Full opinion text
Table of Contents I. Background.11 II. Industry Challenges.18 A. Modification.19 B. Interpretation of 1980 Rule in 2002 Preamble.20 C. Source-Specific Allowable Emissions.21 III. Baseline Emissions. CO H-1 A. Statutory Interpretation. to tO B. Environmental Impact. to —•3 IV. Methodology and Enforceability. CO M A. Demand Growth Exclusion. CO H B. Recordkeeping and Reporting Requirements CO CO V. Plantwide Applicability Limitations.36 VI. Clean Units.38 VII. Pollution Control Projects.40 VIII. State and Local Authority.42 A. Alternative NSR Standards.42 B. Anti-backsliding. Cq C. Notice re Menu of Alternatives ^ IX. Conclusion. .44 Before: ROGERS and TATEL, Circuit Judges, and WILLIAMS, Senior Circuit Judge. Opinion for the Court filed by Circuit Judge PER CURIAM. Concurring opinion filed by Senior Circuit Judge WILLIAMS. PER CURIAM. In 1977, Congress amended the Clean Air Act (“CAA” or “the Act”) to strengthen the safeguards that protect the nation’s air quality. Among other things, these amendments directed that major stationary sources undertaking modifications must obtain preconstruction permits, as must major new sources, through a process known as “New Source Review” (“NSR”). According to a preexisting definition referenced in the 1977 amendments, a source undertakes a modification when “any physical change ... or change in the method of operation ... which increases the amount of any air pollutant emitted by such source” occurs. 42 U.S.C. § 7411(a)(4) (2000). The Environmental Protection Agency (“EPA”) has interpreted this rather terse definition in numerous rules, including ones issued in 1980, 1992, and most recently in 2002. Industry, government, and environmental petitioners now challenge this 2002 rule, which departs sharply from prior rules in several significant respects. Roughly speaking, industry petitioners argue that the 2002 rule interprets “modification” too broadly, while government and environmental petitioners argue that the rule’s interpretation is too narrow. Industry petitioners have also revived previously stayed challenges to EPA’s earlier rules. Today, we reject challenges to substantial portions of the 2002 rule. Specifically, we find the following elements permissible interpretations of the CAA and not otherwise arbitrary and capricious: the use of past emissions and projected future actual emissions, rather than potential emissions, in measuring emissions increases; the use of a ten-year lookback period in selecting the two-year baseline period for measuring past actual emissions; the use of a five-year lookback period in certain circumstances; the abandonment of a provision authorizing states to use source-specific allowable emissions in measuring baseline emissions; the exclusion of increases due to unrelated demand growth from the measurement of projected future actual emissions; and the Plantwide Applicability Limitations (“PAL”) program. We also find meritless certain procedural challenges related to lack of notice. We conclude, however, that two aspects of the 2002 rule rest on impermissible interpretations of the Act and a third is arbitrary and capricious. Specifically, EPA erred in promulgating the Clean Unit applicability test, which measures emissions increases by looking to whether “emissions limitations” have changed. Congress directed the agency to measure emissions increases in terms of changes in actual emissions. EPA also erred in exempting from NSR certain Pollution Control Projects (“PCPs”) that decrease emissions of some pollutants but cause collateral increases of others. The statute authorizes no such exception. 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. The agency failed to provide a reasoned explanation for how, absent such records, it can ensure compliance with NSR. Finally, industry challenges to passages in the preambles to the 2002 and 1992 rules, as well as government challenges to the implementation of the 2002 rule, are unripe for review. I. Background The 1977 CAA amendments define “modification” by reference to a statutory provision added in 1970. Seeking to understand what the 1977 Congress meant by modification — the central issue in this case — we thus begin with the 1970 CAA amendments and their implementing regulations. Congress passed the 1970 amendments “to protect and enhance the quality of the Nation’s air resources so as to promote the public health and welfare and the productive capacity of its population.” 42 U.S.C. § 7401(b). The amendments set out a two-step process for achieving this goal: EPA first develops “National Ambient Air Quality Standards” (“NAAQS”) for various pollutants, and states then create and implement plans, known as “State Implementation Plans” (“SIPs”), to ensure their air meets these standards. See id. §§ 7409-7410. The amendments also required new or modified sources to conform to emissions limits, known as “New Source Performance Standards” (“NSPS”), set by EPA. See id. § 7411. Because “[t]he Act contemplated” that these criteria would be “more stringent than those needed to meet ... NAAQS,” Alabama Power Co. v. Costle, 636 F.2d 323, 346 (D.C.Cir.1979), the meaning of “modified sources” took on particular significance: if an existing source made a “modification,” it needed to conform its change to NSPS, whereas an unmodified source only needed to meet whatever lesser requirements (if any) the SIP imposed for attaining NAAQS. Congress provided the following definition for “modification”: any physical change in, or change in the method of operation of, a stationary source 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. 42 U.S.C. § 7411(a)(4). This definition requires both a change — whether physical or operational — and a resulting increase in emissions of a pollutant. EPA’s 1975 NSPS regulation, like its earlier 1971 regulation, elaborated upon this statutory definition, doing so in provisions whose meaning the parties debate today. One part of the 1975 regulation provided that “ ‘[m]odification’ means any physical change in, or change in the method of operation of, an existing facility which increases the amount of any air pollutant (to which a standard applies) emitted into the atmosphere by that facility.” 40 Fed.Reg. 58,416, 58,418 (Dec. 16, 1975); see also 36 Fed.Reg. 24,876, 24,877 (Dec. 23, 1971). Using somewhat different terms, another part of the 1975 regulation stated that “any physical or operational change to an existing facility which results in an increase in the emission rate to the atmosphere of any pollutant to which a standard applies shall be considered a modification within the meaning ... of the Act,” with “[e]mission rate ... expressed as kg/hr of any pollutant discharged into the atmosphere.” 40 Fed.Reg. at 58,419. Yet neither the 1975 regulation nor its preamble explained why EPA found it necessary to offer these two separate glosses on “modification.” Adding to the confusion, EPA put forth yet another definition of “modification” in a 1974 regulation implementing what became known as the regulatory “Prevention of Significant Deterioration” (“PSD”) program. Seeking to prevent backsliding in regions whose air quality met NAAQS, this program required new sources and sources undertaking modifications to obtain preconstruction permits. See Alabama Power, 636 F.2d at 346-49 (describing the regulatory PSD program). The regulation defined “modification” in a manner that closely tracked — but didn’t precisely mirror — the NSPS regulatory definition, stating that “[t]he phrases ‘modification’ or ‘modified source’ mean any physical change in,-or change-in the method of operation of, a stationary source which increases the emission rate of any pollutant for which a national standard has been promulgated.” 