Because EPA’s treatment of the CISWI-NESHAP
interaction was arbitrary and capricious, we grant the petition
for review with respect to EPA’s denial of reconsideration,
and remand for further action consistent with this decision.
We decline to stay the NESHAP rule pending reconsideration.
The CAA does not mandate a stay, 42 U.S.C. §
7607(d)(7)(B), and because it is unlikely that significant
changes will be made to the standards upon reconsideration,
we see little chance of PCA suffering irreparable harm.

We will, however, enter a stay of the NESHAP standards
applicable to clinker storage piles. EPA has conceded that it
“did not give sufficient notice” of those standards and has
 19
granted PCA’s request for reconsideration, but it denied
PCA’s request for a stay. 76 Fed. Reg. 28,318, 28,325–26
(May 17, 2011). Because EPA will now be receiving
comments for the first time, the standards could likely change
substantially. Thus, industry should not have to build
expensive new containment structures until the standard is
finally determined.

III

Turning to PCA’s challenge to the NSPS rulemaking, we
begin by addressing its contention that for all regulated
pollutants, EPA failed to “consider . . . the range of relevant
variables that may affect emissions in different plants.” PCA
Opening Br. 33 (quoting Nat’l Lime Ass’n v. EPA, 627 F.2d
416, 433 (D.C. Cir. 1980)). PCA argues that EPA failed
adequately to consider the impact of its NSPS standards on
kilns of older design that, if modified, could become subject
to NSPS. Instead, PCA argues that EPA illegitimately
focused solely on kilns with preheater/precalciner technology.

This argument fails on its own terms because contrary to
PCA’s contention, EPA demonstrated how all regulated kilns
could meet NSPS standards. EPA based its PM and sulfur
dioxide limits “on control technologies that can be applied to
any kiln type and achieve the same control levels that would
be expected with a new kiln at similar costs.” 75 Fed. Reg.
54,970, 54,995–96 (Sept. 9, 2010) (emphasis added). PCA
nowhere even attempts to dispute this point. As to nitrous
oxide, EPA did note that it would be more difficult for older
kilns to meet its final emissions limits, and indeed
“investigated whether [it] should set a different [nitrogen
oxide] emissions limit for modified kilns.” Id. at 54,996. But
based on detailed studies, EPA ultimately determined that
older kilns could avoid increasing their nitrogen oxide
 20
emissions—and thus, remain in compliance with NSPS
standards—by utilizing a variety of different controls. Id.
See ASARCO, Inc. v. EPA, 578 F.2d 319, 328–29 (D.C. Cir.
1978) (“[T]he operator of an existing facility can make any
alterations he wishes in the facility without becoming subject
to the NSPS as long as the level of emissions from the altered
facility does not increase. . . . The record does not indicate
why more flexibility than this is necessary or even
appropriate.”) (emphasis added). We thus reject as unfounded
PCA’s contention that EPA failed to consider the effects of its
standards on older kilns.

It is true, as PCA notes, that EPA expected the NSPS
limits would primarily apply to preheater/precalciner kilns
and focused its rulemaking accordingly—for example, by
using data primarily derived from preheater/precalciner kilns.
See 73 Fed. Reg. 34,072, 34,075 (June 16, 2008) (“EPA
believes that the limits proposed today are appropriate for
new, modified, and reconstructed kilns since the
preheater/precalciner design will be utilized in each of these
instances.”). But this was an eminently reasonable decision
based on the facts EPA had before it. As EPA explained,
industry statistics show that virtually all older kilns are being
replaced by newer preheater/precalciner units. Id. Indeed,
during the past 20 years only two long wet or long dry kilns
were modified, rather than replaced, and both were modified
to include preheater/precalciner technology. Id. At its core,
then, PCA’s argument is that EPA abused its discretion by
failing adequately to consider the effects of its standards on an
entirely conjectural species of kiln: a newly modified long
wet or long dry kiln without preheater/precalciner technology.
But given the universal movement in the portland cement
industry towards adoption of preheater/precalciner
technology, we have no basis for concluding that EPA’s
decision to focus primarily—but not exclusively—on
 21
preheater/precalciner kilns was arbitrary or capricious. 42
U.S.C. § 7607(d)(9)(A).

