American Petroleum Institute and Utility Air Regulatory Group v. Environmental Protection Agency

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued February 21, 2012 Decided July 17, 2012
No. 10-1079
AMERICAN P ETROLEUM INSTITUTE AND U TILITY A IR
R EGULATORY GROUP ,
P ETITIONERS
v.
ENVIRONMENTAL P ROTECTION AGENCY ,
R ESPONDENT
NATURAL R ESOURCES D EFENSE C OUNCIL,
INTERVENOR
Consolidated with 10-1080
On Petitions for Review of Final Action
of the Environmental Protection Agency
Allison D. Wood argued the cause for petitioner
American Petroleum Institute. William Pedersen argued the
cause for petitioner Interstate Natural Gas Association of
America. With them on the briefs was Lucinda Minton
Langworthy. Joan Dreskin, Timm L. Abendroth, Daniel J.
Regan Jr., and Janice K. Raburn entered appearances.

-- 1 of 21 --

2
Norman L. Rave Jr., U.S. Department of Justice, argued
the cause for respondent. On the brief was Angeline Purdy,
Attorney. Michael Augustini, Attorney, entered an
appearance.
Abigail M. Dillen, Colin O’Brien, John Walke, and
Adrian Martinez were on the brief for intervenor Natural
Resources Defense Council in support of respondent.
Before: R OGERS , Circuit Judge, and EDWARDS and
G INSBURG , Senior Circuit Judges.
Opinion for the Court filed by Senior Circuit Judge
G INSBURG .
G INSBURG , Senior Circuit Judge: In 2010, the EPA
promulgated a final rule adopting a new, one-hour primary
national ambient air quality standard (NAAQS) for nitrogen
dioxide. The American Petroleum Institute, the Utility Air
Regulatory Group, and the Interstate Natural Gas Association
of America (collectively the API) petition for review of that
rule, claiming the EPA, in adopting the NAAQS, was
arbitrary and capricious and violated the Clean Air Act. The
API also challenges a statement in the preamble to the final
rule regarding the EPA’s intended implementation of the
NAAQS. We deny the petitions insofar as they challenge the
EPA’s adoption of the NAAQS, but because the EPA’s
statement in the preamble was not final, we lack jurisdiction
to consider those portions of the petitions.
I. Background
The Clean Air Act requires the EPA to establish a
primary and a secondary NAAQS for any pollutant
“reasonably ... anticipated to endanger public health or

-- 2 of 21 --

3
welfare.” 42 U.S.C. § 7408(a)(1)(A). The EPA must set the
primary NAAQS at a level “requisite to protect the public
health” with “an adequate margin of safety.” 42 U.S.C. §
7409(b)(1).
In 1971, in order to control the emission of harmful
nitrogen oxides,* the EPA established a primary NAAQS for
nitrogen dioxide (NO2 ) of 53 parts per billion (ppb) for the
annual average in any given area. See National Primary and
Secondary Ambient Air Quality Standards, 36 Fed. Reg.
8186, 8187 (April 30, 1971). Then as now the NAAQS
focused specifically upon NO2 as an indicator for the broader
category of nitrogen oxides; because those gases are typically
emitted together and in similar proportions, detection of one
usually indicates the presence of the others. Combustion
processes, especially those occurring in automobile and truck
engines and electricity-generating plants, account for most of
the production of these compounds. Nitrogen oxides have a
variety of documented adverse effects upon human health,
including increased airway hyperresponsiveness (contraction
of the bronchioles) in asthmatics and increased respiratory
illness in children.
The EPA began a review of the NAAQS for NO2 in 2005
and revised the primary NAAQS in 2010. In the time since
its prior review, accumulated epidemiological and clinical
evidence suggested adverse health effects occurred at lower
concentrations of NO2 , and for exposures of a much shorter
duration, than scientists previously had suspected. For that
reason, the EPA began to consider whether, because the
* The EPA defines this class broadly to include “all forms of
oxidized nitrogen (N) compounds.” ENVIRONMENTAL PROTECTION
AGENCY, INTEGRATED S CIENCE ASSESSMENT FOR OXIDES OF
NITROGEN – HEALTH CRITERIA 2:1 (2008).