39 Fed.Reg. 42,510, 42,514 (Dec. 5, 1974). The regulation’s preamble further provided that the term “modified source” was meant “to be consistent with the definition used in [NSPS].” Id. at 42,513. Both the NSPS and PSD regulations listed certain exceptions to what constitutes a “modification,” though once again the precise content of the regulations varied. The 1974 PSD and the 1971 NSPS regulations provided that: (1) Routine maintenance, repair, and replacement shall not be considered a physical change, and (2) The following shall not be considered a change in the method of operation: (i) An increase in the production rate, if such increase does not exceed the operating design capacity of the source; (ii) An increase in the hours of operation; (iii) Use of an alternative fuel or raw material [under certain conditions]. Id. at 42,514; accord 36 Fed.Reg. at 24,877. The 1975 NSPS regulation not only phrased its exceptions differently, but also added a few additional ones: The following shall not, by themselves, be considered modifications under this part: (1) Maintenance, repair, and replacement which the Administrator determines to be routine ...; (2) An increase in production rate of an existing facility, if that increase can be accomplished without a capital expenditure on the stationary source containing that facility; (3) An increase in the hours of operation; (4) Use of an alternative fuel or raw material [under certain conditions] ...; (5) The addition or use of any system whose primary function is the reduction of air pollutants ...; (6) The relocation or change in ownership of an existing facility. 40 Fed.Reg. at 58,419-20. In its various permutations, this regulatory framework had not been long in place when, in 1977, Congress amended the CAA yet again. These amendments drew upon, expanded, and superceded the regulatory PSD program. In particular, the amendments strengthened the Act by (1) expressly creating a preconstruction review process for new or modified major sources located in “nonattainment” areas (i.e., areas which failed to meet NAAQS), see generally 42 U.S.C. §§ 7501-7515; and (2) expressly providing a parallel preconstruction review process in PSD areas (i.e., areas which met NAAQS or where there was insufficient information to evaluate whether NAAQS were met), see generally id. §§ 7470-7492. The parties refer to the first as “Nonattainment New Source Review” (“NNSR”), to the second as “Prevention of Significant Deterioration” (“PSD”), and to both collectively as “New Source Review” (“NSR”). We shall do the same. Under the amendments, sources seeking NNSR permits must meet stricter requirements than sources seeking PSD permits. Most notably, for NNSR permits, sources must achieve the “lowest achievable emission rate” (“LAER”) for new or modified units, whereas sources seeking PSD permits need only use the less demanding “best available control technology” (“BACT”). At a minimum, LAER and BACT are as restrictive as NSPS. Id. § 7479(3) (“In no event shall application of [BACT] result in emissions of any pollutants which will exceed the emissions allowed by any applicable standard established pursuant to” NSPS); accord id. § 7501(3) (for LAER). In certain circumstances, however, BACT and LAER can be more stringent than NSPS. See id. § 7479. Moreover, to obtain NNSR permits, sources must arrange for emissions reductions at other sources such that the modifications produce no increase in overall regional emissions. Id. § 7503. Sources must also demonstrate that any other sources owned by the same company comply with CAA requirements. Id. To obtain PSD permits, sources must undergo ambient air quality analyses to show that they will neither violate NAAQS increments nor adversely affect air quality in national parks or other areas that EPA has designated as needing particularly high-quality air. Id. § 7475. Congress meant NSR to apply to both new and modified sources. Due to a technical defect, however, Congress initially achieved this goal only in the NNSR portion of the amendments, which defined modification by reference to the NSPS definition: “The terms ‘modifications’ and ‘modified’ mean the same as the term ‘modification’ as used in section 7411(a)(4) of this title.” Id. § 7501(4). By contrast, the PSD portion of the amendments applied initially to new sources only. Congress corrected this in a technical amendment passed several months later, which applied the PSD program to sources that were to undergo modifications “as defined in section 7411(a) of this title.” Pub.L. No. 95-190, § 14(a)(54), 91 Stat. 1393, 1402 (1977) (codified at 42 U.S.C. § 7479(2)(C)). As the legislative history explains, this “technical and conforming” amendment “[implements conference agreement to cover ‘modification’ ... [in] conform[ance with] usage in other parts of the Act.” 123 Cong. Rec. 36,250, 36,253 (Nov. 1, 1977). In sum, the 1977 amendments carved out a significant difference between existing sources on the one hand and new or modified sources on the other. The former faced no NSR obligations — in the common phrase, they were “grandfathered” — while the latter were subject to strict standards. Limiting NSR to new or modified sources was one method of accomplishing the amendments’ goal of “a proper balance between environmental controls and economic growth,” id. at 27,076 (Aug. 04, 1977) (statement of Rep. Waxman) (quoted in Chevron U.S.A., Inc. v. Natural Res. Def. Council, 467 U.S. 837, 852 n. 25, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984)). EPA promulgated an NSR regulation in 1978. (Although at this time and later ones, EPA issued multiple sets of regulations — those applying to PSD in states without approved SIPs, those applying to NNSR in states without approved SIPs, those applying to PSD in states with approved SIPs, and those applying to NNSR in states with approved SIPs — these sets are sufficiently similar that for simplicity we typically reference the first of these as a shorthand for them all.) The 1978 regulation defined a major “modification” as a “physical change, change in the method of operation of, or addition to a stationary source which increases the potential emission rate of any air pollutant regulated under the act.” 43 Fed.Reg. 26,380, 26,403-04 (June 19, 1978). The phrase “potential emission rate,” though new to EPA regulations relating to “modification,” went unchallenged during ensuing litigation over other aspects of the 1978 regulation. That litigation culminated in this circuit’s Alabama Power Co. v. Costle decision, issued initially as a brief opinion, 606 F.2d 1068 (D.C.Cir.1979), that was superceded six months later by a much longer one, 636 F.2d 323 (D.C.Cir.1979). In the period between the two Alabama Poiver opinions, EPA proposed a new NSR regulation. The proposed definition of modification continued focusing on potential emissions rates rather than actual emissions. 