Full opinion text
Opinion for the Court filed PER CURIAM. PER CURIAM: These consolidated petitions for review challenge various aspects of the Clean Air Interstate Rule. Because we find more than several fatal flaws in the rule and the Environmental Protection Agency (“EPA”) adopted the rule as one, integral action, we vacate the rule in its entirety and remand to EPA to promulgate a rule that is consistent with this opinion. I. Background A. Title I of the Clean Air Act Title I of the Clean Air Act (“CAA”), 42 U.S.C. §§ 7401 et seq., requires EPA to issue national ambient air quality standards (“NAAQS”) for each air pollutant that “eause[s] or contribute^] to air pollution which may reasonably be anticipated to endanger public health or welfare [and] the presence of which in the ambient air results from numerous or diverse mobile or stationary sources ...,” id. § 7408(a)(1)(A), (B). It also requires EPA to divide the country into areas designated as “nonattainment,” “attainment,” or “unclassifiable” for each air pollutant, depending on whether the area meets the NAAQS. Id. § 7407(c), (d). Title I gives states “the primary responsibility for assuring air quality” within their borders, id. § 7407(a), and requires each state to create a state implementation plan (“SIP”) to meet the NAAQS for each air pollutant and submit it to EPA for its approval, id. § 7410. If a state is untimely in submitting a compliant SIP to EPA, EPA must promulgate a federal implementation plan (“FIP”) for the state to follow. Id. § 7410(c)(1). One provision of Title I requires SIPs to contain adequate provisions—(i) prohibiting, consistent with the provisions of this subchapter, any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will—(I) contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to any [NAAQS].... 42 U.S.C. § 7410(a)(2)(D)(i)(I) (statutory provision to which we refer throughout this opinion as “section 110(a)(2)(D)(i)(I)”). In 1998, EPA relied on this provision to promulgate the NOx SIP Call, which imposed a duty on certain upwind sources to reduce their NOx emissions by a specified amount so that they no longer “ ‘contribute significantly to nonattainment in, or interfere with maintenance by,’ a downwind State.” Finding of Significant Contribution and Rulemaking for Certain States in the Ozone Transport Assessment Group Region for Purposes of Reducing Regional Transport of Ozone, 63 Fed.Reg. 57,356, 57,358 (Oct. 27, 1998) (“NOx SIP Call”). The NOx SIP Call created an optional cap- and-trade program for nitrogen oxides (“NOx”). Id. at 57,359. Like the NOx SIP Call, the Clean Air Interstate Rule—Rule To Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule); Revisions to Acid Rain Program; Revisions to the NOx SIP Call, 70 Fed.Reg. 25,162 (May 12, 2005) (“CAIR”)—which is the rule at issue in these consolidated petitions for review, also derives its statutory authority from section 110(a)(2)(D)(i)(I). B. Title IV of the Clean Air Act Title TV of the CAA, 42 U.S.C. §§ 7651-7651 o, aims to reduce acid rain deposition nationwide and in doing so creates a eap- and-trade program for sulfur dioxide (“S02 ”) emitted by fossil fuel-fired combustion devices. Congress capped S02 emissions for affected units, or electric generating units (“EGUs”), at 8.9 million tons nationwide, id. § 7651b(a)(l), and distributed “allowances” among those units. One “allowance” is an authorization for an EGU to emit one ton of S02 in a year. Id. § 7651a(3). Title IV includes detailed provisions for allocating allowances among EGUs based for the most part on their share of total heat input of all Title IV EGUs during a 1985-87 baseline period. Id. §§ 7651a(4), 7651c, 7651d, 7651e, 7651h, 7651i. Whenever an EGU emits one ton of S02 in a year, it must surrender one allowance to EPA. See id. § 7651b(g). But Title IV also permits EGUs to transfer unused allowances to deficient EGUs throughout the nation or to “bank” excess allowances and use or sell them in future years. Id. § 7651b(b). Title IV exempts EGUs that are “simple combustion turbines, or units which serve a generator with a nameplate capacity of 25 Mwe [megawatt electrical] or less,” 42 U.S.C. § 7651a(8), those that are not fossil fuel-fired, id. § 7651a(15), those that do not sell electricity, id. § 7651a(17)(A)(i), and those that cogenerate steam and electricity unless they sell a certain amount of electricity, id. § 7651a(17)(C). It also provides that certain exempt units—“qualifying small power production facilities” and “qualifying cogeneration facilities,” defined in 16 U.S.C. § 796(17)(C), (18)(B) (delegating power to FERC to define the terms), and certain “new independent power production facilities,” defined in 42 U.S.C. § 7651o (a)(1)—may elect to become a part of Title IV. 42 U.S.C. § 7651d(g)(6)(A); see id. § 7651i (detailing “electing—in” provisions). C. Clean Air Interstate Rule Pursuant to its Title I authority to ensure that states have plans in place that implement the requirements in section 110(a)(2)(D)(i)(I), EPA promulgated CAIR. CAIR, 70 Fed.Reg. at 25,165. CAIR’s purpose is to reduce or eliminate the impact of upwind sources on out-of-state downwind nonattainment of NAAQS for fine particulate matter (“PM2.5”), a pollutant associated with respiratory and cardiovascular problems, and eight-hour ozone, a pollutant commonly known as smog. Id. at 25,162. For the most part, EPA defines sources at the state level. EPA determined that 28 states and the District of Columbia (“upwind states”) contribute significantly to out-of-state downwind nonattainment of one or both NAAQS. Id. Because S02 “is a precursor to PM2i6 formation, and NOx is a precursor to both ozone and PM2.5 formation,” CAIR requires upwind states “to revise their [SIPs] to include control measures to reduce emissions” of S02 and NOx. Id. CAIR requires upwind states to reduce their emissions in two phases. Id. at 25,165. NOx reductions are to start in 2009, S02 reductions are to start in 2010, and the second reduction phase for each air pollutant is to start in 2015. Id. at 25,162. To implement CAIR’s emission reductions, the rule also creates optional interstate trading programs for each air pollutant, to which, in the absence of approved SIPs, all upwind sources are now subject. Id.; see Rulemaking on Section 126 Petition from North Carolina To Reduce Interstate Transport of Fine Particulate Matter and Ozone; Federal Implementation Plans To Reduce Interstate Transport of Fine Particulate Matter and Ozone; Revisions to the Clean Air Interstate Rule; Revisions to the Acid Rain Program, 71 Fed.Reg. 25,328, 25,328 (Apr. 28, 2006) (“FIP”) (in the absence of approved SIPs for CAIR, applying the rule’s model trading programs via EPA’s Federal Implementation Plan to all sources in upwind states). In addition, CAIR revises Title IV’s Acid Rain Program regulations governing the S02 cap-and-trade program and replaces the NOx SIP Call with the CAIR ozone-season NOx trading program. At issue in much of this litigation is the definition of the term “contribute significantly.” In other words, in order to promulgate CAIR, EPA had to determine what amount of emissions constitutes a “significant contribution” to another state’s nonattainment problem. See 42 U.S.C. § 7410(a)(2)(D)(i)(I). CAIR uses several factors to define “contribute significantly,” including one state’s impact on