-- 3 of 21 --

4
existing NAAQS focused upon the annual average
concentration in an area, an additional NAAQS was necessary
to protect against the adverse effects of short-term spikes in
exposure to NO2 . In 2005, the EPA made a general call for
information, 70 Fed. Reg. 73,236 (Dec. 9, 2005), and in 2007,
after notice and comment, it published the methodology it
would use to review the NAAQS for NO2, see EPA,
INTEGRATED R EVIEW P LAN FOR THE P RIMARY NATIONAL
AMBIENT A IR QUALITY S TANDARD FOR N ITROGEN D IOXIDE
(Aug. 2007). The Review Plan described the EPA’s plans to
synthesize the results of existing epidemiological and clinical
research regarding the health effects of exposure to NO2 ,
develop forecasts of improved air quality under a hypothetical
new NAAQS set at various levels then under consideration,
submit such analyses for external review, and after public
notice and comment adopt a new NAAQS. Id. at 2–3.
In 2008, pursuant to the Review Plan, the EPA released
its Integrated Science Assessment, in which it undertook “to
critically evaluate and assess the latest scientific information
published since [the review it conducted in 1993].”
ENVIRONMENTAL P ROTECTION AGENCY, INTEGRATED
S CIENCE ASSESSMENT FOR OXIDES OF N ITROGEN – H EALTH
C RITERIA xxvii (2008) (hereinafter ISA). The ISA discussed
epidemiological evidence that showed “positive associations
of short-term ambient NO2 concentrations below the current
NAAQS level with increased numbers of [emergency room]
visits and hospital admissions for respiratory causes.” Id. at
5:11. Many studies observed such effects in areas with
average daily concentrations of NO2 between 3 and 50 ppb.
Id. The EPA also presented its updated version of a meta-
analysis*
* A meta-analysis synthesizes the results of multiple studies by
performing statistical analyses of the results of those studies.
of clinical studies on the health effects of NO2 that

-- 4 of 21 --

5
had been done by L.J. Folinsbee in 1992 and that it had
reviewed in its earlier assessment of the NAAQS. Id. at 3:14–
16. The agency made three changes to the 1992 meta-
analysis: It removed one underlying study involving specific
airway responses to ragweed, added a new study involving
non-specific airway responses, and measured the effects at
short-term concentration levels as low as 100 ppb. Id. at 3:16.
The results of the updated meta-analysis showed a statistically
significant 66 percent of resting asthmatics experienced an
increase in hyperresponsiveness in the presence of NO2
concentrations of 100 ppb. Id. The results did not, however,
reveal a dose-response relationship – one in which the
measured health effect, here the proportion of asthmatics
experiencing hyperresponsiveness, increases due to an
increased concentration or dose of some agent, here NO2 –
which would have provided a stronger indication that short-
term exposure to NO2 causes hyperresponsiveness in
asthmatics. Indeed, at levels of 200-300 ppb and over 300
ppb still about 60 percent of asthmatics experienced
hyperresponsiveness. Id. Considering the various clinical
and epidemiological studies together, however, the EPA
concluded the evidence was “sufficient to infer a causal
relationship” between short-term exposure to NO2 at levels as
low as 100 ppb and various types of respiratory morbidity; it
also concluded the data were “suggestive but not sufficient to
infer a causal relationship” between short-term exposures and
mortality. Id. at 5:5.
The EPA also assessed risks from NO2 exposure under
three different assumptions about future air quality: (1) future
air quality remains at its current level (the “as is” assumption)
(2) future air quality just meets the existing NAAQS of 53
ppb (the “just meets” assumption), and (3) future air quality
just meets several different potential hourly NAAQS, to wit,
50, 100, 150, and 200 ppb (the “new NAAQS” assumption).