44 Fed.Reg. 51,924, 51,952 (Sept. 5, 1979). After the issuance of the revised Alabama Power opinion, however, EPA changed its definition of modification. The final 1980 rule defined the term as follows: “ ‘[m]ajor modification’ means any physical change in or change in the method of operation of a major stationary source that would result in a significant net emissions increase of any pollutant subject to regulation under the Act.” 45 Fed.Reg. 52,676, 52,735 (Aug. 7, 1980) (emphasis added). The regulation defined “[n]et emissions increase” as “any increase in actual emissions from a particular physical change or change in method of operation” that occurred after taking into account, through a process known as “netting,” “any other increases and decreases in actual emissions at the source that are contemporaneous with the particular change and are otherwise creditable.” Id. at 52,736. The regulation then defined “actual emissions” as follows: (ii)In general, actual emissions as of a particular date shall equal the average rate, in tons per year, at which the unit actually emitted the pollutant during a two-year period which proceeds the particular date and which is representative of normal source operation. The Administrator shall allow the use of a different time period upon a determination that it is more representative of normal source operation. Actual emissions shall be calculated using the unit’s actual operating hours, production rates, and types of materials processed, stored, or combusted during the selected time period. (iii) The Administrator may presume that source-specific allowable emissions for the unit are equivalent to the actual emissions of the unit. (iv) For any emissions unit which has not begun normal operations on the particular date, actual emissions shall equal the potential to emit of the unit on that date. Id. at 52,737. In contrast to the proposed regulation’s approach, this regulation emphasized “actual emissions.” Justifying the shift, EPA explained in the regulation’s preamble that while the initial Alabama Power decision had used the phrase “potential to emit,” the later opinion used language that, “like the [statutory] definition, suggested] changes in actual emissions,” and that EPA had followed suit. Id. at 52,700. Finally, the 1980 regulation provided that “[a] physical change or change in the method of operation shall not include ... an increase in the hours of operation or in the production rate.” Id. at 52,735-36. Several parties petitioned this court for review of the 1980 rule, but we stayed that challenge because of ongoing settlement discussions with EPA. Ultimately, EPA and the parties entered into an agreement providing that the agency would undertake a new rulemaking and that if the new rule failed to meet certain conditions, the parties could revive their stayed petitions. In the proceedings before us today, industry petitioners and EPA dispute what the 1980 rule meant. Both agree that for a source to undertake a modification, it must first make a physical or operational change other than an increase in the hours of operation. They disagree over how to measure an “increase” in emitted pollutants once a change has occurred. According to industry petitioners, the 1980 regulation provided that an emissions “increase” occurs only if the maximum hourly emissions rate goes up as a result of the physical or operational change. According to EPA, however, an increase occurs under the 1980 regulations if, after netting, a source’s past annual emissions (typically measured by averaging out the two “baseline” years prior to the change) are less than future annual emissions (measured by calculating the source’s potential to emit after the change). EPA proffered this interpretation, which quickly became known as the “actual-to-potential” test, in proceedings leading up to Puerto Rican Cement Co. v. EPA, 889 F.2d 292 (1st Cir.1989), and Wisconsin Electric Power Co. v. Reilly, 893 F.2d 901 (7th Cir.1990) (“WEPCo”). EPA also referred to this interpretation in its preambles to later rules, see 57 Fed.Reg. 32,314, 32,328 (July 21, 1992); 67 Fed.Reg. 80,186, 80,199 (Dec. 31, 2002). Puerto Rican Cement’s facts illustrate the practical difference between industry’s and EPA’s interpretations. In that case, a factory sought to make a physical change: it would replace old cement kilns that operated 60% of the time with a new kiln that would emit fewer pollutants per hour. “If operated to achieve about the same level of production [as the old ones], the new kiln will pollute far less than the older kilns; but, if the Company operates the new kiln at significantly higher production levels, it will emit more pollutants than did the older kilns.” 889 F.2d at 293. Under the actual-to-potential test, the company “increased” its emissions after the change, making it subject to NSR: operated at full potential, the new kiln would emit more pollutants than the old kilns had emitted when actually in operation. Under the interpretation urged by industry petitioners, however, the company had not undergone an “increase” in emissions — and thus would not trigger NSR — since the new kiln would have a lower hourly emissions rate than the old ones. Siding with EPA, the First Circuit agreed that, the company had to obtain an NSR permit to make the intended change. Id. at 296-99. WEPCo, which is important because of EPA’s response to it, addressed whether EPA could apply the actual-to-potential test to utility plants undergoing extensive renovations. The petitioner argued that given the particular nature of the utility market, it was unfair to compare a utility’s past actual emissions with its future potential emissions. Instead, the petitioner argued — and the Seventh Circuit agreed— that EPA should measure future emissions by projecting future actual emissions rather than by assuming, as it had done under the actual-to-potential test, that the source would operate at full capacity in the future. 893 F.2d at 916-18. The Seventh Circuit decided WEPCo shortly before Congress enacted the 1990 amendments to the CAA. In those amendments, Congress added several programs — distinct from NSR — aimed at further securing good air quality through regulating existing sources. See generally Pub.L. No. 101-549, 104 Stat. 2399 (1990) (creating, among other things, programs aimed at reducing acid rain and at decreasing regional haze). Though it also made some changes related to NSR, Congress ultimately neither addressed the issues raised in WEPCo, see H.R. Conf. Rep. No. 101-952, at 344-45 (1990), nor revisited its statutory definition of modification, instead leaving it up to EPA to respond to that decision. EPA dealt with WEPCo by issuing a 1992 rule that changed the test utilities used for measuring emissions increases. 