Commonwealth of Virginia v. Environmental Protection Agency, State of Connecticut, Intervenors

108 F.3d 1397, 323 U.S. App. D.C. 368

Opinion

Opinion for the Court filed by Circuit Judge RANDOLPH.

RANDOLPH, Circuit Judge:

These are consolidated petitions for review of a final rule issued by the Environmental Protection Agency under the Clean Air Act (“CAA”). The rule is aimed at reducing ozone pollution in the northeastern United States. It requires the twelve states in the region and the District of Columbia to adopt what is essentially California’s vehicle emission program. Opposing the rule, and appearing here as petitioners, are the Commonwealth of Virginia, and three associations representing automobile manufacturers and dealers. Intervening to defend the rule are the Commonwealth of Massachusetts, the States of New York, Connecticut, Rhode Island and Vermont, the City of New York, and two associations.

Petitioners believe EPA’s rule is unsupported by the record, contrary to the statute, and constitutionally defective. The evidence regarding ozone pollution, they maintain, does not support EPA’s demand for region-wide emission reductions and, in any event, the means EPA has chosen for realizing those reductions — mandating that these states prohibit the sale of new cars that do not satisfy California’s standards — is something Congress has barred the Agency from imposing. A majority of the governors of the twelve northeastern states recommended, over the objection of Virginia and other states, that EPA impose this solution. The group operated as a regional commission under a new section of the Clean Air Act, a section petitioners say is unconstitutional because it gave the commission significant legislative and executive powers without regard to the Appointments Clause of the Constitution, the non-delegation doctrine, the Joinder and Compact Clauses and other constitutional limitations. Furthermore, in petitioners’ view EPA’s interpretation of the Act renders it unconstitutional under the Tenth Amendment because it compels the states to enact and administer a federal regulatory program.

I

A

EPA believed that if the states in the Northeast passed legislation adopting the California ear program this would help alleviate the ozone hazard. Ozone (03) in the upper atmosphere was not the problem. There it absorbs harmful ultraviolet rays. Ozone at ground level is another matter entirely. It is one of the primary constituents of smog. Ozone’s three-atom arrangement is highly unstable: the third oxygen atom, in a process called oxidation, has an aggressive tendency to react with whatever substance is available. Ozone’s high reactivity, evident in the stratosphere where ozone reacts with chlorofluorocarbons, has harmful effects at ground level. Much of the ozone inhaled reacts with sensitive lung tissues, irritating and inflaming the lungs, and causing a host of short-term adverse health consequences including chest pains, shortness of breath, coughing, nausea, throat irritation, and increased susceptibility to respiratory infections. Final Rule, 60 Fed.Reg. 4712, 4712-13 (1995). “The mechanisms of ozone-induced impairment of lung function are only partly understood,” but “the effects of a single exposure to” ozone “are reversible and last up to 48” hours. Editorial, Ozone: Too Much in the Wrong Place, The Lancet, July 27, 1991, at 221. On the other hand, “[c]hronic effects that have resulted from recurrent seasonal exposure to ozone have been studied only to a limited degree. Most of the current evidence is derived from animal responses to chrome ozone exposure.” Committee ON TROPOSPHERIC OZONE FORMATION and Measurement, National Research Council, Rethinking the Ozone PROBLEM in Urban and Regional Air Pollution 33 (1991) [hereinafter National Research Council]. In addition to its direct effect on humans, excessive ozone can also damage forests and food crops.

Exactly how ozone is created and transported in the lower atmosphere, and how it decays, is a matter of extreme complexity. Ozone is not a direct pollutant. Vehicles do not emit it, and it does not billow out of smokestacks. Instead, it is formed mostly from the mixture of two chemical precursors emitted by automobiles and industry: nitrogen oxides (NOx) and a large group of hydrocarbon pollutants called volatile organic compounds (VOCs). 1 National Research Council, supra, at 153. These precursors cook in the sun — they cook best in a “sea” of warm stagnant air — and produce ozone through a complex chain of reactions. The creation of ozone can thus be seen as a seasonal phenomenon, with concentrations peaking in the summer, and as a diurnal occurrence, with concentrations peaking during the afternoon and falling during the night. The precursor- and ozone-laden air slowly moves downwind, and as the air mass moves, ozone levels often continue to increase, in part because the ozone has more time to develop, in part because the air mass picks up more precursors along the way. Ultimately, this process can bring high ozone levels to areas hundreds of miles downwind of the pollution sources.- See id. at 100. Therein lies the source of much of the difficulty in controlling it.