-- 5 of 21 --

6
See EPA, R ISK AND EXPOSURE ASSESSMENT TO S UPPORT THE
R EVIEW OF THE NO2 P RIMARY NATIONAL A MBIENT A IR
QUALITY S TANDARD 59, 120 (Nov. 2008) (hereinafter REA).
The EPA explained that, although it had few actual data
showing any areas experienced short-term exposures above
the levels of the NAAQS under consideration in the third
scenario, REA at 89–95, its simulation procedure showed that
at the current level of air quality, people near roads are now
and would be in the future exposed numerous times in a year
to concentrations above 100 ppb (the short-term exposure
level at which the ISA concluded adverse health effects were
likely to occur), REA at 97–99. The number of such
exposures would rise dramatically under an alternate scenario
in which each area was forecasted to just meet the existing
annual standard. Id. at 144. The agency’s projections
indicated a one-hour standard (defined by the 3-year average
of the 98th percentile of hourly values) of 100 ppb measured
area-wide would improve upon the “just meets” but not upon
the “as is” scenario; a one-hour standard of 50 ppb would
improve upon the “as is” scenario. Id.
The EPA submitted both the ISA and the REA to the
Clean Air Scientific Advisory Committee (CASAC) for
review, as required by the Clean Air Act, 42 U.S.C.
7409(d)(2)(A). The CASAC agreed with the EPA’s
assessment that the current annual NAAQS was not adequate
to protect human health, and it suggested the agency adopt a
one-hour standard for NO2 of no greater than 100 ppb.
In 2009 the EPA proposed to set a new hourly NAAQS
with allowable maximum concentration levels between 80
and 100 ppb. 74 Fed. Reg. 34,404 (July 15, 2009). The
petitioners each submitted comments criticizing the EPA for
proposing a revision to the NAAQS based upon an
unpublished study, i.e., the updated meta-analysis, and for

-- 6 of 21 --

7
discounting a published and peer-reviewed study that did not
conclude exposures to NO2 at 100 ppb caused a measureable
adverse health effect. They also expressed skepticism about
the EPA’s interpretation of the epidemiological evidence,
questioned the assumptions built into the forecasts in the
REA, and pointed out the proposed rule provided no guidance
as to how a permit applicant for a new or modified source of
NO2 pollution should demonstrate compliance with the new
NAAQS.
In its Final Rule, the EPA adopted a new one-hour
primary NAAQS, requiring in effect that “the three-year
average of the annual 98th percentile of the daily maximum 1-
hour average concentration [be] less than or equal to 100
ppb.” Primary National Ambient Air Quality Standards for
Nitrogen Dioxide, 75 Fed. Reg. 6474, 6531 (Feb. 9, 2010)
(codified at 40 C.F.R. § 50.11(f)). The EPA concluded this
standard was needed “to provide protection for asthmatics and
other at-risk populations against an array of adverse
respiratory health effects related to short-term NO2 exposure.”
Id. at 6502.
II. Analysis
The API petitioned for review of the Final Rule under 42
U.S.C. § 7607(b), which gives this court exclusive jurisdiction
to hear a challenge to a NAAQS. We review the EPA’s
setting of a NAAQS to determine whether it was “arbitrary,
capricious, an abuse of discretion, or otherwise not in
accordance with law.” 42 U.S.C. § 7607(d)(9)(A); see also 5
U.S.C. § 706(2)(A). According to the API, the EPA was
arbitrary and capricious in how it dealt with the record
evidence and the NAAQS it adopted is unlawful because
more stringent than “requisite to protect the public health”
with “an adequate margin of safety,” 42 U.S.C. § 7409(b)(1).

-- 7 of 21 --

8
In addition, the API argues the EPA’s implementation of the
NAAQS was arbitrary and capricious because, when the EPA
stated a permit applicant for a new or modified source of
pollution must demonstrate compliance with the new
NAAQS, the agency – or so the API asserts – did not consider
whether it would be able to resolve applications within the
statutorily required time period or what effect such a
requirement might have upon economic growth.
A. The EPA’s Adoption of the One Hour NAAQS
The API claims the process by which the EPA adopted
the new NAAQS was flawed and the standard must therefore
be vacated. More specifically, it faults the EPA for (1)
relying upon an unpublished, non-peer-reviewed meta-
analysis of clinical studies, (2) discounting a published meta-
analysis that called into question the EPA’s conclusions, (3)
treating the same epidemiological study differently in reviews
of the NAAQS for NO2 and for ozone, and (4) projecting the
benefits to air quality from the new NAAQS based upon
faulty assumptions.
1. Peer Review of the Meta-Analysis
The API first contends the EPA, by relying upon an
internal meta-analysis that was not published, “did not follow
its own requirements ... that it rely only on peer-reviewed and
published studies in reviewing NAAQS.” Pet. Br. at 26.
Perhaps the API should have had its brief peer-reviewed. In
quoting the EPA’s Review Plan, see Pet. Br. at 28, the API
omits the first and most relevant word of the following
sentence: “Generally, only information that has undergone
scientific peer review and that has been published (or
accepted for publication) in the open literature will be
considered,” Review Plan at 11; see also ISA Annexes at 1:1.