another’s air quality, the cost of “highly cost-effective” emissions controls, fairness, and equity in the balance between regional and local controls. CAIR, 70 Fed.Reg. at 25,-174-75. The air quality factor is the threshold step in the analysis, determining whether an upwind state is subject to CAIR, and the other factors help EPA determine the quantitative level of emissions reductions required of upwind sources. CAIR uses a different air quality threshold for each of the two pollutants it regulates. A state meets the air quality threshold for PM2.5 (and is therefore subject to CAIR) if it contributes 0.2 micrograms per cubic meter (Vg/m3”) or more of PM2.6 to out-of-state downwind areas that are in nonattainment. Id. at 25,174-75, 25,191. CAIR uses a more complicated process to define the air quality threshold for ozone NAAQS. CAIR first eliminates a state from inclusion in the CAIR ozone program if it has the following characteristics: (1) it contributes less than 2 parts per billion (“ppb”) to a nonattainment area’s ozone concentration as measured using either a “zero-out method” or a “source apportionment method,” or (2) its relative contribution to the nonattainment area’s excess ozone concentration (the number of particles exceeding 85 ppb) is less than one percent. Id. at 25,191; see also Rule to Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule): Reconsideration, 71 Fed. Reg. 25,304, 25,320 (Apr. 28, 2006) (“Reconsideration”). States that survive the screening criteria are then assessed to determine if they contribute significantly to ozone nonattainment in another state using three metrics: (1) magnitude of contribution, (2) frequency of contribution, and (3) relative amount of contribution to the area’s ozone concentration that exceeds attainment levels. CAIR, 70 Fed.Reg. at 25,191-92. States that “contribute significantly” to nonattainment for ozone NAAQS are subject to CAIR’s ozone-season limits for NOx and those that “contribute significantly” to nonattainment for PM2.5 NAAQS are subject to CAIR’s annual limits for NOx and S02. The ozone-season NOx limits are a percentage reduction in the annual limits for NOx calculated for PM25 contributors. In order to eliminate a state’s significant contribution to PM2.5 NAAQS, CAIR sets an annual cap on NOx and S02 emissions in the region. Each state participating in CAIR’s allowance-trading programs receives a budget of allowances, calculated according to a different formula for S02 and NOx. If a state develops a SIP that opts out of the trading programs to which all its upwind sources are now subject in the absence of an approved SIP, see FIP, 71 Fed.Reg. at 25,328, the state must limit its emissions to a cap specified by CAIR. CAIR sets each state’s NOx emissions budget by allocating the regionwide NOx budget among CAIR states according to each state’s proportion of oil-, gas-, and coal-fired facilities. CAIR, 70 Fed.Reg. at 25,230-31. The regionwide budget is equal to the upwind states’ average annual heat input for EGUs from 1999 to 2002 multiplied by the uniform emissions rate if EGUs were to use “highly cost-effective” emissions controls. Id. at 25,231. For Phase One, which starts in 2009, the multiplier is 0.15 pounds per million British thermal units (“Ib/mmBtu”) and for Phase Two, which starts in 2015, the multiplier is 0.125 Ib/mmBtu. Id. at 25,230. Even though EPA determined that emissions controls in both phases are “highly cost effective,” it only deemed Phase Two to eliminate the upwind states’ “significant contribution” to downwind nonattainment. Id. at 25,198. In 2009, EPA has supplemented the budget of 1.5 million tons of NOx emissions with a one-time Compliance Supplement Pool of 200,000 NOx allowances. Id. at 25,231-32. Like S02 allowances in Title IV, one CAIR NOx allowance permits an EGU to emit one ton of NOx in one year. State budgets are based on their average annual heat input, adjusted by fuel type (coal, gas, oil) during the 1999-2002 time period. Id. at 25,231. The use of fuel-adjustment factors means states with higher percentages of gas- and oil-fired facilities receive comparably fewer NOx allowances than states with higher percentages of coal-fired facilities. States have discretion to accomplish their NOx emissions caps as they see fit in their SIPs, but if a state takes part in the EPA-administered trading program for NOx, it must follow EPA’s rules for that program. CAIR sets each state’s S02 budget using a process similar to the one used for NOx budgets; it allocates the regionwide S02 budget among upwind states. However, EPA used a different method to determine the regionwide budget for S02. Instead of using 1999-2002 data, the agency summed all the Title IV allowances allotted to EGUs in the covered states and reduced them by 50% for 2010 (Phase One) and 65% for 2015 (Phase Two). Id. at 25,229. As stated above, Title TV allocates allowances among EGUs based for the most part on their share of the total heat input of all Title IV EGUs during a 1985-87 baseline period, not the later time period used for NOx allowances in CAIR. 42 U.S.C. §§ 7651a(4), 7651c, 7651d, 7651e, 7651h, 7651i. States subject to CAIR may opt into the EPA-administered trading program for S02, but if they do not opt in and at the same time choose to regulate EGUs, their SIPs must include a mechanism for retiring Title IV S02 allowances in excess of the budget CAIR allocates to each state. CAIR, 70 Fed.Reg. at 25,259. A state not participating in CAIR’s trading program but regulating other sources of S02 in addition to EGUs, does not need to surrender quite as many of its Title IV S02 allowances. Id. Any surrendered allowance may not be used for Title IV compliance purposes and is forever out of circulation. Id. at 25,291. A state does not have to surrender any Title IV S02 allowances if it adopts a SIP that regulates only non-EGUs to accomplish its S02 cap, id. at 25,295, but EPA notes that EGUs are projected to contribute 70% of S02 emissions in 2010, id. at 25,214, making such a scenario unlikely. EPA issued two additional rules clarifying CAIR that are also under review in this proceeding. One rule responds to various petitions for reconsideration, which are discussed in more detail below. Reconsideration, 71 Fed.Reg. 25,304. Another rule, inter alia, sets forth a FIP to regulate EGUs until upwind states implement EPA-approved SIPs that conform with CAIR requirements. FIP, 71 Fed.Reg. 25,328. D. Petitions for Review Section 307 of the CAA requires petitions for judicial review of CAIR to be filed within 60 days of the rule’s publication in the Federal Register. 