We next turn to PCA’s various challenges to the final PM
limits. PCA first argues that in promulgating the PM NSPS
EPA improperly “incorporated the entirely distinct new
source PM limit from the NESHAP rulemaking in lieu of
undertaking the analysis and considering the factors required
by [NSPS].” PCA Opening Br. 20. In particular, PCA
contends that EPA adopted the 0.01 pound/ton NESHAP PM
standard as the NSPS PM standard without “consideration of
the cost or other non-air impacts that [NSPS] requires.” PCA
Opening Br. 22.

This assertion is incorrect. Although both the NSPS and
NESHAP rulemaking resulted in a PM emissions limit of 0.01
pounds per ton, EPA arrived at that limit using two different
mechanisms. Under NESHAP, EPA set the PM emissions
limit at 0.01 pounds per ton because that was the level
achieved by the best-performing existing source. 75 Fed.
Reg. 54,970, 54,987 (Sept. 9, 2010). By contrast, under
NSPS, EPA determined that the “best system of emission
reduction,” 42 U.S.C. § 7411(a)(1), for PM was “well-
operated and maintained fabric filters” with membrane bags.
73 Fed. Reg. 34,072, 34,076–77 (June 16, 2008). Expressly
considering the cost of these filters, EPA’s proposed NSPS
rule determined the technology was “well within the range of
cost-effectiveness . . . accepted as reasonable for other [non
cement kiln] stationary sources.” Id. at 34,077. And once
EPA determined that fabric filter technology could result in
greater emissions reductions than previously thought, its final
rule stated the self-evident proposition that fabric filter
“technology would now be evaluated as more cost-effective
than at proposal, since greater PM reductions will result from
its use.” 75 Fed. Reg. 54,970, 54,995 (Sept. 9, 2010)
 22
(emphasis added). To be sure, the final rule also noted that
kilns would have to install fabric filter technology to comply
with NESHAP, concluding that the parallel NSPS rule would
therefore have no additional cost. Id. But this statement
hardly means that EPA “adopted” the NESHAP standards,
nor does it somehow invalidate EPA’s earlier cost analysis of
fabric filter technology. We therefore reject PCA’s
contention that EPA failed to consider cost when
promulgating its NSPS standard. Nor do we find merit in
PCA’s novel argument—unsupported by any authority—that
EPA was required to “reanalyze costs . . . in promulgating the
final PM limit,” and thus improperly relied on the cost
analysis it had previously conducted in the proposed rule.
PCA Reply Br. 5. Neither law nor logic requires EPA to
spend its time and resources conducting a perfunctory cost
analysis when doing so would duplicate information the
agency already has before it.

Similarly, we have little trouble rejecting PCA’s
argument that EPA failed to consider the other countervailing
factors required by NSPS: “nonair quality health[,]
environmental impact and energy,” 42 U.S.C. § 7411(a)(1).
EPA’s final order included sections surveying the PM
standard’s 1) water quality impact, 2) solid waste impact, 3)
secondary environmental impacts, 4) energy impacts, and 5)
cost impacts. 75 Fed. Reg. 54,970, 55,022–23 (Sept. 9,
2010). Although PCA correctly notes that these sections are
commingled with discussion of various NESHAP regulations,
nothing in the Clean Air Act requires a segmented discussion
of NSPS factors. Instead, the statute requires only that EPA
“tak[e] into account” health, environmental, and energy
impacts. 42 U.S.C. § 7411(a)(1). The final order’s discussion
of these factors shows that EPA did just that. The fact that the
final order also discussed the health, environmental, and
energy impacts of NESHAP regulations is immaterial.
 23