-- 8 of 21 --

9
Of course, “generally” here indicates the practice in question
will not invariably be followed, see Kurke v. Oscar Gruss and
Son, Inc., 454 F.3d 350, 355–56 (D.C. Cir. 2006) (“as ... the
word ‘generally’ suggests, there are exceptions to the rule”);
Bernhardt v. City & Suburban Ry. Co., 263 F. 1009, 1015
(D.C. Cir. 1920) (“the word ‘generally’ ... indicates that there
may be [exceptions to the stated rule]”). A bad start for the
petitioners.
The API also points to guidelines the EPA promulgated
pursuant to the Information Quality Act (IQA), Pub. L. 106-
554, § 515(b)(2)(A) (requiring each federal agency to issue
guidelines “for ensuring and maximizing the quality,
objectivity, utility, and integrity of information disseminated
by the agency”), which it contends also require peer review.
See EPA, GUIDELINES FOR ENSURING AND M AXIMIZING THE
QUALITY, OBJECTIVITY , UTILITY, AND INTEGRITY OF
INFORMATION D ISSEMINATED BY THE E NVIRONMENTAL
P ROTECTION AGENCY (Oct. 2002). By their terms, however,
the Guidelines provide only “non-binding policy and
procedural guidance.” Id. at 4. Such a statement would not
override a specific commitment made elsewhere in the
document, see Appalachian Power Co. v. EPA, 208 F.3d
1015, 1022–23 (D.C. Cir. 2000), but the petitioners point to
none. In keeping with the Review Plan and the ISA, the
Guidelines also state that “major scientifically and technically
based work products ... related to Agency decisions should be
peer-reviewed.” Guidelines at 11. The use of the phrase
“should be” rather than “shall” suggests but does not
necessarily mean the Guidelines are not binding. Compare
Doe v. Hampton, 566 F.2d 265, 281 (D.C. Cir. 1977) (“that
the provision in question employs the directory ‘should be’
rather than the mandatory ‘shall’ or ‘must’ ... should not be
automatically determinative of the issue”), with Jolly v.
Listerman, 672 F.2d 935, 945 (D.C. Cir. 1982) (“use of the

-- 9 of 21 --

10
word ‘should’ ... detracts significantly from any claim that
this guideline is more than merely precatory”), and Military
Toxics Project v. EPA, 146 F.3d 948, 958 (D.C. Cir. 1998)
(accepting as permissible EPA’s interpretation of “word
‘should’ ... as calling for an exercise of judgment and hence
conferring discretion upon the Administrator”). More
important, the Guidelines themselves expressly commit “the
decision whether to employ peer review” to the discretion of
agency management. Guidelines at 11. Finally, the
Guidelines note the EPA’s Peer Review Handbook “provides
detailed guidance for implementing” the agency’s peer review
policy, id., and the Handbook in turn states specifically the
relevant decision-makers “need[] to make a judgment”
whether peer review is appropriate in a specific case because
“[t]here is no easy single yes/no test,” EPA, PEER R EVIEW
HANDBOOK § 2.2.3 (2000). No doubt the EPA believes peer
review is important and it intended to impress that value upon
its staff, but the agency did not bind itself to a judicially
enforceable norm.
We need not decide the extent, if any, to which an
agency must account for any departure from a non-binding
guideline, compare Sitka Sound Seafoods, Inc. v. NLRB, 206
F.3d 1175, 1182 (D.C. Cir. 2000) (because manual was non-
binding, question was whether, apart from requirements of
manual, agency acted reasonably), with Edison Elec. Inst. v.
EPA, 391 F.3d 1267, 1269 & n.3 (D.C. Cir. 2004) (agency
must account for departure from non-binding plan), because
the EPA, contrary to the API’s claim, did not depart from its
non-binding peer review policy. The EPA merely updated the
Folinsbee meta-analysis, which was originally peer-reviewed
and published; the only data it added to the meta-analysis
were the results of a study that was itself peer-reviewed and
published; and the CASAC peer-reviewed the results of the
updated meta-analysis. See Peer Review Handbook § 2.4.3(d)