57 Fed.Reg. 32,314. Under the new test, known as the “actual-to-projected-actual test,” utilities would determine whether they had post-change increases in emissions' — -and thus whether they needed NSR permits' — by comparing actual emissions before the change to their projections of actual post-change emissions. See id. at 32,323-26. In measuring projected emissions, EPA permitted utilities to exclude increases stemming from unrelated demand growth, reasoning that such increases would in no way be caused by physical or operational changes. See id. at 32,326-28. The parties call this the “demand growth exclusion.” Applying the actual-to-projected-actual test and the demand growth exclusion to utilities only, EPA left the actual-to-potential test in place for other sources. Various petitioners challenged the 1992 rule, but once again we stayed the proceedings as EPA began a new rulemaking process. This new process went slowly. EPA issued a proposed rule in 1996, 61 Fed.Reg. 38,250 (July 23, 1996), followed by a 1998 Notice of Availability (“NOA”) requesting additional comment on several issues, 63 Fed.Reg. 39,857 (July 24, 1998), followed in turn by a four-year hiatus. In the meantime, EPA began investigating numerous sources for noncompliance with the existing NSR program. It ended up bringing complaints against thirty-two utilities in ten states. In 2002, EPA issued a new final rule to “reduce burden, maximize operating flexibility, improve environmental quality, provide additional certainty, and promote administrative efficiency.” 67 Fed.Reg. at 80,189. This rule departed from the prior rules in several significant respects relevant to this litigation. First, it adopted the actual-to-projected-actual test for all existing sources, id. at 80,275 (codified at 40 C.F.R. § 52.21(a)(2)(iv)(c) (2004)), though leaving sources the option to continue using the actual-to-potential test if they preferred, id. at 80,277 (codified at 40 C.F.R. § 52.21(b)(41)(ii)(d)). Second, it altered the method for measuring past actual emissions. Under the 1980 rule, sources determined past actual emissions by averaging their annual emissions during the two years immediately prior to the change, though they could use either different, more representative periods or source-specific allowable emissions levels, if they could convince the permitting authorities. In contrast, under the 2002 rule, sources other than electric utilities determine past actual emissions by averaging annual emissions of any two consecutive years during the ten years prior to the change. Id. at 80,278 (codified at 40 C.F.R. § 52.21(b)(48)(ii)). EPA determined that this change eliminated the need for case-specific alternatives. See id. at 80,200. Adopting a statement from the 1992 rule’s preamble, the 2002 rule also set a five-year lookback period for electric utilities. Id. at 80,278 (codified at 40 C.F.R. § 52.21(b)(48)(i)); see also 57 Fed.Reg. at 32,323. Third, the 2002 rule expanded the 1992 rule’s demand growth exclusion, making it applicable to all sources, not just utilities. See 67 Fed.Reg. at 80,277 (codified at 40 C.F.R. § 52.21(b)(41)(ii)(c)). Fourth, the rule provided that sources that saw no reasonable possibility that post-change emissions would prove higher than past actual emissions need keep no records of actual post-change emissions. See id. at 80,279 (codified at 40 C.F.R. § 52.21(r)(6)). Fifth, the rule set forth three specific situations in which sources, without undergoing NSR, could make changes that might otherwise constitute modifications: the Plantwide Applicability Limitations (“PAL”) program, the Clean Unit option, and the Pollution Control Project (“PCP”) exemption. The PAL program permits sources that opt in to make whatever changes they wish during the next ten years without triggering NSR,' provided that each year these sources remain below a certain level of emissions. See id. at 80,284-89 (codified at 40 C.F.R. § 52.21(aa)). Under the Clean Unit option, sources that install technology “comparable to” BACT (if in PSD regions) or LAER (if in NNSR regions) may make whatever changes they want over the next ten years without triggering NSR, provided that these changes do not cause them to exceed the “emissions limitations” set by their comparable technology. See id. at 80,279-83 (codified at 40 C.F.R. § 52.21(x)-(y)). The PCP exemption shields from NSR those sources that install technology that, though substantially reducing emissions of some pollutants, has the effect of causing increases in emissions of other pollutants. See id. at 80,275-77, 80,283-84 (codified at 40 C.F.R. §§ 52.21(b)(2)(iii)(h), 52.21(b)(32), 52.21(z)). EPA denied petitions for reconsideration on all matters of significance. 68 Fed.Reg. 63,021 (Nov. 7, 2003). Numerous petitioners now challenge the 2002 rule. Industry petitioners object to the actual-to-projected-actual test, arguing that the CAA requires EPA to compare past potential emissions with future potential emissions (i.e., use a “potential-to-potential” test). They also challenge the readings of the 1980 rule contained in the preambles to the 1992 and 2002 rules, arguing that these preambles impermissibly interpret the 1980 rule as using an actual-to-potential test rather than a potential-to-potential test. One petitioner, Newmont Mining Corporation (“Newmont”), argues that the 2002 rule is arbitrary and capricious because sources may no longer use either source-specific allowable emissions or a “more representative period” for their two-year baseline that occurred more than ten years before the proposed modification. Between them, government and environmental petitioners challenge virtually all aspects of the 2002 rule, including the use of a ten-year lookback period for selecting the two-year baseline, the use of this ten-year lookback period in the netting context, the use of a five-year look-back period for electric utilities, the demand growth exclusion, the recordkeeping standards, and the PAL, Clean Unit, and PCP provisions. They also raise several procedural challenges involving lack of notice. Environmental petitioners additionally challenge the 1992 rule’s five-year lookback period in the NSPS context. Government petitioners argue that EPA’s presumption that all states must incorporate the 2002 rule’s elements into their SIPs violates several statutory provisions. Finally, several intervenors and amici have joined the fray, attacking or defending various aspects of EPA’s rules. We consolidated these petitions and now consider them, first addressing industry petitioners’ contentions and then turning to the arguments of government and environmental petitioners. In considering these challenges, we apply a highly deferential standard of review. We may set aside a regulation only if it exceeds EPA’s “statutory jurisdiction, authority, or limitations” or is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 42 U.S.C. § 7607(d)(9). As to EPA’s interpretation of the CAA., we proceed under Chevron’s familiar two-step process. See 467 U.S. at 842-43, 104 S.Ct. 2778. In the first step (“Chevron Step 1”), we determine whether, based on the Act’s language, legislative history, structure, and purpose, “Congress has directly spoken to the precise question at issue.” Id. at 842, 104 S.Ct. 2778. If so, EPA must obey. But if Congress’s intent is ambiguous, we proceed to the second step (“Chevron Step 2”) and consider “whether the agency’s [interpretation] is based on a permissible construction of the statute.” Id. at 843, 104 S.Ct. 2778. If so, we will give that interpretation “controlling weight unless [it is] arbitrary, capricious, or manifestly contrary to the statute.” Id. at 844, 104 S.Ct. 2778. Aside from statutory interpretation, we evaluate EPA’s actions based on traditional administrative law principles. See Ethyl Corp. v. EPA, 51 F.3d 1053, 1064 (D.C.Cir.1995) (noting that the CAA’s review provisions are identical to those in the Administrative Procedure Act). “Where, as here, the issue before us requires a high level of technical expertise, we must defer to the informed discretion of the responsible