Next, we address PCA’s claims that we should vacate the
PM standards “because EPA did not give . . . notice of its
methodology for setting the NSPS limit, and because the final
PM limit is not a logical outgrowth of the one EPA
proposed.” PCA Opening Br. 29. This notice-based
argument rests primarily on the premise that EPA set NSPS
standards by “incorporat[ing] . . . the new source NESHAP
limit for PM as the NSPS.” PCA Opening Br. 26. Having
already rejected that argument, we have little difficulty
rejecting PCA’s parallel claims that EPA “never provided
notice that it would adopt the new source NESHAP limit for
PM as the NSPS limit.” PCA Opening Br. at 30.

But we do see merit in one of PCA’s notice-based claims:
that EPA failed to provide notice that it would require
continuous monitoring of PM emissions from cement kilns.
EPA proposed requiring kilns to demonstrate compliance with
the PM standard by conducting periodic stack tests. The only
mention of continuous monitoring in the proposed rule came
when EPA proposed providing an “option” for plants to
demonstrate compliance with the PM standard by installing a
CEMS. 73 Fed. Reg. 34,072, 34,082 (June 16, 2008). In its
final NSPS rule, however, EPA required plants to demonstrate
compliance with the standard through continuous emissions
monitoring. PCA contends that this “change in limit and
fundamental approach” contravenes this court’s directive that
a proposed rule must “describe the range of alternatives being
considered with reasonable specificity.” PCA Opening Br. 29
(quoting Horsehead Res. Dev. Inc. v. Browner, 16 F.3d 1246,
1268 (D.C. Cir. 1994) (per curiam) (quotation omitted)). We
agree. The fact that EPA proposed providing kilns with a
CEMS option hardly placed PCA on notice that kilns could be
required to demonstrate NSPS compliance through continuous
emissions monitoring. On this point, we find it instructive to
 24
compare EPA’s proposed NSPS rule to its proposed NESHAP
rule. As here, EPA’s final NESHAP rule required kilns to
demonstrate compliance with a PM standard through periodic
stack tests. But unlike here, the proposed NESHAP rule
expressly invited comment on whether to require CEMS
monitoring. 74 Fed. Reg. 21,136, 21,157 (May 6, 2009)
(“[W]e are specifically soliciting comment on making the use
of a PM CEMS a requirement.”) (emphasis added).

But although EPA gave inadequate notice that it might
adopt a CEMS requirement under NSPS, this error was
harmless precisely because the proposed NESHAP rule put
PCA on notice that EPA might require kilns to install CEMS
systems. During NESHAP rulemaking, PCA commented on
the propriety of requiring CEMS. In response, EPA made
changes to the way in which CEMS limits were calculated
from raw stack test data for both NESHAP and NSPS rules.
75 Fed. Reg. 54,970, 54,975 (Sept. 9, 2010). Thus, PCA had
an opportunity to comment on the potential for a required
CEMS system—and did so. We “may invalidate [a] rule” for
“alleged procedural errors” only if “there is a substantial
likelihood that the rule would have been significantly changed
if such errors had not been made.” 42 U.S.C. § 7607(d)(8).
PCA does not argue that repeating the comments it made in
response to the proposed NESHAP rulemaking would have
resulted in a “substantial likelihood” that NSPS standards
would have “significantly changed,” and we fail to see how
this could have been the case. Given EPA’s harmless
procedural error, we thus have no basis for invalidating the
NSPS standard.

Finally, PCA contends that having granted
reconsideration on the final PM standard as applied to
modified sources, EPA abused its discretion by refusing to
stay the implementation of that standard. But because the
 25
fabric filter cost analysis EPA conducted—which applied to
both new and modified kilns—was sufficient to support the
0.01 lb/ton PM standard, supra p. 22, we think it entirely
unlikely that EPA will impose a different standard for
modified sources on reconsideration. As a result, and as with
the NESHAP standards, we see little chance that PCA will
suffer irreparable harm. We therefore deny PCA’s request for
a stay. See 42 U.S.C. § 7607(d)(7)(B).