-- 10 of 21 --

11
(listing the CASAC among acceptable sources of external
peer review); cf. City of Portland, Or. v. EPA, 507 F.3d 706,
716 (D.C. Cir. 2007) (holding “advice from [EPA’s] Science
Advisory Board [a group of outside scientists, similar to the
CASAC, organized by the EPA to review its scientific
analyses] ... [was an] acceptable form of peer review”). The
EPA also relied upon epidemiological studies, as well as
individual clinical studies underlying the meta-analysis that
had been published and peer-reviewed. The EPA’s staff
conducting the review of the proposed NAAQS judged the
CASAC’s review of the meta-analysis was sufficient, and the
API has presented no reason for us to disturb that judgment.
2. Treatment of the Goodman Study
Next, the API argues the EPA inappropriately discounted
the results of a published meta-analysis by Dr. Julie E.
Goodman et al., Meta-Analysis of Nitrogen Dioxide Exposure
and Airway Hyper-Responsiveness in Asthmatics, 39 C RIT.
R EV. TOXICOLOGY 719 (2009). According to the API, the
study suggests, contrary to the EPA’s findings, there is no
causal relationship between an increase in ambient
concentrations of NO2 and an increase in health effects, such
as airway hyperresponsiveness in asthmatics. The EPA
defends its treatment of the study as follows: The agency did
not receive the study until after it had conducted the analyses
described in the ISA and the REA and submitted them to the
CASAC for review; it nevertheless considered the study but
found its methodology wanting and therefore not a reason for
reopening its review process.
An agency’s action is arbitrary and capricious if it
“entirely failed to consider an important aspect of the problem
[or] offered an explanation for its decision that runs counter to
the evidence before the agency.” North Carolina v. EPA, 531

-- 11 of 21 --

12
F.3d 896, 906 (D.C. Cir. 2008) (quoting Motor Vehicle Mfrs.
Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43
(1983)). The API has not shown the EPA’s treatment of the
Goodman study fell below these standards.
First, the results of the Goodman study did not “run[]
counter to the evidence before the agency,” North Carolina,
531 F.3d at 906. Contrary to the API’s description, the study
did not establish there was “no dose-response relationship”; it
simply failed to reject the null hypothesis to that effect. That
is, the authors could not tell whether there is no such
relationship or their test merely lacked sufficient power to
detect the relationship. See David H. Kaye & David A.
Freedman, Reference Guide on Statistics, in R EFERENCE
M ANUAL ON S CIENTIFIC EVIDENCE 211, 253–54 (Fed. Judicial
Ctr. 3d ed. 2011), available at
http://www.fjc.gov/public/pdf.nsf/lookup/SciMan3D01.pdf/$f
ile/SciMan3D01.pdf. The Goodman study, moreover, did not
test for the possibility of a non-linear dose-response
relationship.* See Goodman at 733.
* See Daniel L. Rubinfeld, Reference Guide on Multiple Regression,
in REFERENCE M ANUAL ON SCIENTIFIC EVIDENCE at 303, 316
(explaining "[f]ailure to account for nonlinearities [in the estimated
equation] can lead to either overstatement or understatement of the
effect of a change in the value of an explanatory variable on the
dependent variable"). The underlying data in the EPA’s meta-
analysis indeed suggest that roughly the same proportion of
asthmatics, 58 to 66 percent, experience airway
hyperresponsiveness whether exposed to levels of 100 ppb, 200
ppb, or even 300 or more ppb. See ISA at 3:14–16. These results
are consistent with a non-linear dose-response relationship that
increases sharply at a low concentration of ambient NO2 and then
flattens out as the dose nears a concentration of 100 ppb.

-- 12 of 21 --

13
Second, because the EPA gave the specific reasons for
which it disagreed with Goodman’s methodology, it did not
“fail[] to consider” the study, nor did it fail to “offer[] an
explanation” for not relying upon that study. North Carolina,
531 F.3d at 906. The EPA explained that it had decided not
to focus upon detecting a dose-response relationship in its
meta-analysis because differences in the assumptions made
and in the methodologies used in the underlying studies
would likely make it impossible to derive a reliable estimate.
See Final Rule, 75 Fed. Reg. at 6487, 6498. The EPA did
acknowledge the limitations inherent in its own study, noting
“uncertainty with regard to the magnitude and [to] the
clinical-significance of NO2 -induced increases in airway
responsiveness,” id. at 6488, but it explained the Clean Air
Act requires the agency to promulgate a primary NAAQS to
protect the public health even where, as here, the risks from
the pollutant could not be quantified or “precisely identified
as to nature or degree,” Am. Trucking Ass’ns v. EPA, 283 F.3d
355, 369 (D.C. Cir. 2002) (internal quotation marks and
citation omitted). The EPA was therefore justified in revising
the NAAQS considering the evidence of a statistically
significant relationship between relevant health conditions
and NO2 exposure at various concentrations, even if the
agency did not know the precise dose-response relationship
between NO2 and airway responsiveness, among other health
effects.
The API mistakenly places much weight upon our recent
decision in Business Roundtable v. SEC, 647 F.3d 1144 (D.C.
Cir. 2011). As the foregoing discussion makes clear, the
EPA’s analysis of the proposed NAAQS was materially better
than the analysis for which we faulted the SEC in that case.
There the agency had ignored “numerous studies submitted by
commenters that reached the opposite result” and relied
instead upon “two relatively unpersuasive studies.” 647 F.3d