At the moment, dozens of areas throughout the United States have not attained the national ambient air quality standard for ozone. The standard permits a maximum one-hour average ozone concentration of 0.12 parts per million. 40 C.F.R. § 50.9. If the ambient air exceeds this standard more than once per year averaged over three years, the area is considered in “nonattainment.” Id. The ozone season of 1994 was a good one; EPA estimated that there was twelve percent less ozone than in 1985. EPA Air Quality Trends Brochure 1994 (EPA-454/F-95-003) (1995). But the warmer 1995 season was worse, with national ozone concentrations increasing four percent from 1994. EPA Air Quality Trends Brochure 1995 (EPA-454/F-96-008) (1996). According to press reports, New Jersey and Maryland each exceeded the ambient air standard fourteen times. See Northeast Ozone Problems Continue, Show Long-Term Gain, Octane WK., Oct. 9,1995, at 2.

Because of the ozone transport phenomenon, an area in nonattainment may be unable to do much about it. Even if the state implements the strictest practicable control measures, it might still fail to satisfy the ambient air standard. Across the country there are what one might call interstate ozone transport regions: around Chicago; in the region from Cleveland to Erie, Pennsylvania; in the Texas-Louisiana area; and in the northeast from Virginia to Maine, an area that became subject to special legislation in 1990.

B

The “Northeast Ozone Transport Region,” defined in the 1990 amendments to the Clean Air Act, consists of Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Virginia, Vermont, and the District of Columbia. 2 To decrease ozone pollution in this “Region,” EPA promulgated the rule we have before us. Each of these states had an implementation plan, adopted after public hearings, containing measures to control air pollution, and approved by EPA under section 110. CAA § 110(a)(1), 42 U.S.C. § 7410(a)(1). EPA’s rule declared all of these implementation plans “substantially inadequate” and therefore in need of revision. Final Rule, 60 Fed.Reg. 4712, 4736 col. 2 (1995). The plans were inadequate, according to EPA, because nitrogen oxide and volatile organic compound emissions had to be reduced by 50% to 75% from a 1990 baseline “throughout” this densely populated Region to attain the national ambient air quality standard for ozone in “serious and severe areas.” Id. at 4720 col. 1. EPA acknowledged that it was “enormously complicated to determine which reductions” were needed where. Id. col. 2. But, it said, “wind trajectory data” supported the following conclusions. Id. col. 3. During the summer, ozone and emissions of precursors in the Washington, D.C. area contribute to the ozone problem in Baltimore, which lies to the northeast. Baltimore and the rest of Maryland contribute to ozone pollution in the Philadelphia area, which contributes to the problem in New York, which contributes to the problem in Boston, and so on up the Eastern seaboard from one metropolitan area to another. Id. at 4720-21. (Of course the northeastern section of the country is not some isolated island; and there are indications that western Pennsylvania receives ozone from Ohio and West Virginia.)

Congress, too, had found the entire subject very complex. In the 1990 amendments it ordered EPA, “in conjunction with the National Academy of Sciences, [to] conduct a study on the role of ozone precursors in tropospheric ozone formation and control. The study shall examine the roles of NOx and VOC emission reductions, the extent to which NOx reductions may contribute (or be counterproductive) to achievement of attainment in different nonattainment areas....” CAA § 185B, 42 U.S.C. § 7511f. One part of the study, completed in 1991, found that reducing NOx emissions can sometimes increase, rather than decrease, ambient ozone levels. See National RESEARCH Council, supra, at 11. “This seemingly contradictory prediction, that lowering NOx can, under certain conditions, lead to increased ozone, results from the complex chemistry involved in ozone formation in VOC-NOx mixtures_” Id. at 167.