federal agencies.” Transmission Access Policy Study Group v. FERC, 225 F.3d 667, 714 (D.C.Cir.2000) (internal quotation marks and citation omitted). After a “searching and careful inquiry” into the facts, Am. Trucking Ass’n v. EPA, 283 F.3d 355, 362 (D.C.Cir.2002), we will find EPA’s actions arbitrary and capricious if the agency has failed to “examine the relevant data and articulate a satisfactory explanation for its action, including a rational connection between the facts found and the choice made,” Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (internal quotation marks and citation omitted), or has reached a conclusion unsupported by substantial evidence, Ass’n of Data Processing Serv. Orgs., Inc. v. Bd. of Governors of the Fed. Reserve Sys., 745 F.2d 677, 683-84 (D.C.Cir.1984). The standard of review “does not,” however, “permit us to substitute our policy judgment for that of the Agency.” Bluewater Network v. EPA, 370 F.3d 1, 11 (D.C.Cir.2004). IL Industry Challenges Various firms and industry associations advance three main- challenges. First, industry petitioners attack the 2002 rule’s definition of “modification” for NSR purposes on the ground that it unlawfully differs from its definition for NSPS purposes. While the NSPS regulatory definition of modification allegedly focuses on the hourly rate of emissions, the NSR definition focuses on net emissions increases measured in tons per year. Compare 40 C.F.R. § 60.14 (NSPS), with id. § 52.21(b)(2)(ii)(NSR). Industry claims that this divergence is. unlawful because Congress intended to adopt for NSR purposes the NSPS regulatory ' definition in existence at the time of the 1977 amendments. (Industry petitioners also challenge the 1980 rule’s definition of modification in the NSR context to the extent that it differs from the NSPS definition.) We are not convinced. Second, industry petitioners argue that statements in the preamble to the 2002 rule constitute an unlawful interpretation of the 1980 rule. Because of multiple uncertainties about the existence or likely application of any such interpretation, let alone any burden to petitioners from delay of adjudication, we find the issue unripe. Third, the previous rules allowed states to use source-specific emissions limitations as proxies for actual emissions. 45 Fed.Reg. at 52,737 (previously codified at 40 C.F.R. § 52.21(b)(21) (1981)). Petitioner Newmont challenges the elimination of this provision in the 2002 rule, arguing that EPA’s decision lacks adequate reasoning and violates the statute. We find neither argument convincing. A. Modification. Industry rests its claim that modification must have the same regulatory meaning for NSR as prevailed for NSPS in 1977 on the fact that Congress, by a cross-reference, used the same language in both statutory contexts. Thus, the NNSR portion of the Act provided: The terms “modifications” and “modified” mean the same as the term “modification” as used in section 7411(a)(4) of this title. 42 U.S.C. § 7501(4). Similarly, the PSD portion of the statute provides that “construction” includes “the modification (as defined in section 7411(a) of this title) of any source or facility.” Id. § 7479(2)(C). So far as appears, then, these incorpo-rations by reference are the equivalent of Congress’s having simply repeated in the NSR context the definitional language used before in the NSPS context. We have (naturally) required indications in the statutory language or history to infer that Congress intended to incorporate into a statute a preexisting regulatory definition. See Continental Air Lines, Inc. v. Dep’t of Transp., 843 F.2d 1444, 1454 (D.C.Cir.1988). Industry suggests there is “abundant indication” of such intent, pointing to Congress’s having said that modification (in the NNSR portion of the statute) has the meaning of the same word “as used in” the NSPS portion of the statute. It also cites a conference committee report that explains agreement to cover modification as well as construction in Part C of the Act (PSD) (a point apparently originally excluded unintentionally) by saying that construction is being defined “to conform to mage in other parts of the Act.” See 123 Cong. Rec. 32,253 (Nov. 1, 1977) (emphasis added). But the phrases “usage” and “used in” refer not to regulatory usage, but only to usage in the statute itself. They tell us no more than if Congress had used a little more ink and repeated the NSPS definitions verbatim. Elsewhere in the Act, moreover, Congress did incorporate regulatory provisions expressly by reference. See, e.g., Pub.L. No. 95-95, § 129(a)(1), 91 Stat. 685, 745 (1977) (“the interpretative regulation of the Administrator of the Environmental Protection Agency published in 41 Federal Register 55524 ... shall apply ....”) (incorporating EPA’s offset ruling); 42 U.S.C. § 7502 note. Congress’s failure to use such an express incorporation of prior regulations for “modification” cuts against the proposed inference. Industry petitioners also invoke Bragdon v. Abbott, 524 U.S. 624, 632, 118 S.Ct. 2196, 141 L.Ed.2d 540 (1998), for the proposition that when Congress repeats a well-established term, it implies that Congress intended the term to be construed in accordance with preexisting regulatory interpretations. But that proposition does industry little good here, as the regulatory definitions in the NSPS and PSD programs already differed at the time of the 1977 amendments. See Part I, supra, at 10-12 (comparing regulatory definitions of NSPS and PSD programs). In fact, the NSPS regulations adopted in 1975 and in force at the time of the 1977 CAA amendments themselves used two different (and possibly inconsistent) definitions of modification. Section 60.2(h) defined modification to include “any physical change in, or change in the method of operation of, an existing facility which increases the amount of any air pollutant (to which a standard applies) emitted into the atmosphere by that facility.” 40 Fed.Reg. at 58,418 (previously codified at 40 C.F.R. § 60.2(h) (1976)). But 40 C.F.R. § 60.14(a) provided that “any physical or operational change to an existing facility which results in an increase in the emissions rate to the atmosphere of any pollutant to. which a standard applies shall be considered a modification,” and § 60.14(b) specified that the emissions rate should be expressed in “kg/hr of any pollutant discharged into the atmosphere.” 40 Fed.Reg. at 58,419; see also Part I, supra, at 10. Industry’s briefs, curiously, mention only § 60.14, never § 60.2(h). Given the two quite differently worded regulatory definitions of “modification” unthin the NSPS program at the time of the 1977 amendments, it would take a rather pointed indication from Congress to support the idea that it expressly adopted one of them for NSR. No such indication exists. We express no opinion as to whether Congress intended to require that EPA use identical regulatory definitions of modification across the NSPS and NSR programs. Cf. United States v. Duke Energy Corp., No. 04-1763, slip op. at 11-19 (4th Cir. June 15, 2005). That argument was not made by industry petitioners in their opening brief and is therefore waived. See Verizon Tel. Cos. v. FCC, 292 F.3d 903, 911-12 (D.C.Cir.2002). As industry makes no attack at all on the reasonableness of EPA’s definition of modification for NSR (apart from its divergence from one of the 1975 NSPS definitions), we reject this portion of industry’s challenge to the 1980 and the 2002 rules. B. Interpretation of 1980 Rule in 2002 Preamble. Industry petitioners also challenge an allegedly new interpretation of the 1980 rule contained in the preamble to the 2002 rule. Specifically, industry objects to the following sentence in the preamble: Prior to today, the regulations applied an actual-to-future-actual applicability test for EUSGUs [Electric Utility Steam Generation Units] and an actual-to-potential applicability test for all other emissions units. 