-- 13 of 21 --

14
at 1150–51. Putting aside the analytical incoherence of the
SEC’s rationale, which would have been fatal by itself, the
evidentiary problem in Business Roundtable was not limited
to the agency’s insufficient treatment of any one study,
though there was that, see id. at 1151; it was the agency’s
larger failure to deal with the weight of the evidence against
it. The EPA’s analysis at issue here was in no way
comparable to the botched job on display in Business
Roundtable. The EPA, in addition to performing a meta-
analysis of 19 underlying clinical studies of the effects NO2
exposure has upon health, relied upon numerous
epidemiological studies, which evidenced a relationship
between an increase in local ambient NO2 concentrations and
an increase in local emergency room visits, see Final Rule, 75
Fed. Reg. at 6488–89. The API has pointed to nothing
arbitrary or capricious either in the agency’s handling of the
Goodman study in particular, or in its treatment of this other
record evidence supporting the EPA’s conclusion.
3. Treatment of the Schildcrout Study
The API also argues the EPA acted inconsistently, and
therefore arbitrarily and capriciously, by relying upon an
epidemiological study by Jonathan S. Schildcrout et al.,
Ambient Air Pollution and Asthma Exacerbations in
Children: An Eight City Analysis, 164 AM. J. EPIDEMIOLOGY
505–17 (2006), in its review of the NAAQS for NO2 although
the agency allegedly had decided in 2006 not to rely upon the
same study when it was reviewing the NAAQS for ozone.
According to the EPA, the study came too late in its 2006
review process for ozone but the agency did consider it in a
later assessment of recent studies relevant to the NAAQS for
ozone. See Final Rule, 75 Fed. Reg. at 6486. The agency’s
explanation is rational, and the API makes no attempt to rebut
it; enough said about this issue.

-- 14 of 21 --

15
4. Alternate Scenarios
The API next contends the EPA used faulty assumptions
in projecting the degree to which air quality would be
improved under the new NAAQS and therefore exceeded its
authority under the Clean Air Act by adopting a NAAQS
more stringent than is “requisite to protect the public health”
with “an adequate margin of safety,” 42 U.S.C. § 7409(b)(1).
In particular, the API points to the EPA’s comparison of air
quality under various potential new NAAQS against several
different projections of air quality without a new NAAQS,
including one scenario in which it assumed all areas just meet
the current air quality standards. The API underscores that
the EPA acknowledged, contrary to this “just meets”
assumption, current air quality is significantly better than
what the existing annual NAAQS for NO2 requires, and the
EPA has even projected that if it took no action air quality
would continue to improve in this respect. Therefore, the API
reasons, the EPA should have measured the likely benefits of
the new NAAQS relative to a projection of air quality more
accurate than its “just meets” scenario, which change it claims
would have shown the one-hour NAAQS was not necessary
“to protect the public health” with “an adequate margin of
safety.”
The EPA says the API misunderstands the agency’s
analysis. In predicting a benefit to air quality from adopting
the new NAAQS for NO2 , it did not rely solely upon the
assumption that air quality would just meet existing NAAQS
if a new NAAQS was not in place; it also measured the
improvement in air quality under the “as is” assumption, in
which the agency assumed air quality would remain at current
levels.