42 U.S.C. § 7607(b)(1). On May 12, 2005, EPA published CAIR and on April 28, 2006, EPA published its Reconsideration and FIP, which describes the Federal Implementation Plan required of sources while states formulate their SIPs. CAIR, 70 Fed.Reg. 25,162; Reconsideration, 71 Fed.Reg. 25,-304; FIP, 71 Fed.Reg. 25,328. In the 60 days after EPA published CAIR and its Reconsideration, several petitions for review were filed in this Court. Among those petitions are North Carolina’s objections to EPA’s trading programs, EPA’s interpretation of the “interfere with maintenance” language in section 110(a)(2)(D)(i)(I), Phase Two’s 2015 compliance date, the NOx Compliance Supplement Pool, EPA’s interpretation of “will” in “will contribute significantly,” and the air quality threshold for PM26. Several electric utility companies (“S02 Petitioners”) contest EPA’s authority under Title I and Title IV to limit the number of Title IV allowances in circulation, to set state S02 budgets as percentage reductions in Title IV allowances, and to require units exempt from Title IV to acquire Title IV allowances. Petitioners Entergy Corporation and FPL Group, to which we refer as “Entergy,” contest EPA’s authority to base state NOx budgets on the number of coal-, oil-, and gas-fired facilities a state has compared to other states in the CAIR region. Electric utilities operating in Texas, Florida, and Minnesota and one municipality argue against the inclusion of all or part of those States in CAIR. And Florida Association of Electric Utilities petitions for review of EPA’s 2009 start date for Phase One of NOx restrictions. We consider these petitions below. II. Analysis Our jurisdiction derives from the CAA, which also establishes our standard of review. We “may reverse any such action found to be ... arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law; ... [or] in excess of statutory jurisdiction, authority, or limitations, or short of statutory right....” 42 U.S.C. § 7607(d)(9). We refer to the review standard in 42 U.S.C. § 7607(d) instead of the similar standard of review set forth in the Administrative Procedure Act (“APA”) because the CAA directs that its review standard apply to “such ... actions as the Administrator may determine.” Id. § 7607(d)(l)(V); see Supplemental Proposal for the Rule To Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule), 69 Fed.Reg. 32,684, 32,686 (June 10, 2004) (applying section 307(d), 42 U.S.C. § 7607(d), “to all components of the rulemaking”). The petitions under review involve EPA’s construction of the CAA, a statute it administers. Where the statute speaks to the direct question at issue, we afford no deference to the agency’s interpretation of it and “must give effect to the unambiguously expressed intent of Congress.” Chevron U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). But where the statute does “not directly address[] the precise question at issue, ... the question for the court is whether the agency’s answer is based on a permissible construction of the statute,” and we only reverse that determination if it is “arbitrary, capricious, or manifestly contrary to the statute.” Id. at 843, 104 S.Ct. 2778. An action is “arbitrary and capricious” if it has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise. 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); see Motor Vehicle Mfrs. Ass’n. v. EPA, 768 F.2d 385, 389 n. 6 (D.C.Cir.1985) (noting that “the standard we apply (i.e., whether the EPA’s actions were in excess of statutory authority or arbitrary and capricious) is the same under” the CAA and the APA). A. North Carolina Issues Petitioner North Carolina challenges CAIR’s programs for pollution-trading, EPA’s interpretation of the “interfere with maintenance” provision in section 110(a)(2)(D)(i)(I), the 2015 compliance deadline for Phase Two of CAIR, the NOx Compliance Supplement Pool, EPA’s interpretation of the word “will” that precedes “contribute significantly” in section 110(a)(2)(D)(i)(I), and EPA’s use of a 0.2 |xg/m3 air quality threshold for including upwind states in CAIR’s PM2,5 program. We grant North Carolina’s petition as to the trading programs, the “interfere with maintenance” language, and the 2015 compliance deadline, deny its petition as to its interpretation of “will” and the air quality threshold, and take no action on the NOx Compliance Supplement Pool issue. 1. Pollution-Trading Programs North Carolina challenges the lawfulness of CAIR’s trading programs for S02 and NOx. North Carolina contests the lack of reasonable measures in CAIR to assure that upwind states will abate their unlawful emissions as required by section 110(a)(2)(D)(i)(I), but does not submit that any trading is per se unlawful. EPA designed CAIR to eliminate the significant contribution of upwind states, as a whole, to downwind nonattainment. CAIR, 70 Fed.Reg. at 25,195. EPA did not purport to measure each state’s significant contribution to specific downwind nonattainment areas and eliminate them in an isolated, state-by-state manner. Reasoning that capping emissions in each state would not achieve reductions in the most cost-effective manner, EPA decided to take a re-gionwide approach to CAIR and include voluntary emissions trading programs. In modeling the CAIR ... EPA assumes interstate emissions trading. While EPA is not requiring States to participate in an interstate trading program for EGUs, we believe it is reasonable to evaluate control costs assuming States choose to participate in such a program since that will result in less expensive reductions. Id. at 25,196. In CAIR’s trading system, states are given initial emissions budgets, but sources can choose to sell or purchase emissions credits from sources in other states. As a result, states may emit more or less pollution than their caps would otherwise permit. Because EPA evaluated whether its proposed emissions reductions were “highly cost effective,” at the regionwide level assuming a trading program, it never measured the “significant contribution” from sources within an individual state to downwind nonattainment areas. Using EPA’s method, such a regional reduction, although equivalent to the sum of reductions required by all upwind states to meet their budgets, would never equal the aggregate of each state’s “significant contribution” for two reasons. State budgets alone, without trading, would not be “highly cost effective.” And although EPA has measured the “air quality factor” to include states in CAIR, it has not measured the unlawful amount of pollution for each upwind-downwind linkage. “As noted earlier in the case of S02, EPA recognizes that the choice of method in setting State budgets, with a given regionwide total annual budget, makes little difference in terms of the levels of resulting regionwide annual S02 and NOx emissions reductions.” Id. at 25,230-31. Thus EPA’s apportionment decisions have nothing to do with each state’s “significant contribution” because under EPA’s method of analysis, state budgets do not matter for significant contribution purposes. But according to Congress, individual state contributions to downwind nonattainment areas do matter. Section 110(a)(2)(D)(i)(I) prohibits sources “within the State ” from “contributfing] significantly to nonattainment in ... any other State ...” (emphasis added). Yet under CAIR, sources in Alabama, which contribute to nonattainment of PM2 6 NAAQS in Davidson County, North Carolina, would not need to reduce their emissions at all. See CAIR, 70 Fed.Reg. at 25,247 tbl. VI-8. Theoretically, sources in Alabama could purchase enough NOx and S02 allowances to cover all their current emissions, resulting in no change in Alabama’s contribution to Davidson County, North Carolina’s non-attainment. CAIR only assures that the entire region’s significant contribution will be eliminated. It is possible that CAIR would achieve section 110(a)(2)(D)(i)(I)’s goals. EPA’s modeling shows that sources contributing to North Carolina’s nonattainment areas will at least reduce their emissions even after opting into CAIR’s trading programs. 