67 Fed.Reg. at 80,199 (emphasis added). Industry petitioners’ claim is that by uttering the above sentence, EPA attempted to interpret the 1980 rule retroactively to require a “universal actual-to-potential test.” Such an interpretation would be, industry claims, substantively inconsistent with the 1980 rule and the Act, and in violation of various procedural requirements for amendments of agency rules. Industry petitioners raise a similar objection to the 1992 rule’s preamble. Br. for Industry Pet’rs at 29 n.46, 32 n.52. These claims are unripe. Ripeness depends on (1) the fitness of the issue for judicial review, and (2) the hardship to the parties of withholding a judicial decision. See Abbott Labs. v. Gardner, 387 U.S. 136, 148, 87 S.Ct. 1507, 18 L.Ed.2d 681 (1967). Fitness is highly questionable here, as the disputed sentence appears to be — as EPA claims — no more than a short-hand reference to the 1980 rule, not a formal interpretation. If industry’s fears should prove well-grounded, review could proceed more intelligibly on a clearer record. See Am. Iron & Steel Inst. v. EPA, 115 F.3d 979, 990 (D.C.Cir.1997); State Farm Mut. Auto. Ins. Co. v. Dole, 802 F.2d 474, 479 (D.C.Cir.1986). Nor has industry shown that delay of review will inflict any hardship. The usual form of hardship is to put a regulated firm to a choice between submission and violation, each with its attendant nonreeovera-ble costs. But the new (2002) rule has been applicable for three years now. For planning purposes the 1980 rule appears moot. If there are still pending applications of the 1980 rule in which EPA attempts to employ the disputed sentence (which seems improbable in light of its express disclaimer), judicial proceedings addressed to the application could solve the problem of any affected firm. C. Source-Specific Allowable Emissions. The. previous rules allowed state SIPs to provide for calculation of baseline emissions by using a unit’s “source-specific allowable emissions” as the unit’s actual emissions. See 45 Fed.Reg. at 52,737 (previously codified at 40 C.F.R. § 52.21(b)(21) (1981)). Petitioner Newmont challenges the elimination of this provision in the 2002 rule, arguing that EPA’s decision lacks adequate reasoning and violates the statute. EPA’s reasoning was simple enough— that the baseline is intended to be an indicator of emissions associated with utilization “actually achieved.” See EPA, TECHNICAL SuPPORT ÜOCUMENT FOR THE PREVENTION of Significant Deterioration anb Nonattainment Area New Source Review Regulations 1-3-11 (2002) (“TSD”). Otherwise changes increasing emissions beyond historic levels would avoid NSR. Id.; see also id. 1-5-9, II — 3—9. Newmont makes the counterargument that EPA’s decision imposes a foolhardy “use it or lose it” regime in which sources are encouraged to continue emitting at high levels to avoid losing the “right” to emit. A closer approximation is that the rule imposes a “use it for twenty-four months in ten years or lose it” regime, in which “lose it” entails an obligation to comply with review procedures for modifications at the source. In any event, such choices are for EPA to make so long as the agency engages in reasoned decision-making. See Bluewater Network, 370 F.3d at 11. Although EPA never expressly addressed this possibly perverse incentive, its resolute focus1 on the significance of changes in “actual” emissions suggests that it found the risk of firms’ strategic use of emissions ceilings relatively minor when compared with the benefits of catching actual increases and subjecting them to NSR. See Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 285-86, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974) (“we will uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned”); ACS of Anchorage Inc. v. FCC, 290 F.3d 403, 408 (D.C.Cir.2002). Newmont’s statutory claim is that eliminating the states’ discretion to use source-specific allowable emissions as the emissions baseline violates the Act’s principles of power sharing between the states and the federal government. Indeed the Act does have roles for both levels of government. See Virginia v. EPA, 108 F.3d 1397, 1408 (D.C.Cir.1997). While states are responsible for writing SIPs, the Act gives EPA responsibility for developing basic rules for the NSR program, see 42 U.S.C. § 7503(a)(1), a responsibility that clearly includes choosing a methodology for calculating baseline emissions. We see no violation of Congress’s assignment of duties. III. Baseline Emissions The NSR provisions of the CAA require “new and modified major stationary sources” of air pollution to obtain precon-struction permits and to install pollution control technology in order to protect and enhance air quality. 42 U.S.C. §§ 7475, 7502, 7503. An, existing source triggers NSR when it makes a “modification,” defined as: any physical change in, or change in the method of operation of, a stationary source 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). To determine whether a change “increases” emissions, the source must first calculate its baseline level of “actual emissions.” See 57 Fed.Reg. at 32,316. The 1980 rule defined “actual emissions” as “the average rate, in tons per year, at which the unit actually emitted the pollutant during a two-year period which precedes the [change] and which is representative of normal source operation.” 45 Fed.Reg. at 52,737 (codified at 40 C.F.R. § 52.21(b)(21)(ii)). The 1980 rule also provided for “the use of a different time period upon a determination that it is more representative of normal source operation.” Id. While EPA historically used the two-year period immediately preceding the change to calculate baseline actual emissions, “in some cases” it allowed use of “an earlier period.” 67 Fed.Reg. at 80,188. The 2002 rule reinterprets the term “increases” by adopting a new method for calculating baseline actual emissions. See id. at 80,191. For sources other than electric utilities, “baseline actual emissions” are defined as “the average rate, in tons per year, at which the emissions unit actually emitted the pollutant during any consecutive 24-month period selected by the [source] within the 10-year period immediately preceding [the change].” Id. at 80,278 (codified at 40 C.F.R. § 52.21(b)(48)(ii)). A source must adjust its baseline downward to reflect any legally enforceable emissions limitations that have been imposed since the baseline period, see id. (codified at 40 C.F.R. § 52.21(b)(48)(ii)(c)), and it may not use a more “representative” baseline period outside the ten-year “lookback period,” see id. at 80,195. A source may use a different baseline period for each regulated pollutant. See id. (codified at 40 C.F.R. § 52.21(b)(48)(ii)(d)). The 2002 rule also codifies the presumption established in the 1992 rule that for an electric utility, “any 2 consecutive years within the 5 years prior to the proposed change is representative of normal source operations.” 