IV

This brings us to Environmental Petitioners’ challenge to
EPA’s failure to adopt greenhouse gas emissions standards as
part of its portland cement NSPS. We agree with PCA
intervenors that we lack jurisdiction to hear this challenge.
The Clean Air Act gives us jurisdiction to review only “final”
agency actions, 42 U.S.C. § 7607(b), and there was nothing
“final” in EPA’s decision to collect additional information
before proposing greenhouse emissions standards.

EPA’s final NSPS rule states that: 1) EPA did “not yet
have adequate information about [greenhouse gas] emissions
sufficient to set a standard,” but 2) “based on our initial
evaluation it appears that there are cost-effective control
strategies . . . that would provide an appropriate basis for
establishing a standard of performance for [greenhouse gas]
emissions.” 75 Fed. Reg. 54,970, 54,996–97 (Sept. 9, 2010).
EPA, the rule explains, “is working towards a proposal for
[greenhouse gas] standards from Portland cement facilities”—
a proposal it will promulgate after it receives data necessary
“to develop proposed standards.” Id. at 54,997. We fail to
understand how explicitly tentative and conditional
statements—which expressed certainty only as to EPA’s
decision to continue the process of studying greenhouse
gases—could possibly be considered “final.” Indeed, as the
 26
final rule states, “[t]his is not the end of the matter.” Id. at
54,996.

As an alternative, Environmental Petitioners attempt to
recast the final rule as a reviewable final decision to defer
performance of a duty pursuant to Section 7607(b)(2). This
section provides: “Where a final decision by the
Administrator defers performance of any nondiscretionary
statutory action to a later time, any person may challenge the
deferral . . . . ” 42 U.S.C. § 7607(b)(2) (emphasis added). As
environmental petitioners point out, EPA undertook the
instant NSPS rulemaking pursuant to its nondiscretionary
duty to “at least every 8 years, review and, if appropriate,
revise [NSPS] standards.” See Environmental Pets’ Reply Br.
7 (quoting 42 U.S.C. § 7411(b)(1)(B)). Arguing that this
same section confers upon EPA a nondiscretionary duty to
“complete [any] revision within the same period,”
Environmental Petitioners contend that EPA’s decision to
collect additional data on greenhouse gas emissions
constitutes a reviewable “final decision” to defer performance
of that duty. Id. at 7–8.

We are unconvinced. First, it is unclear whether EPA has
such a duty with respect to pollutants it has not previously
regulated, but in any event, nothing in the NSPS rule indicates
that EPA has made a final decision to defer performance of its
duty to “review and revise” standards. Quite to the contrary:
EPA began the process of reviewing its NSPS standards for
greenhouse gases, decided it needed further information, and
is now continuing that process of review. This might be a
different case if EPA had stated that it was deferring
consideration of greenhouse gas emissions standards until its
next mandatory NSPS review. But EPA did no such thing.
Instead, it reviewed the information it had, decided its data
was insufficient, and continued “working towards a proposal
 27
for [greenhouse gas] standards from Portland cement
facilities.” 75 Fed. Reg. 54,970, 54,997 (Sept. 9, 2010). This
court has never considered an agency decision to continue the
rulemaking process to be a “final agency action,” nor has any
court held that we have jurisdiction to review such a decision
under Section 7607(b)(2). But see Maine v. Thomas, 875 F.2d
883 (1st Cir. 1989) (decided prior to promulgation of
7607(b)(2)).

At various points in their brief, Environmental Petitioners
also appear to recast their petition as a challenge to EPA’s
“refus[al] to act,” see, e.g., Environmental Pets’ Opening Br.
35, noting that since promulgation of its NSPS rule, EPA “has
taken no steps towards either information collection or
regulating cement plants’ [greenhouse gas] emissions.” Id. at
20. But if environmental petitioners are indeed challenging a
“refusal to act,” they should have brought their case in the
district court. The Clean Air Act provides that any individual
may file suit alleging that EPA has failed “to perform any act
or duty . . . which is not discretionary with the
Administrator,” 42 U.S.C. § 7604(a)(2), and that “[t]he
district courts shall have jurisdiction” over these suits, id. §
7604(a) (emphasis added).