Nonetheless, having declared the state plans inadequate, EPA required the states to revise their plans and enact a Low Emission Vehicle program, or LEV as the parties call it. In EPA’s words, the final rule orders “all of the northeastern states to adopt the. California car program to reduce significantly the pollution emitted by new cars and' light-duty trucks.” Final Rule, 60 Fed.Reg. at 4713 col. 1. To appreciate the rationale of this order one must understand that there are two, and only two, permissible sets of regulations limiting emissions from new cars sold in the United States. There are the California regulations; and there are the federal regulations, which preempt the laws of the other 49 states. CAA §§ 202, 209(a), 42 U.S.C. §§ 7521, 7543(a). The California standards — the “Low Emission Vehicle” standards — are considerably more restrictive than the federal standards and include ceilings on motor vehicle emission of nitrogen oxides, see 13 Cal. Admin.Code § 1960.1(f)(2), (g)(1), and volatile organic compounds, see 13 Cal. Admin.Code § 1960.1(f)(2), (g)(1), (g)(2), the two chief ozone precursors.

C

In finding the twelve state plans inadequate, and in requiring the states to adopt the California solution, EPA acted on the recommendation of the “Northeast Ozone Transport Commission,” a body established by section 184 of the Act. CAA § 184, 42 U.S.C. § 7511c. As part of the 1990 amendments, section 184 set up a panel consisting of the governors of the twelve states (and the Mayor of the District of Columbia) or their delegates. This panel, or Ozone Commission as we shall call it, was to develop, through majority vote, proposals for additional control measures for ozone pollution in the Region “necessary to bring any area in such region into attainment.” Id. § 7511c(e)(l). 3

After EPA receives a Commission proposal, EPA must “publish in the Federal Register a notice stating that the recommendations are available and provide an opportunity for public hearing within 90 days beginning on the receipt date”; “commence a review of the recommendations to determine whether the control measures in the recommendations are necessary to bring any area into attainment” with national standards; and “consult with members of the commission” and take into account data and comments received during the notice and comment process. Id. § 7511c(c)(2), (c)(3).

From the time of receiving the Commission’s recommendations, EPA has nine months: to “determine whether to approve, disapprove,” or approve in part and disapprove in part, the recommendation; to “notify the commission in writing” of the Administrator’s determination; and to “publish such determination in the Federal Register.” Id. § 7511c(c)(4). Absent complete approval, EPA must explain “why any disapproved additional control measures are not necessary to bring any area in such region into attainment ... or are otherwise not consistent with” the remainder of the Act, and must recommend “actions that could be taken by the commission to conform the disapproved portion of the recommendations to the requirements” of section 184. Id. § 7511c(c)(4)(i) & (c)(4)(H).

If EPA approves the recommendation, it must declare each state’s implementation plan inadequate and it must order the states to include the approved control measures in their revised plans pursuant to section 11000(5). CAA § 184(c)(5), 42 U.S.C. § 7511c(c)(5). Failure to heed EPA’s order unleashes a panoply of sanctions upon the noncomplying state. We will have more to say on the subject of sanctions later in the opinion.

EPA’s final rule in this case resulted from the section 184 process just described. In August 1993, Maine, Maryland, and Massachusetts petitioned the Ozone Commission to adopt a recommendation calling for the application of the California Low Emission Vehicle program throughout the Region. See Notice of Availability, 59 Fed.Reg. 12,914, 12,915 col. 1 (1994). After holding several “public forums,” and a hearing, the Commission voted on February 1, 1994, “to recommend that EPA mandate the California” vehicle program “throughout the” Region. Id. A minority of Ozone Commission members— those from Virginia, Delaware, New Jersey, and New Hampshire — voted against the recommendation. See id. On February 10, 1994, the Commission submitted the recommendation to EPA Id. at 12,914 col. 1.

In its final rule, EPA approved the Commission’s recommendation. 60 Fed.Reg. at 4713 col. 1. Therefore, as section 184(c)(5) required, EPA found “that the State Implementation Plans” for the twelve states in the Region “are substantially inadequate to comply with the requirements of section 110(a)(2)(D) of the Clean Air Act, 42 U.S.C. § 7410(a)(2)(D), and to mitigate adequately the interstate pollutant transport described in section 184 of the Clean Air Act, 42 U.S.C. § 7511c, to the extent that they do not provide for emission reductions from new motor vehicles in the amount that would be achieved by the” Low Emission Vehicle program. Final Rule, 60 Fed.Reg. at 4736 col. 2 (to be codified at 40 C.F.R. § 51.120(a)).