-- 15 of 21 --

16
The API is correct to the extent that, as the word
“requisite” in § 109(b)(1) of the Clean Air Act indicates, the
EPA is to set a NAAQS that is “not lower or higher than is
necessary ... to protect the public health,” Whitman v. Am.
Trucking Ass’ns, 531 U.S. 457, 475–76 (2001) (interpreting
42 U.S.C. § 7409(b)(1)). The same statutory provision,
however, unhorses the API’s argument because it enjoins the
EPA to set the standard with “an adequate margin of safety,”
which means the agency is to “err on the side of caution.”
Am. Farm Bureau Fed’n v. EPA, 559 F.3d 512, 533 (D.C. Cir.
2009) (internal quotation marks and citation omitted). In
other words the Act contemplates the agency “should set
standards providing ‘a reasonable degree of protection ...
against hazards which research has not yet identified.’”
Coalition of Battery Recyclers Ass’n v. EPA, 604 F.3d 613,
618 (D.C. Cir. 2010) (quoting Natural Res. Def. Council v.
EPA, 824 F.2d 1146, 1152 (D.C. Cir. 1987) (en banc)). The
uncertainty inherent in predicting the future is particularly
vexing when one is making a projection of air quality, the
actual future of which depends upon regulatory policy,
technological change, economic performance, and political
outcomes, among other variables. Although air quality had
improved and was expected to keep improving, it was
certainly possible this trend would be reversed. Therefore, it
was not unreasonable for the EPA to measure expected
benefits from the new NAAQS in part upon the assumption
that, if the new NAAQS were not adopted, then each area
would in the future just meet the existing standard.
Moreover, the EPA maintains its comparison in the REA
of expected benefits under a new 100 ppb hourly NAAQS
(the “new NAAQS” scenario) against the more realistic “as
is” scenario does not, as the API contends, show the new
NAAQS would provide no benefit. As the agency explains,
the API disregards a critical difference between the

-- 16 of 21 --

17
hypothetical 100 ppb standard in the REA and the 100 ppb
standard the EPA eventually adopted: The “new NAAQS”
scenario in the REA assumed the standard would be set at an
area-wide average, i.e., the average value recorded by the
monitors in an area equaled the level set by the NAAQS, so
that some monitors would record concentrations of NO2
above and some below that standard. The new NAAQS the
EPA actually adopted, however, applies to peak rather than to
average concentrations, i.e., it requires that all monitors in an
area be below the 100 ppb level. Accordingly, the assumption
in the REA that an area meets a hypothetical new NAAQS of
100 ppb does not fully capture the expected improvement in
air quality from the hourly 100 ppb peak concentration
standard the agency ultimately adopted.
Because a peak hourly concentration of 100 ppb is
roughly equivalent to an area-wide hourly average
concentration of between 50 and 75 ppb, see Final Rule, 75
Fed. Reg. at 6494, the EPA concluded the standard it adopted
corresponds more closely to the “new NAAQS” scenario in
the REA with a standard of 50 ppb than the scenario using the
100 ppb assumption upon which the API focuses its criticism.
In the REA the agency had projected a new NAAQS of 50
ppb area-wide would provide a substantial improvement over
current air quality. See REA at 120.
Considering its duty to err on the side of caution, we
conclude the EPA did not act unreasonably by comparing the
benefits of the one-hour standard against not only a scenario
based upon existing air quality but also upon an alternate
scenario in which areas just meet the annual NAAQS set in
1971. For that reason, and because the record adequately
supports the EPA’s conclusion that material negative health
effects result from ambient air concentrations as low as the
100 ppb level, we cannot conclude the agency was arbitrary

-- 17 of 21 --

18
and capricious or violated the Act in adopting that level as the
new one-hour NAAQS for NO2 .*
B. Statement Regarding Permitting
Finally, the API claims the EPA was arbitrary and
capricious when it allegedly decided to require applicants for
new or modified sources of pollution under § 165(a) of the
Clean Air Act, 42 U.S.C. § 7475(a) (prohibiting construction
of a “major emitting facility” without a permit), to
demonstrate their compliance with the new NAAQS despite
the lack of an adequate technique to model compliance.
According to the API, the EPA should have considered the
effect of that decision upon the agency’s ability to resolve
each application within one year of its filing, as required by §
165(c) of the Act, 42 U.S.C. § 7475(c) (“Any completed
permit application ... for a major emitting facility ... shall be
granted or denied not later than one year after the date of
filing”), and upon its alleged duty under § 160(3) of the Act to
consider the effect of its permitting decisions upon economic
growth, see 42 U.S.C. § 7470(3) (“purposes of this part
[include] ... insur[ing] that economic growth will occur in a
manner consistent with the preservation of existing clean air
resources”). The API argues that until such time as methods
for modeling compliance with the new one-hour NAAQS are
developed and have been approved by the agency, the EPA
should allow applicants to demonstrate compliance with the
* We note the API does not take issue with the EPA’s simulation of
air quality at and near roadways as routinely exceeding the new 100
ppb standard. Indeed, counsel for the API confirmed at oral
argument that, so long as we accept the agency's findings on the
health effects of NO2 at concentrations as low as 100 ppb and we
reject the API's criticism of the agency's forecasts, it has raised no
challenge to the EPA's setting of the new NAAQS at 100 ppb. So
be it.