Because we lack statutory jurisdiction over
environmental petitioners’ claims, we have no need to
consider PCA’s alternative argument that environmental
petitioners lack Article III standing. See Nat’l Ass’n of Home
Builders v. Norton, 415 F.3d 8, 12 n. 4 (D.C. Cir. 2005).

V

For the aforementioned reasons, we grant PCA’s petition
for review with respect to EPA’s denial of reconsideration of
the NESHAP rule and remand the rule for further action, deny
 28
PCA’s petition for review with respect to the NSPS rule, and
dismiss Environmental Petitioners’ petition for lack of
jurisdiction. All of the standards will remain in place except
for the NESHAP standards applicable to clinker storage piles,
which are stayed pending reconsideration. We nonetheless
urge EPA to act expeditiously on remand. See 42 U.S.C. §
7604(a) (“any person may commence a civil action” in district
court “to compel . . . agency action unreasonably delayed”);
NRDC v. EPA, 902 F.2d 962, 985 (D.C. Cir. 1990) (“the
Clean Air Act’s citizen suit [provision] . . . may in appropriate
circumstances provide a check against indefinite stalling by
EPA.”).
So ordered.
 BROWN, Circuit Judge, concurring: I fully join the per
curiam decision, but I write separately to observe that there is
much to be said for Petitioners’ argument that EPA should not
be permitted to base NESHAP standards on bare emissions
data, and that EPA should instead isolate the effect of
emissions control technology by controlling for input quality.
Because kilns are co-located with raw material quarries and
because there is significant variability in the pollutant content
of those raw materials, a kiln may have low emissions simply
because it happens to be blessed with good inputs, not
because it is using a superior control technology. But when
the CAA directs EPA to set floors based on “the emission
control that is achieved in practice by the best controlled
similar source,” 42 U.S.C. § 7412(d)(3) (emphases added), it
would seem to be specifically directing EPA’s attention to the
active steps a kiln has taken to “control” its emissions, not
simply to the level of emissions itself. In addition to the text,
the structure of the statute also suggests that the quality of
inputs should not be permitted to affect the calculation of
floors: the “substitution of materials”—in other words, the
degree to which EPA can require kilns to switch inputs in
order to comply with a standard—is listed as a factor to be
considered in the second, beyond-the-floor determination, not
in the antecedent floor-setting determination. Id. §
7412(d)(2)(A).

As the per curiam decision notes, however, this argument
has already been rejected by the Court in Sierra Club v. EPA,
479 F.3d 875, 883 (D.C. Cir. 2007), and that decision
controls, Maxwell v. Snow, 409 F.3d 354, 358 (D.C. Cir.
2005). I am simply puzzled as to how we arrived at our
conclusion. First, the text and structure of the statute seem to
me to compel the opposite result. Second, Sierra Club relied
on our holding in National Lime Ass’n v. EPA, 233 F.3d 625,
640 (D.C. Cir. 2000), that the CAA does not require “that
 2
[the] achievement . . . be the product of a specific intent.” But
I do not read National Lime to have held that the achievement
need not be the product of any intent. Instead, context reveals
that the National Lime Court was referring to emissions of
one sort that are “controlled only incidentally by controls
placed upon” another sort of emission. Id. The incidental
control of one emission as the result of controlling another
still certainly counts as an “achievement” of emission control.
But the Court did not state—or even imply—that emissions
levels determined by inputs alone count as an “achievement”
of emission control within the meaning of the statute.