D

EPA purported to offer the states an alternative to enacting the California program. Instead of restricting the type of new cars sold within its jurisdiction, a state may choose its own mix of programs. To be an acceptable “Substitute Program,” however, it must reduce ozone precursors anywhere from 3.5 to 6.5 times more than would the California program. See infra part II.A 4

II

Of petitioners’ many challenges , to EPA’s rule, one is aimed at something known in Clean Air Act parlance as a “SIP call”— EPA’s declaration that a state’s implementation plan is substantially inadequate and must be revised. Virginia maintains that the record does not support the SIP call to it. If Virginia is correct, EPA’s regulation requiring Virginia to adopt the California car program would fall. However, the issues raised by the other petitioners — the automobile manufacturers and dealers — reach beyond Virginia. If the states implement the California program in response to EPA’s rule, these petitioners claim their members will be adversely affected throughout the Northeast Region as new car price increases take effect. 5 Furthermore, it is not clear whether EPA would have issued a SIP call to any state if EPA knew it lacked a legal basis for requiring the California program. We therefore think it best to start our analysis with questions other than the SIP call, questions relating mainly to EPA’s statutory authority.

We will discuss those questions in this order and explain why we have reached the following conclusions.. First, did EPA condition its approval of a state’s revision of its implementation plan on the state’s adoption of the California program? We think the answer is yes. Second, does section 110 give EPA the authority to condition approval of a state’s plan on the state’s adoption of control measures EPA has chosen? We conclude that section 110 does not give EPA this authority. Third, does section 184 authorize EPA to so condition its approval of plan revisions? We believe the answer is yes. Fourth, did other provisions of the Clean Air Act nevertheless bar EPA from ordering states to enact the California program? We hold that EPA was so barred and that this portion of its final rule is therefore invalid.

A

Did EPA condition approval of a state’s revised implementation plan on the state’s adoption of the California ear program? EPA tells us it did not because it offered the states a choice — they could either enact the California program or submit a “Substitute Program.” It turns out, however, that EPA’s alternative is no alternative at all.

Having determined that, at a minimum, 50 percent reductions of each precursor are necessary to bring each northeastern state into ozone attainment, 60 Fed.Reg. at 4718 col. 3, EPA required any Substitute Program to reduce precursor emissions by the difference between that 50 percent and the percentage reduction “achievable through implementation of all of the Clean Air Act-mandated and potentially broadly practicable control measures,” id. at 4737 col. 2 (to be codified at 40 C.F.R. § 51.120(d)(1) & (d)(2)). The phrase “potentially broadly practicable control measures” refers to other options — “those control measures that could potentially render the” California program “unnecessary in the” Northeast. Supplemental Notice, 59 Fed. Reg. at 48,678 col. 2. EPA has calculated that “potentially broadly practicable control measures” yield a 35.9 percent reduction in nitrogen oxides 6 and a 37 percent reduction in volatile organic compounds. Final Rule, 60 Fed.Reg. at 4722. Since none of those options (taken individually or cumulatively) could produce the necessary 50 percent reduction for precursors, EPA takes the position that all the measures are necessary, and presupposes that those measures will be included in state implementation plans submitted sometime in the future. Id. at 4723 col. 3.

Thus, a state that chooses not to adopt the California program must legislate a state implementation plan containing measures to make up this “shortfall” — the difference between the 50 percent reductions EPA says is necessary and the 35.9 percent and 37 percent that it has already targeted for nitrogen oxides and volatile organic compounds, respectively. Virginia or any other state seeking to implement a program not involving California vehicles, therefore, must pass legislation or issue regulations or orders reducing nitrogen oxide production in the year 2005 by 14.1 percent (50 percent minus 35.9 percent), and volatile organic compound production by 13 percent (50 percent minus 37 percent).

In contrast, the California program, by 2005, reduces' nitrogen oxides by 4 percent, and volatile organic compounds by 2 percent. Supplemental Notice, 59 Fed.Reg. at 48,683 col. 1. States adopting the California program need legislate nothing more. 7 Virginia, therefore, must reduce nitrogen oxides 3.5 times (14.1%/4%) and volatile organic compounds 6.5 times (13%/2%) more than do states that pass legislation incorporating the California program, and Virginia must do so by means EPA has already concluded are not practicable. 8 In sum, the Substitute Program is expected to solve a shortfall EPA itself has no idea how to remedy, a shortfall states enacting the California program are not expected to make up.