-- 18 of 21 --

19
pre-existing annual NAAQS, as they previously had to do.
The EPA maintains the Final Rule does not constitute a final
decision concerning the permitting of new or modified
sources under the new NAAQS, and is therefore not subject to
judicial review, see Portland Cement Ass’n v. EPA, 665 F.3d
177, 193 (D.C. Cir. 2011) (“The Clean Air Act gives [the
court] jurisdiction to review only ‘final’ agency actions”
(citing 42 U.S.C. § 7607(b))).
The only reason the API has for suggesting the EPA has
taken any final action regarding the permitting of a new or
modified source is the statement in the preamble to the Final
Rule that “major new and modified sources applying for
[permits under § 165 of the Act] will initially be required to
demonstrate that their proposed emissions increases of NOx
will not cause or contribute to a violation of ... the [new] 1-
hour NO2 NAAQS.” 75 Fed. Reg. at 6525. Although “there
is [no] categorical bar to judicial review of a preamble,”
Kennecott Utah Copper Corp. v. U.S. Dep’t of Interior, 88
F.3d 1191, 1222 (D.C. Cir. 1996) (citation omitted), it “is not
the norm,” Natural Res. Def. Council v. EPA, 559 F.3d 561,
565 (D.C. Cir. 2009). The operative question when faced
with such a challenge is “whether the [preambular statement]
has independent legal effect, which in turn is a function of the
agency’s intention to bind either itself or regulated parties.”
Kennecott, 88 F.3d at 1223.
Any action of an agency, including a statement in a
preamble, is “final” only if it (1) “mark[s] the consummation
of the agency’s decisionmaking process” and (2) is “one by
which rights or obligations have been determined, or from
which legal consequences will flow[.]” Bennett v. Spear, 520
U.S. 154, 178 (1997) (internal quotation marks and citations
omitted). The preambular statement challenged here has
neither effect, as indicated both on its face and, more clearly,

-- 19 of 21 --

20
by the context in which it was made. To be sure, one could
reasonably read as mandatory the isolated statement that
permit applicants “will initially be required” to meet the new
NAAQS. At the same time, the statement could reasonably
be read to mean the EPA intends in the future to establish
such a requirement, in which case the statement falls short of
being the consummation of the agency’s decisionmaking
process. The Supreme Court similarly has said “a statement
in [an agency’s land management] plan that [it] ‘will’ [rather
than “shall”] take this, that, or the other action ... is not [a
binding commitment] ... absent clear indication” to the
contrary, Norton v. S. Utah Wilderness Alliance, 542 U.S. 55,
69 (2004), which suggests the statement that applicants “will
initially be required” is predictive of the agency’s future
actions, not one from which “legal consequences w[ould]
flow.”
Read in context, the lack of finality in the statement is
more obvious:
The EPA acknowledges that a decision to promulgate
a new short-term NO2 NAAQS will clearly have
implications for the air permitting process. The full
extent of how a new short-term NO2 NAAQS will
affect the [new source review] process will need to be
carefully evaluated. First, major new and modified
sources applying for [new source review or prevention
of significant deterioration] permits will initially be
required to demonstrate that their proposed emissions
increases of NOX will not cause or contribute to a
violation of ... the annual or 1-hour NO2 NAAQS ....
Final Rule, 75 Fed. Reg. at 6525. By acknowledging it had
not yet, but “w[ould] need to[,] ... carefully evaluate[]” the
effect of the new NAAQS on the permitting process, the EPA

-- 20 of 21 --

21
made clear it was not making a final decision. The subject
statement does not express a final agency action, and so we
lack jurisdiction under the Clean Air Act, 42 U.S.C. §
7607(b), to consider the API’s challenge to it.
III. Conclusion
Because the API has not shown the EPA’s adoption of
the one-hour NAAQS for NO2 was either arbitrary and
capricious or in violation of the Clean Air Act, we shall deny
the petitions in that respect. The portions of the petitions
challenging the EPA’s non-final statement regarding
permitting in the preamble to the Final Rule we shall dismiss
for lack of jurisdiction.
So ordered.

-- 21 of 21 --