57 Fed.Reg. at 32,323; see 67 Fed.Reg. at 80,278 (codified at 40 C.F.R. § 52.21(b)(48)(i)). Government and environmental petitioners raise two sets of challenges to the ten-year lookback period. First, they contend that the ten-year lookback period reflects an impermissible interpretation of the statutory term “increases” because it allows sources to increase their emissions beyond their most recent levels without triggering NSR. Second, they contend that EPA’s selection of a ten-year lookback period is arbitrary and capricious because it contravenes the statutory purpose of protecting and enhancing air quality. For the following reasons, we conclude that petitioners’ challenges to the ten-year lookback period fail to overcome the presumption of validity afforded to EPA regulations under the CAA. See Int’l Fabricare Inst. v. EPA, 972 F.2d 384, 389 (D.C.Cir.1992). A. Statutory Interpretation. While the CAA defines a “modification” as any physical or operational change that “increases” emissions, it is silent on how to calculate such “increases” in emissions. 42 U.S.C. § 7411(a)(4). According to government petitioners, the lack of a statutory definition does not render the term “increases” ambiguous, but merely compels the court to give the term its “ordinary meaning.” See Engine Mfrs. Ass’n v. S. Coast Air Quality Mgmt. Dist., 541 U.S. 246, 124 S.Ct. 1756, 1761, 158 L.Ed.2d 529 (2004); Bluewater Network, 370 F.3d at 13; Am. Fed’n of Gov’t Employees v. Glickman, 215 F.3d 7, 10 (D.C.Cir.2000). Relying on two “real world” analogies, government petitioners contend that the ordinary meaning of “increases” requires the baseline to be calculated from a period immediately preceding the change. They maintain, for example, that in determining whether a high-pressure weather system “increases” the local temperature, the relevant baseline is the temperature -immediately preceding the arrival of the weather system, not the temperature five or ten years ago. Similarly, in. determining whether a new engine “increases” the value of a car, the relevant baseline is the value of the car immediately preceding the replacement of the engine, not the value of the car five or ten years ago when the engine was in perfect condition. EPA maintains that its choice of the ten-year lookback period is entitled to deference under Chevron Step 2 because it is based on a permissible construction of the ambiguous term “increases.” 67 Fed.Reg. at 80,199. EPA disputes the validity of government petitioners’ analogies, pointing out, for example, that if the weather system arrives in the evening, it is inappropriate to compare the nighttime temperature immediately following the arrival of the system to the daytime temperature immediately preceding the arrival of the system. The important point is that the period immediately preceding a change may not be analogous to the period following the change and thus may not yield a meaningful comparison for the purpose of determining whether the change “increases” emissions. Hence, government petitioners’ reliance on the “ordinary meaning” of “increases” fails to address a practical reality. Indeed, during oral argument, counsel for government petitioners agreed that the provision in the 1980 rule for use of a “more representative” period not immediately preceding the change is consistent with the statutory language because some flexibility is needed to account for anomalous disruptions in operations. It follows that the statutory term “increases” does not plainly and unambiguously require the baseline period to immediately precede the change. Rather, the statute is silent or ambiguous on how to calculate baseline emissions, and the issue is whether the ten-year lookback period is based on a permissible interpretation of the statute under Chevron Step 2. Under Chevron Step 2, a court must defer to the agency’s interpretation of the ambiguous statutory term if it “represents a reasonable accommodation of conflicting policies that were committed to the agency’s care by the statute.” Chevron, 467 U.S. at 845, 104 S.Ct. 2778 (quoting United States v. Shimer, 367 U.S. 374, 383, 81 S.Ct. 1554, 6 L.Ed.2d 908 (1961)). In particular, the agency’s interpretation is entitled to deference when “the regulatory scheme is technical and complex, the agency considered the matter in a detailed and reasoned fashion, and the decision involves reconciling conflicting policies.” Id. at 865, 104 S.Ct. 2778. There can be no doubt that EPA is entitled to balance environmental concerns with economic and administrative concerns, at least to a point. The Supreme Court recognized in Chevron that, in enacting the NSR program, “Congress sought to accommodate the conflict between the economic interest in permitting capital improvements to continue and the environmental interest in improving air quality,” id. at 851, 104 S.Ct. 2778, and delegated the responsibility of balancing those interests to EPA, id. at 865, 104 S.Ct. 2778. Different interpretations of the term “increases” may have different environmental and economic consequences, and in administering the NSR program and filling in the gaps left by Congress, EPA has the authority to choose an interpretation that balances those consequences. See id. at 843, 104 S.Ct. 2778. In so doing, the Supreme Court has instructed, EPA may “properly rely upon the incumbent administration’s view of wise policy to inform its judgments.” Id. at 865, 104 S.Ct. 2778. Furthermore, as there is no question that the NSR program is technical and complex, id. at 848, 104 S.Ct. 2778, EPA may properly rely on its extensive experience and expertise in administering the program. Cf. Nuclear Energy Inst., Inc. v. EPA, 373 F.3d 1251, 1296 (D.C.Cir.2004) (per curiam). Based on what EPA describes in its brief as more than twenty years of experience with the NSR program under the 1980 rule and more than “ten years of review, analysis, and communications with stakeholders,” Br. for Resp’t at 69, EPA responded to industry complaints that the 1980 rule was “too complex and burdensome” and adopted the ten-year lookback period as part of an effort to simplify and streamline the NSR program without sacrificing air quality. 61 Fed.Reg. at 38,252. Based on their own experience with the 1980 rule, state intervenors Alaska, Indiana, Kansas, Nebraska, North Dakota, South Carolina, South Dakota, Utah, and Virginia concur with' EPA’s conclusion that the NSR program has been “broken for many years and [is] long overdue to be fixed.” Br. for State Intervenors at 17. It is EPA’s position that the ten-year lookback period is based on a permissible interpretation of the CAA because it “fulfills the statutory goal of balancing economic growth with the need to protect air quality.” Br. for Resp’t at 69. According to EPA, the ten-year lookback period promotes economic growth and administrative efficiency by affording sources the flexibility to respond rapidly to market changes, focusing limited regulatory resources on changes most likely to harm the environment, and eliminating conflicts over whether a proposed baseline period is “more representative of normal source operations.” 