71 Fed.Reg. at 25,344-45. But EPA is not exercising its section 110(a)(2)(D)(i)(I) duty unless it is promulgating a rule that achieves something measurable toward the goal of prohibiting sources “within the State” from contributing to nonattainment or interfering with maintenance “in any other State.” In Michigan v. EPA, 213 F.3d 663 (D.C.Cir.2000), we deferred to EPA’s decision to apply uniform emissions controls to all upwind states despite different levels of contribution of NOx to nonattainment areas caused by the differing quantities of emissions produced in upwind states and the varying distances of upwind sources to downwind nonattainment areas. Id. at 679. We did so because these effects “flow[] ineluctably from the EPA’s decision to draw the ‘significant contribution’ line on a basis of cost differentials” and “[o]ur upholding of that decision logically entails upholding this consequence.” Id. But the flow of logic only goes so far. It stops at the point where EPA is no longer effectuating its statutory mandate. In Michigan we never passed on the lawfulness of the NOx SIP Call’s trading program. Id. at 676 (“Of course we are able to assume the existence of EPA’s allowance trading program only because no one has challenged its adoption.”). It is unclear how EPA can assure that the trading programs it has designed in CAIR will achieve section 110(a)(2)(D)(i)(I)’s goals if we do not know what each upwind state’s “significant contribution” is to another state. Despite Michigan’s approval of emissions controls that do not correlate directly with each state’s relative contribution to a specific downwind nonattainment area, CAIR must include some assurance that it achieves something measurable towards the goal of prohibiting sources “within the State” from contributing to nonattainment or interfering with maintenance in “any other State.” Because CAIR is designed as a complete remedy to section 110(a)(2)(D)(i)(I) problems, as EPA claims, FIP, 71 Fed.Reg. at 25,340, CAIR must do more than achieve something measurable; it must actually require elimination of emissions from sources that contribute significantly and interfere with maintenance in downwind nonattainment areas. To do so, it must measure each state’s “significant contribution” to downwind nonattainment even if that measurement does not directly correlate with each state’s individualized air quality impact on downwind nonattainment relative to other upwind states. See Michigan, 213 F.3d at 679. Otherwise, the rule is not effectuating the statutory mandate of prohibiting emissions moving from one state to another, leaving EPA with no statutory authority for its action. Whether EPA could promulgate a section 110(a)(2)(D)(i)(I) remedy that would bar alternate relief, such as would be available under section 126, 42 U.S.C. § 7426, is a question that is not before the court. 2. “Interfere With Maintenance” Section 110(a)(2)(D)(i)(I) requires EPA to ensure that SIPs “contain adequate provisions” prohibiting sources within a state from emitting air pollutants in amounts which will “contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to any [NAAQS].” 42 U.S.C. § 7410(a)(2)(D)(i)(I) (emphasis added). North Carolina argues that EPA unlawfully ignored the “interfere with maintenance” language in section 110(a)(2)(D)(i)(I), divesting it of independent effect in CAIR. It contends that instead of limiting the beneficiaries of CAIR to downwind areas that were monitored to be in nonattainment when EPA promulgated CAIR and were modeled to be in non-attainment in 2009 and 2010, when CAIR goes into effect, CAIR, 70 Fed.Reg. at 25,244, EPA should have also included in CAIR upwind states, such as Georgia, that send pollution into downwind areas that are projected to barely meet attainment levels of NAAQS in 2010. North Carolina only contests EPA’s interpretation of the “interfere with maintenance” prong as applied to EPA’s determination of which states are beneficiaries of CAIR for the ozone NAAQS. North Carolina explains that even though all of its counties are projected to attain NAAQS for ozone by 2010, several of its counties are at risk of returning to nonattainment due to interference from upwind sources. Specifically, it notes that Mecklenburg County, which projections show will have ozone levels of 82.5 ppb in 2010 (2.5 ppb below the 85.0 ppb NAAQS) without help from CAIR, could fall back into nonattainment because of the historic variability in the county’s ozone levels. Technical Support Document for the Final Clean Air Interstate Rule, Air Quality Modeling, at Appendix E (March 2005) (“Technical Support Document”). EPA has stated that “historical data indicates that attaining counties with air quality levels within 3 ppb of the standard are at risk of returning to nonattainment.” EPA, Corrected Response to Significant Public Comments on the Proposed Clean Air Interstate Rule, at 148 (April 2005) (“Corrected Response”). “The information also indicates that even if CAIR receptors were to [be] 3-5 ppb below the standard, they would have a reasonable likelihood of returning to nonattainment.” Id. And in the case of Fulton County, Georgia, EPA determined that the “interfere with maintenance” provision justified imposing controls on upwind states in 2015 even though it is projected to attain the NAAQS by a margin of 7 or 8 ppb because its ozone levels have varied by at least that margin several times in the recent past. Id. at 150. North Carolina argues that EPA must utilize this “historic variability” standard to determine which downwind areas suffer interference with their maintenance in 2010, not just 2015. If it did so, EPA would see that Mecklenburg County, North Carolina, has varied by at least 3 ppb (the relevant margin between attainment and nonattainment for that county in 2010) six times in the recent past and consequently would include in CAIR any state, such as Georgia, that is contributing an unlawful amount of pollution to this downwind area. Id. at 1042. EPA contends that it interpreted “interfere with maintenance” just as it did in the NOx SIP Call, in which it gave the term a meaning “much the same as” the one given to the preceding phrase, “contribute significantly to nonattainment.” CAIR, 70 Fed. Reg. at 25,193 n. 45. EPA maintains that “the ‘interfere with maintenance’ prong may come into play only in circumstances where EPA or the State can reasonably determine or project, based on available data, that an area in a downwind state will achieve attainment, but due to emissions growth or other relevant factors is likely to fall back into nonattainment.” Id. In the NOx SIP Call, it meant that areas monitored to be in attainment when that rule was promulgated but which were modeled to be in nonattainment in 2007, when the rule went into effect, were considered downwind areas with which upwind sources’ emissions interfered. NOx SIP Call, 63 Fed.Reg. at 57,379. EPA states it gave effect to the “interfere with maintenance” prong in CAIR by using it as a basis for implementing further emissions reductions in Phase Two of CAIR, by which time some downwind states will have attained NAAQS. CAIR, 70 Fed. Reg. at 25,195. First, we note that we did not consider EPA’s interpretation of “interfere with maintenance” in Michigan. Thus any interpretation it used in that rulemaking cannot provide support for EPA’s contention that its current