Senior Judge Williams concurred in Sierra Club to “note
a paradox in the relationship between two key provisions of §
112 of the Clean Air Act”:

What if meeting the “floors” is extremely or
even prohibitively costly for particular plants
because of conditions specific to those plants
(e.g., adoption of the necessary technology
requires very costly retrofitting, or the
required technology cannot, given local inputs
whose use is essential, achieve the “floor”)?
For these plants, it would seem that what has
been “achieved” under § 112(d)(3) would not
be “achievable” under § 112(d)(2) in light of
the latter’s mandate to EPA to consider cost.

479 F.3d at 884 (Williams, J., concurring) (emphasis added).
He was quite right that ignoring input quality when
determining the floors subverts the statutory scheme by
allowing EPA to establish a floor that some kilns simply
cannot meet. The CAA permits EPA to do this at the
“beyond-the-floor” stage as long as it considers the relevant
costs, but the very existence of that secondary phase indicates
 3
that EPA should not be permitted to set a standard at the
floor-setting stage which is unachievable due to input quality.

But it was our decision, not Congress’s, to demand that
EPA ignore input variability when it sets emissions floors. It
was our decision to not only permit but to require EPA to
ignore the costs of achieving those floors—to enact them, in
other words, even if some kilns would never be able to meet
them. Because of our decision, these “maximum achievable
control technology” floors have little to do with achievability,
controls, or technology, even though, as Senator Domenici
stated during the consideration of this law, the “initial level of
tight controls . . . is [to be] determined strictly on the basis of
the availability of technology.” 136 Cong. Rec. S17,120–24
(daily ed. Oct. 26, 1990) (statement of Sen. Pete Domenici).1

In contrast to our interpretation, the Congress that
enacted the current NESHAP program in 1990 was quite
concerned about the costs of regulation—and those costs
presumably included the economic impact of putting going
concerns out of business. The straightforward text and
structure of the floor-setting provisions convey as much.
Moreover, Congress included a specific requirement in the
1990 Amendments that EPA prepare a “comprehensive
analysis” of the “costs, benefits and other effects associated
with compliance” with, among other things, NESHAP
standards. 42 U.S.C. § 7612(a). Speaking in support of this
provision and of cost-benefit analysis more generally, Senator
Moynihan described the Amendments as “the first
environmental legislation in history to require extended and
1
In fact, EPA’s own website also refers to the standards as
“technology-based,” rather than emissions-based. Envtl. Prot.
Agency, Summary of the Clean Air Act,
http://www.epa.gov/lawsregs/laws/caa.html (last visited Nov. 7,
2011).
 4
intensive cost benefit analysis” and said, “Until now, we have
too often feared facts. This fear has not served [us] well.
Environmental programs that prohibit the EPA from taking
the costs of compliance into account have, more often than
not, resulted in deadlock.” 136 Cong. Rec. S16,895–97 (daily
ed. Oct. 27, 1990) (statement of Sen. Daniel Patrick
Moynihan). It would be strange indeed if a Congress so
attuned to the importance of cost-benefit analysis intended
EPA to set emissions floors regardless of the cost. Congress
sought to constrain the agency’s discretion; we decided to set
it free.

The truth is that this is no unavoidable paradox: the
statute’s use of terms like “achieved” and “controlled” at the
floor-setting stage urges EPA to focus on what sources have
actually done to ameliorate the pollution caused by their
particular set of inputs. If the outcome of that analysis is not
strong enough for EPA’s satisfaction, the statute’s “beyond-
the-floor” procedures provide it with an outlet to set stricter
standards, so long as it considers the costs of that course of
action. Congress’s very provision of this beyond-the-floor
mechanism and its persistent attention—reflected elsewhere
in the statute and in the legislative history—to the importance
of cost-benefit analysis only bolster this clear reading of the
text. Our holding in Sierra Club was a self-inflicted wound,
and the result of a series of interpretive leaps that I simply
cannot follow. I regret that we have ignored Congress’s
wishes and made life more difficult—for industry and its
employees, for EPA, and for ourselves.