67 Fed.Reg. at 80,191-92. At the same time, EPA believes that the ten-year lookback period protects air quality by eliminating the regulatory disincentive to make physical or operational changes that improve efficiency and reduce emissions rates. Id. at 80,192. We conclude that EPA supports these conclusions with “detailed and reasoned” analysis based on its experience and expertise. Chevron, 467 U.S. at 865,104 S.Ct. 2778. In explaining the benefits of the ten-year lookback period, EPA appropriately refers to the problems experienced under the 1980 rule. EPA notes that under the 1980 rule, establishing a representative baseline period other than the two-year period immediately preceding the change was “complex and time-consuming” and often involved “disputed judgment calls.” 61 Fed.Reg. at 38,258. EPA further notes that under the 1980 rule, sources experiencing periods of low production faced the unwelcome choice of either “surrendering capacity” by capping emissions at unrepresentative low levels or incurring the time and expense of securing NSR permits “for even small, non-excluded changes to a portion of the plant.” Id. According to industry comments on the ten-year lookback period, this dilemma discourages sources from making economically efficient and environmentally beneficial changes during periods of low production. See TSD at I-4-5, 1-4-17. Similarly, as EPA points out in its brief, government petitioner New Jersey explained in comments on the ten-year lookback period that the 1980 rule “results in a baseline that decreases each time production decreases. In other words, if economic downturn temporarily slows production at a facility for a few years, the facility’s baseline actually decreases and the facility loses operational flexibility. It also discourages facilities from voluntarily implementing pollution prevention measures.” Letter from Catherine Cowan, Assistant Comm’r, N.J. Dep’t of Envtl. Protection, to EPA (Dec. 4, 1996) (Docket A-90-37, Entry IV-D-172). EPA confirms that one “common complaint” about the 1980 rule is that sources have “limited ability to consider the operational fluctuations associated with normal business cycles when establishing baseline actual emissions unless [the] reviewing authority agrees that another period is ‘more representative of normal source operation.’ ” 67 Fed.Reg. at 80,191-92. In response to these concerns, EPA commissioned a study of the business cycles of nine major emitting industries, including charcoal production, carbon black manufacturing, Portland cement manufacturing, lime manufacturing, iron and steel manufacturing, primary copper smelting, primary aluminum production, primary zinc and lead smelting, and secondary metal production. See EasteRN ReseaRCH GRoup, InC., Business Cyoles in Major Emitting Source INdustries (1997) (“Business Cycle Study”). The study examined industry output data from 1982 to 1994 and measured each industry’s business cycle from peak to peak and from trough to trough. Id. at 1-2. Peak-to-peak cycles ranged from three to six years, and trough-to-trough cycles ranged from three to eight years. Id. at 16. Govermnent and environmental petitioners contend that the business cycle study does not support EPA’s choice of a ten-year lookback period because none of the industries in the study had business cycles longer than eight years, and the study did not consider whether emissions vary with business cycles. However, petitioners ignore the study’s conclusions that “business cycles differ markedly by industry” and that “a minimum of ten years of data is recommended to capture an entire industry cycle.” Id. Moreover, while the study did not track emissions, it did track output, which generally correlates with emissions. See 67 Fed.Reg. at 80,201; Puerto Rican Cement, 889 F.2d at 297-98. Hence, the business cycle study supports EPA’s conclusion that a ten-year lookback period “is a fair and representative time frame for encompassing a source’s normal business cycle.” 67 Fed.Reg. at 80,200. Based on “their experience over the years in implementing the NSR program,” state interve-nors agree that a ten-year lookback period is reasonable, Br. for State Intervenors at 10, and government and environmental petitioners provide no basis for the court to determine whether a particular time frame is reasonable under the CAA. Absent such an explanation, the court must defer to EPA’s policy choice because it is supported by the business cycle study and not “manifestly contrary to the statute.” Chevron, 467 U.S. at 844, 104 S.Ct. 2778. Environmental petitioners further contend that the ten-year lookback period does not ensure a representative baseline because it allows sources with shorter business cycles to choose among two or three peaks, not just the most recent one. Similarly, petitioner Newmont contends that the ten-year lookback period does not ensure a representative baseline because it fails to capture the entire business cycle of the gold industry, which it claims is longer than ten years. Newmont contends in its brief that the gold industry has not completed a full business cycle since 1980 because the price of gold has not returned to $700 per ounce. At oral argument, counsel for Newmont admitted the implausibility of this contention. Business cycles are measured from peak to peak or from trough to trough based on comparative fluctuations in output; nothing requires the peaks to reach the same level of output, much less the same price. According to Newmont’s graph of gold prices, the price of gold peaked at $500 per ounce in 1988 and 1988, and at $400 per ounce in 1990, 1994, and 2004. Thus, Newmont provides no basis for the court to conclude that the gold industry’s business cycle is longer than ten years. EPA recognizes that “business cycles differ markedly by industry,” 67 Fed.Reg. at 80,200, as the business cycle study itself indicates, see Business Cyole Study at 16. But in an effort to promote operational flexibility and administrative efficiency, EPA chose to apply a fixed ten-year look-back period to all sources in order to lend “clarity and certainty to the process” and to avoid the administrative burden of determining “representative” baselines on a case-by-case basis. 67 Fed.Reg. at 80,200; TSD at I-2-10. This policy choice, which reconciles conflicting interests in accuracy and efficiency, based on years of regulatory experience, is entitled to deference under Chevron Step 2, for petitioners fail to demonstrate that EPA’s choice is impermissible under the CAA. See Chevron, 467 U.S. at 844, 864-66, 104 S.Ct. 2778. In addition to challenging EPA’s business cycle study, environmental petitioners contend that the ten-year lookback period violates this court’s interpretation of the CAA in Alabama Power, 636 F.2d 323. Under Alabama Power and the 1980 rule, a physical or operational change constitutes a “modification” subject to NSR only if it results in a net increase in emissions; thus, a source making a change that increases emissions from one unit can “net out” of NSR based on a “contemporaneous” change that decreases emissions from another unit. See id. at 401-02; 45 Fed.Reg. at 52,736 (codified at 40 C.F.R. § 52.21(b)(3)). The court stated in Alabama Power that EPA has “discretion, within reason, to define which changes are contemporaneous,” 636 F.2d at 402, and the 1980 rule defines “contemporaneous” as within a five-year period, see 45 Fed.Reg. at 52,736 (codified at 40 C.F.R. § 52.21(b)(3)(ii)). The 2002 rule retains this definition of “contemporaneous” bu