interpretation, even if identical to that in the NOx SIP Call, comports with the statute. So we analyze EPA’s interpretation of “interfere with maintenance” for the first time here. Despite using “interfere with maintenance” as a justification for imposing further emissions controls in 2015, CAIR gave no independent significance to the “interfere with maintenance” prong of section 110(a)(2)(D)(i)(I) to separately identify upwind sources interfering with downwind maintenance. Under EPA’s reading of the statute, a state can never “interfere with maintenance” unless EPA determines that at one point it “contribute[d] significantly to nonattainment.” EPA stated clearly on two occasions “that it would apply the interfere with maintenance provision in section 110(a)(2)(D) in conjunction with the significant contribution to nonattainment provision and so did not use the maintenance prong to separately identify upwind States subject to CAIR.” FIP, 71 Fed.Reg. at 25,337 (citing CAIR, 70 Fed.Reg. at 25,193); see also Corrected Response, at 63. EPA reasoned that this interpretation “avoid[s] giving greater weight to the potentially lesser environmental effect” and strikes “a reasonable balance between controls in upwind states and in-state controls.” FIP, 71 Fed.Reg. at 25,337. EPA stated that an interpretation that permitted states that are able to attain NAAQS on their own to benefit from CAIR “could even create a perverse incentive for downwind states to increase local emissions.” Id. All the policy reasons in the world cannot justify reading a substantive provision out of a statute. See Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 485, 121 S.Ct. 903, 149 L.Ed.2d 1 (2001). Areas that find themselves barely meeting attainment in 2010 due in part to upwind sources interfering with that attainment have no recourse under EPA’s interpretation of the interference prong of section 110(a)(2)(D)(i)(I).2010 is not insignificant because that is the deadline for downwind areas to attain ozone NAAQS. See 42 U.S.C. § 7511 (setting forth deadlines for attaining ozone NAAQS). An outcome that fails to give independent effect to the “interfere with maintenance” prong violates the plain language of section 110(a)(2)(D)(i)(I). The provision at issue is written in the disjunctive: SIPs must “contain adequate provisions prohibiting ... any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will contribute significantly to nonattainment in, or interfere with maintenance by, any other State....” 42 U.S.C. § 7410(a)(2)(D)(i)(I) (emphasis added). “Canons of construction ordinarily suggest that terms connected by a disjunctive be given separate meanings, unless the context dictates otherwise .... ” Reiter v. Sonotone Corp., 442 U.S. 330, 339, 99 S.Ct. 2326, 60 L.Ed.2d 931 (1979). There is no context in section 110(a)(2)(D)(i)(I) directing an alternate result; therefore EPA must give effect to both provisions in the statute. EPA contends in its brief that CAIR is just one step in carrying out its section 110(a)(2)(D)(i)(I) duties, hinting that it may later choose to give independent effect to the “interfere with maintenance” language. There is some general language in the record to support this contention. See CAIR, 70 Fed.Reg. at 25,175 (“This overall plan is well within the ambit of EPA’s authority to proceed with regulation on a step-by-step basis.”). But more specific language in the rule belies this claim. “The [section 110(a)(2)(D)(i)(I) ] violation is eliminated once a State adopts a SIP containing the CAIR trading programs (or a SIP containing other emission reduction options meeting the requirements specified in CAIR), or EPA promulgates a FIP to achieve those same reductions.” FIP, 71 Fed.Reg. at 25,340. Because EPA describes CAIR as a complete remedy to a section 110(a)(2)(D)(i)(I) violation and does not give independent significance to the “interfere with maintenance” language to identify upwind states that interfere with downwind maintenance, it unlawfully nullifies that aspect of the statute and provides no protection for downwind areas that, despite EPA’s predictions, still find themselves struggling to meet NAAQS due to upwind interference in 2010. For this reason, we grant North Carolina’s petition on this issue. Although North Carolina challenged CAIR on the “interfere with maintenance” issue only with regard to ozone, the rule includes the same flaw with regard to PM2.5. The court does not address North Carolina’s separate contention that EPA failed to comply with notice-and-comment requirements regarding its proposed test for an “interfere with maintenance” violation, or the propriety of the test itself. 3. 2015 Compliance Deadline North Carolina argues that the 2015 deadline for upwind states to eliminate their “significant contribution” to downwind nonattainment ignores the plain language of section 110(a)(2)(D)(i), 42 U.S.C. § 7410(a)(2)(D)®, contradicts EPA’s goal of “balancing] the burden for achieving attainment between regional-scale and local-scale control programs,” CAIR, 70 Fed.Reg. at 25,166, violates the Supreme Court’s holding that EPA may not consider economic and technological infeasibility when approving a SIP, Union Elec. Co. v. EPA, 427 U.S. 246, 96 S.Ct. 2518, 49 L.Ed.2d 474 (1976), and departs from the contrary approach it took in the NOx SIP Call without explanation, NOx SIP Call, 63 Fed.Reg. at 57,449. North Carolina challenges the 2015 Phase Two deadline for upwind states to come into compliance with CAIR as incompatible with section 110(a)(2)(D)(i)(I)’s mandate that SIPs contain adequate provisions prohibiting significant contributions to nonattainment “consistent with the provisions of [Title I].” 42 U.S.C. § 7410(a)(2)(D)(i)(I). Title I dictates the deadlines for states to attain particular NAAQS. PM2.5 attainment must be achieved “as expeditiously as practicable, but no later than 5 years from the date such area was designated nonattainment ... except that the Administrator may extend the attainment date ... for a period no greater than 10 years from the date of designation as nonattainment....” 42 U.S.C. § 7502(a)(2)(A). North Carolina, along with the rest of the CAIR states, must meet PM2.6 NAAQS by 2010. See 40 C.F.R. § 81.301 et seq. Ozone nonattainment areas must attain permissible levels of ozone “as expeditiously as practicable,” but no later than the assigned date in the table the statute provides. 42 U.S.C. § 7511. North Carolina’s statutory deadline is June 2010, but it could be even sooner if EPA upon repromulgating its regulations sets an earlier deadline. See S. Coast Air Quality Mgmt. Dist. v. EPA, 472 F.3d 882 (D.C.Cir.2006). North Carolina argues that despite the statutory mandate that section 110(a)(2)(D)®, 42 U.S.C. § 7410(a)(2)(D)®, be consistent with the rest of Title I, which requires compliance with PM2,5 and ozone NAAQS by 2010, CAIR gives states that “contribute significantly” to nonattainment until 2015 to comply based solely on reasons of feasibility. CAIR, 70 Fed.Reg. at 25,177; see also Corrected Response, at 58, 61; CAIR, 70 Fed.Reg. at 25,222-25 (citing feasibility restraints such as the difficulty of securing project financing and the limited amount of specialized boilermaker labor to install controls). EPA contends that the phrase “consistent with the provisions of [Title I]” does not require incorporating Title I’s NAAQS attainment deadlines into CAIR. It argues that section 110(a)(2)(D)(i)(I) does not mandate any particular time frame and that the language about consistency only requires EPA to make a rule consistent with procedural provisions in Title I, not substantive ones. It comes to this conclusion because the phrase “consistent with the provisions of this title” follows the word “prohibiting.” Due to this placement, EPA argues that the phrase requiring consistency only modifies the word “prohibiting.” EPA does not explain how it jumps from this observation to the conclusion that a phrase modifying the word “prohibiting” can only refer to procedural requirements. The word “procedural” is simply not in the statute. If there were any ambiguity as to Congress’s intent in excluding the limiting language EPA proposes, an examination of the relevant language in the context of the whole CAA dispels any doubts as to its meaning. In the CAA, Congress differentiates between requiring consistency with provisions in a title and requiring consistency “with the procedures established” under a title. Compare 42 U.S.C. § 7410(a)(2)(D)®, with id. § 7661b(c) (emphasis added). Section 110(a)(2)(D)®, 42 U.S.C. § 7410(a)(2)(D)®, is not limited to procedural provisions in Title I; thus it requires EPA to consider all provisions in Title I—■ both procedural and substantive—and to formulate a rule that is consistent with them. Despite section 110(a)(2)(D)(i)’s requirement that prohibitions on upwind contributions to downwind nonattainment be “consistent with the provisions of [Title I],” EPA did not make any effort to harmonize CAIR’s Phase Two deadline for upwind contributors to eliminate their significant contribution with the attainment deadlines for downwind areas. 42 U.S.C. § 7410(a)(2)(D)®. As a result, downwind nonattainment areas must attain NAAQS for ozone and PM2.5 without the elimination of upwind states’ significant contribution to downwind nonattainment, forcing downwind areas to make greater reductions than section 110(a)(2)(D)(i)(I) requires. Because EPA ignored its statutory mandate to promulgate CAIR consistent with the provisions in Title I mandating compliance deadlines for downwind states in 2010, we grant North Carolina’s petition challenging the 2015 Phase Two deadline. We need not address petitioner’s other arguments against this provision. EPA justified the deadline partly on the basis that additional reductions will be required through the year 2015 in order to satisfy the “interfere with maintenance” provision of the statute. Although this may be a valid reason to require maintenance-based emissions reductions beyond the year 2010, EPA does not explain why it did not coordinate the final CAIR deadline to provide a sufficient level of protection to downwind states projected to be in nonattainment as of 2010. 4. NOx Compliance Supplement Pool North Carolina contends that the NOx Compliance Supplement Pool of 200,-000 tons defies section 110(a)(2)(D)(i)(I)’s mandate to eliminate the significant contribution of upwind sources to downwind NAAQS nonattainment and that the Compliance Supplement Pool is an arbitrary exercise of power that contradicts EPA’s own record findings. Under CAIR without the Compliance Supplement Pool, states can only begin to bank CAIR NOx allowances in 2009, the year in which Phase One of the CAIR NOx limits go into effect. The Compliance Supplement Pool gives states an incentive to make emissions cuts early; states that can show “surplus” NOx emissions reductions in 2007 and 2008 can receive bankable (and tradeable) credits for those reductions. CAIR, 70 Fed.Reg. at 25,285. The 200,000 NOx credits are apportioned to states in accordance with their share of the 2009 regionwide NOx budget. Id. at 25,286. States may distribute the credits to sources based on “(1) [a] demonstration by the source to the State of NOx emissions reductions in surplus of any existing NOx emission control requirements; or (2) a demonstration to the State that the facility has a ‘need’ that would affect electricity grid reliability.” Id. EPA created the Compliance Supplement Pool to “mitigat[e] some of the uncertainty regarding the EPA projections of resources to comply with CAIR” and to “provide! ] incentives for early, surplus NOx reductions.” Id. North Carolina first argues that the Compliance Supplement Pool is unlawful because it permits states to emit NOx in excess of the 1.5 million ton annual regional NOx cap, which EPA measured to be the upwind states’ significant contribution to downwind nonattainment in the years 2009 to 2014. See CAIR, 70 Fed.Reg. at 25,210. EPA contends that North Carolina’s argument is flawed. EPA based its measurement of upwind states’ “significant contribution” on the level of reductions that would be “highly cost effective” in 2015, not 2009. The Phase One deadline is simply EPA’s measurement of the reductions that would be feasible by 2009; it is not an independent measurement of “significant contribution” in that year. See id. at 25,-177. Thus any emissions that exceed the 1.5 million ton level due to the extra 200,-000 allowances from the Compliance Supplement Pool do not affect the elimination of upwind states’ “significant contribution.” The elimination of upwind states’ significant contribution will not happen until Phase Two’s 2015 deadline. Because we grant North Carolina’s petition that CAIR’s Phase Two deadline of 2015 is unlawful, we will not pass judgment on the lawfulness of the Compliance Supplement Pool. As EPA explains, it created the Compliance Supplement Pool under the assumption that 2015 was an appropriate deadline for CAIR compliance. It is not. EPA does not argue that it can set a level of emissions that is an upwind state’s “significant contribution” and then allow that state to exceed it. On remand, EPA must determine what level of emissions constitutes an upwind state’s significant contribution to a downwind nonattainment area “consistent with the provisions of [Title I],” which include the deadlines for attainment of NAAQS, and set the emissions reduction levels accordingly. 5. EPA’s Definition of “Will” in “Will Contribute Significantly” North Carolina contends that EPA altered its definition of “will” from a term that meant certainty in the NOx SIP Call to one that denotes the future tense in CAIR and that EPA made this change without any explanation. See 42 U.S.C. § 7410(a)(2)(D)(i)(I). North Carolina also argues that EPA’s interpretation of “will” violates the plain text of the statute. As a result, EPA did not consider upwind states for consideration in CAIR that contributed to monitored (or “certain”) nonattainment in North Carolina counties at the time EPA promulgated CAIR; EPA only included upwind states that contributed to projected nonattainment in 2010. In the NOx SIP Call, EPA stated “that the term ‘will’ means that SIPs are required to eliminate the appropriate amounts of emissions that presently, or that are expected in the future [to], contribute significantly to nonattainment downwind.” NOx SIP Call, 63 Fed.Reg. at 57,375. This isolated phrase provides some support for North Carolina’s contention that EPA considered upwind states that contributed to monitored nonattainment at the time it was promulgating the NOx SIP Call to be subject to the rule even if those states did not contribute to projected nonattainment in 2007, the year the rule went into effect. However, EPA later in the same rulemaking explained its approach to measuring nonattainment in more detail: In determining whether a downwind area has a nonattainment problem under the 1-hour standard to which an upwind area may be determined to be a significant contributor, EPA determined whether the downwind area currently has a nonattainment problem, and whether that area would continue to have a nonattainment problem as of the year 2007 assuming that in that area, all controls specifically required under the CAA were implemented, and all required or otherwise expected Federal measures were implemented. If, following implementation of such required CAA controls and Federal measures, the downwind area would remain in nonat-tainment, then EPA considered that area as having a nonattainment problem to which upwind areas may be determined to be significant contributors. Id. at 57,377. In the NOx SIP Call, EPA interpreted “will” to indicate sources that presently and at some point in the future “will” contribute to nonattainment. Because the NOx SIP Call was to go into effect in 2007, that rule used 2007 as the relevant future year for measuring nonat-tainment. This approach is identical to the one EPA took in CAIR. Because CAIR goes into effect in 2009 and 2010 respectively, those are the future years used in the measurement. See CAIR, 70 Fed.Reg. at 25,241. North Carolina’s claims about an arbitrary change in EPA’s interpretation of “will” are unfounded because there was no change. And because “will” can mean either certainty or indicate the future tense, it was reasonable for EPA to choose to give effect to both interpretations of the word. Simply because CAIR does not include states based upon present-day violations that will be cured by 2010 does not mean that EPA may ignore present-day violations for which there may be another remedy, such as relief pursuant to section 126, 42 U.S.C. § 7426. Therefore we deny North Carolina’s petition on this issue. 6. PM2.5 Contribution Threshold North Carolina argues that EPA acted arbitrarily by proposing an air quality threshold for PM2.5 at 0.15 (xg/m3 but finally settling on an air quality threshold of 0.2 |xg/m3. The air quality threshold for PM2.e is the amount of PM2.5 that sources in a state must contribute to a downwind nonattainment area to be regulated as an upwind state in CAIR’s PM2.5 program. North Carolina also challenges EPA’s decision to truncate, rather than round, the numbers it compared to the threshold. As a result, states that contributed 0.19 p,g/m3 or less to a downwind nonattainment area were not linked with North Carolina by CAIR. EPA contests North Carolina’s standing to raise this issue. It notes that only two states would be affected if EPA were to use the 0.15 |xg/m3 threshold. Illinois, which is already subject to CAIR’s requirements for PM2.6 contributions, would be subject to the exact same requirements for an additional reason—its contributions to Catawba County, North Carolina. Technical Support Document, at Appendix H. This additional upwind-downwind “link” would not change any of Illinois’s duties under CAIR; therefore it would not change any effects felt by Catawba County, North Carolina. The lower threshold would also subject Arkansas to CAIR’s PM2.5 controls. CAIR, 70 Fed.Reg. at 25,-191; Technical Support Document, at 42 tbl. YII-1. EPA states that Arkansas does not contribute at threshold levels to nonat-tainment in North Carolina, but it cites no record support for this assertion. North Carolina has standing to raise this issue for three reasons. First, if in repromulgating CAIR to comply with section 110(a)(2)(D)(i)(I), EPA removes or modifies its interstate trading options, Illinois would be barred outright from contributing significantly to North Carolina’s nonattainment areas. Second, EPA does not provide support for its assertion that Arkansas does not contribute to nonattainment areas in North Carolina because it never modeled the State. North Carolina claims that models for sources in Louisiana, Missouri, and Texas, which are further from North Carolina than those in Arkansas, show that Arkansas contributes at the 0.15 |xg/m3 threshold to nonattainment areas in North Carolina. Third, because EPA designed CAIR to be a complete statutory remedy, whether North Carolina is linked with Illinois by CAIR under section 110(a)(2)(D)(i)(I) is likely to affect related remedies that North Carolina may have against Illinois, for example, pursuant to section 126, 42 U.S.C. § 7426. Although we cannot anticipate what a new rule will look like, there is a “substantial probability” that a favorable decision by this court would redress the injury North Carolina asserts. Because North Carolina has demonstrated an injury-in-fact caused by the rule it is challenging which a favorable decision by this Court could likely remedy, we can turn to the merits of North Carolina’s petition. North Carolina notes that EPA first considered a threshold of 0.1 fxg/m3. NPR, 69 Fed.Reg. at 4584. In the Notice of Proposed Rulemaking, EPA stated that a 0.1 |xg/m3 threshold “is the smallest one that can make the difference between compliance and violation of the NAAQS for an area very near the NAAQS....” Id. EPA then decided that it is “on balance, more appropriate to adopt a small percentage value of the standard level” and chose the percentage of the NAAQS standard of 15.0 jx/m3 that is closest to 0.1 jxg/m3, which was one percent. Id. One percent of 15.0 |xg/m3 is 0.15 jxg/m3, so EPA initially chose that number as the threshold. Id. However, EPA then “requested] comments on the use of higher or lower thresholds for this purpose.” Id. In CAIR, EPA finally settled on a threshold value of 0.2 jxg/m3. It did so because EPA was “persuaded by commentersf] arguments on monitoring and modeling that the precision of the threshold should not exceed that of the NAAQS,” which only measure PM26 concentration to the tenths column. CAIR, 70 Fed.Reg. at 25,191; see id. at 25,190 (com-menters). North Carolina believes it was arbitrary for EPA to round 0.15 |xg/m3 up to 0.2 |xg/m3 instead of reverting to the earlier 0.1 |xg/m3 number that “is the smallest one that can make the difference between compliance and violation of the NAAQS.” See NPR, 69 Fed.Reg. at 4584. EPA did not explain why it chose the larger number instead of the smaller number in the final rule; it only explained why it chose a number that ended at the tenths column. CAIR, 70 Fed.Reg. at 25,191. Based on EPA’s reasoning in the Notice of Proposed Rulemaking, it may have made more sense to return to the 0.1 |xg/m3 threshold instead of “[r]ounding the proposal value of 0.15,” which is what it did. See id. But EPA was concerned that the 0.15 |xg/m3 threshold it originally proposed was too low, requesting comments on “the use of higher or lower thresholds.” NPR, 60 Fed.Reg. at 4584. And in raising the threshold number, EPA was responding to comments citing concerns about the “measurement precision of existing PM25 monitors.” CAIR, 70 Fed.Reg. at 25,190. We cannot say in this circumstance that EPA’s decision to round the 0.15 (xg/m3 threshold to 0.2 |xg/m3 instead of reverting to the original threshold considered of 0.1 |xg/m3 was wholly unsupported by the record. Likewise, we cannot say that EPA’s decision to truncate rather than round the PM2.5 contribution levels it compared to the 0.2 |xg/m3 threshold was arbitrary. The parties dispute which C.F.R. provision applies to the number it compares to the threshold-one mandating rounding, 40 C.F.R. pt. 50, App. N, § 4.3(a) (preferred by petitioner), or another mandating truncating, 40 C.F.R. pt. 50, App. N § 3.0(b) (preferred by EPA). The number EPA compares to the threshold, which is measured as “the average of annual means [of PM2.b contribution] from three successive years,” is the contribution of PM25 from one upwind state to a nonattainment area. CAIR, 70 Fed.Reg. at 25,190. Section 4.3(a) applies to annual PM26 standard design values. Design values “are the metrics (i.e., statistics) that are compared to the NAAQS levels to determine compliance.” 40 C.F.R. pt. 50 App. N § 1.0(c). Design values are composed of the average of annual means of PM2.5 for three consecutive years, 40 C.F.R. pt. 50 App. N § 4.1(b), but design values are measurements of PM2.5 levels in a stationary area—not levels of PM2.5 moving from one area to another. Because the contribution level is not a design value, section 4.3(a)’s rounding mandate does not apply. Similarly, section 3.0(b)’s truncation mandate applies to PM2i6 hourly and daily measurement data and says nothing about the contribution level EPA is assessing in CAIR. Without a rule mandating any particular method, EPA is free to round or truncate the numbers it is comparing to the 0.2 gg/m3 threshold as long as its choice is reasonable. EPA chose to truncate numbers because the “truncation convention for PM2.5 is similar to that used in evaluating modeling results in applying the ozone significance screening criterion of 2 ppb in the NOx SIP call and the CAIR proposal, as well as today’s final action.” CAIR, 70 Fed.Reg. at 25,191 n. 42 (internal citation omitted). EPA’s choice to truncate the numbers is reasonable. As a result, we deny North Carolina’s petition challenging the 0.2 |xg/m3 threshold and EPA’s choice to truncate the numbers compared to it. B. S02 and NOx Budgets S02 Petitioners and petitioner Entergy challenge CAIR’s budgets for the S02 and NOx trading programs. EPA set states’ S02 budgets for 2010 to 50% (35% in 2015) of the allowances the states’ EGUs receive under Title IV. S02 Petitioners argue EPA never explained how these budgets related to section 110(a)(2)(D)(i)(I)’s mandate of prohibiting significant contributions to downwind non-attainment. Therefore, they claim, the budgets and the regionwide cap, are “arbitrary, capricious, ... or otherwise not in accordance with law,” 42 U.S.C. § 7607(d)(9)(A). As for NOx, EPA reduced states’ budgets to the extent their EGUs burned oil or gas. Entergy claims EPA made this adjustment purely in the interests of fairness—an improper reason under section 110(a)(2)(D)(i)(I). We grant the petitions, agreeing EPA chose the budgets for both pollutants in an improper manner. In short, the fact that S02 and NOx are precursors to ozone and PM2.6 pollution does not give EPA plenary authority to reduce emissions of these substances. Section 110(a)(2)(D)(i)(I) obligates states to prohibit emissions that contribute significantly to nonattainment or interfere with maintenance downwind, and EPA must exercise its authority under this provision to make measurable progress towards those goals. 1. S02 Budgets We first address EPA’s choice of S02 budgets. EPA claims to have based state budgets for S02 and NOx on the amount of emissions sources can eliminate by applying controls EPA deems “highly cost-effective controls”—an approach EPA says we approved in Michigan v. EPA, 213 F.3d 663 (D.C.Cir.2000). We observe initially that state S02 budgets are unrelated to the criterion (the “air quality factor”) by which EPA included states in CAIR’s S02 program. Significant contributors, for purposes of inclusion only, are those states EPA projects will contribute at least 0.2 gg/m8 of PM2.5 to a nonattainment area in another state. While we would have expected EPA to require states to eliminate contributions above this threshold, EPA claims to have used the measure of significance we mentioned above: emissions that sources within a state can eliminate by applying “highly cost-effective controls.” EPA used a similar approach in deciding which states to include in the NOx SIP Call, which Michigan did not disturb since “no one quarrelfed] either with its use of multiple measures, or the way it drew the line at” the inclusion stage. 213 F.3d at 675. Likewise here, the S02 Petitioners do not quarrel with EPA drawing the line at 0.2 fji,g/m3 or its different measure of significance for determining states’ S02 budgets. Again, we do not disturb this approach. Even so, EPA’s method in setting the S02 budgets is not what Michigan approved. In that case, the petitioners argued section 110(a)(2)(D)(i)(I) does not permit EPA to consider the cost of reducing ozone. After reconciling petitioners’ shifting (and somewhat conflicting) arguments, we answered a well-defined question: Could EPA, in selecting the “significant” level of “contribution” under section 110(a)(2)(D)(i)(I), choose a level corresponding to a certain reduction cost? Michigan, 213 F.3d at 676-77. Answering that question in the affirmative, we held EPA may “after [a state’s] reduction of all [it] could ... cost-effectively eliminate[ ],” consider “any remaining ‘contribution’ ” insignificant. Id. at 677, 679. Michigan also rejected claims that applying a uniform cost-criterion across states was irrational because both smaller and larger contributors had to make reductions achievable by the same highly cost-effective controls. This, we said, “flow[ed] ineluctably from the EPA’s decision to draw the ‘significant contribution’ line on a basis of cost.” Id. at 679. Upholding that decision “logically entailfed] upholding this consequence.” Id. And while EPA’s approach did not necessarily ensure “aggregate health benefits” at roughly the lowest cost, EPA researched alternatives, and found none that significantly improved air quality or reduced cost. Id. Since no one offered a “material critique” of this research, we did not upset EPA’s judgment. Id. Here, EPA did not use cost in the manner Michigan approved. Even worse, EPA’s choice of S02 budgets does not track the requirements of section 110(a)(2)(D)(i)(I). That much is evident from EPA’s decision to base the budgets on allowances states’ EGUs receive under Title IV. Those allowances are not, as EPA asserts, a “logical starting point” for setting CAIR’s S02 emissions caps, CAIR, 70 Fed.Reg. at 25,229. Congress designed the Title IV allowance scheme using EGU data from 1985 to 1987 to address the national acid rain problem. Nowhere does EPA explain how reducing Title IV allowances will adequately prohibit states from contributing significantly to downwind nonattainment of the PM2.6 NAAQS. And while “Congress chose a policy of not revisiting and revising these allocations and, apparently, believed that its allocation methodology would be appropriate for future time periods,” Reconsideration, 71 Fed.Reg. at 25,308, it is unclear how the quantitative number of allowances created by 1990 legislation to address one substance, acid rain, could be relevant to 2015 levels of an air pollutant, PM2.5. EPA also explains that it chose Title IV as a starting point “to preserve the viability and emissions