Of course, only a very foolish state would see EPA’s offer to accept this Substitute Program as a real alternative. EPA conceded as much. No state, EPA recognized, “would seriously entertain unreasonable or impracticable measures to adopt in place of the” California program “in order to achieve the necessary emissions reductions.” Supplemental Notice, 59 Fed.Reg. at 48,672 col. 2. As a practical matter, then, EPA “require[d] all the northeastern states to adopt the California car program.” Final Rule, 60 Fed.Reg. at 4713 col. 1. And that is how we shall treat the rule.

B

Does section 110 give EPA the authority to condition approval of a state’s plan on the state’s adoption of control measures EPA has chosen? EPA’s discovery that section 110 gave it this authority came late in the rulemaking proceedings. The Ozone Commission, acting pursuant to section 184, started the ball rolling. It forced ozone pollution in the Northeast to the forefront of EPA’s agenda. In EPA’s Notice of Proposed Rulemaking, the agency therefore naturally treated section 184 as the sole source of its authority to approve or disapprove the Commission’s recommendation. See 59 Fed. Reg. 21,720, 21,720-21 (1994).

The agency shifted its position after automobile manufacturers and dealers began questioning the constitutionality of section 184. See Supplemental Notice, 59 Fed.Reg. at 48,670 col. 1 & n.ll. In response to the proposed rule, some argued that EPA deference to the Ozone Commission’s technical and policy judgments would violate the Appointments Clause of the Constitution 9 and the nondelegation doctrine. Commenters also complained that the Joinder and Compact Clauses 10 prevented a multi-state alliance such as the Ozone Commission from exercising significant power. And they maintained that the proposed rule forcing the California program on the states would violate the Tenth Amendment because it would “‘commandeer’ state governments into the service of federal regulatory purposes, and would for this reason be inconsistent with the Constitution’s division of authority between federal and state governments.” New York v. United States, 505 U.S. 144, 175, 112 S.Ct. 2408, 2428, 120 L.Ed.2d 120 (1992).

In response to these comments, EPA issued a süpplemental notice stating that the arguments against section 184’s constitutionality lacked “merit,” and that “in any case, given EPA’s independent authority under section 110(k)(5), any constitutional question regarding the validity of section 184 would not affect the validity of the” proposed rule. Supplemental Notice, 59 Fed.Reg. at 48,670 col. 1. And so, not only section 184, but also section 110 supposedly provided the legal basis for EPA’s action.

C

Enacted more than a quarter of a century ago, section 110 has gone through many changes, but its basic structure has survived. The provision has been the subject of dozens of appellate court decisions. Yet we are aware of no ease (EPA has cited none) supporting the proposition EPA now urges upon us, namely, that under section 110 EPA may condition approval of a state’s implementation plan on the state’s adopting a particular control measure, here the California Low Emission Vehicle program. Despite the volumes of legislative history on the Clean Air Act and its amendments, EPA has offered no committee reports, no statements on the floor of either house, suggesting that anyone in Congress ever shared its view of section 110. Nor has EPA given us reason to believe that it has followed any consistent, longstanding practice of using section 110 to force states to adopt control measures of EPA’s choosing. The final rule provides no explanation of why EPA read section 110 as it did. The rule contains only this bare conclusion: “EPA disagrees with comments claiming that EPA lacks” authority “because section 110 does not authorize EPA to require states to adopt specific measures.... ” Final Rule, 60 Fed.Reg. at 4717 col. 1. EPA’s brief offers little more. We are told the “plain language” of the statute gives EPA this authority, and even if the language is not so plain, EPA’s reading is reasonable and deserves our deference. Final Brief for Respondent at 34. We are not persuaded, and therefore do not reach Virginia’s contention that EPA’s interpretation would render section 110 unconstitutional, in violation of the Tenth Amendment.

Congress added section 110 to the Clean Air Act in 1970. See Clean Air Amendments of 1970, Pub.L. No. 91-604, § 4, 84 Stat. 1676,1680-81 (1970). It has remained one of the key provisions of the Act. The Act authorizes EPA to promulgate national ambient air quality standards for ozone and five other pollutants. CAA §§ 108 & 109, 42 U.S.C. §§ 7408 & 7409. Areas that do not meet the minimum level of air quality mandated by these national standards are considered to be “nonattainment areas.” CAA §§ 107(d) & 171(2), 42 U.S.C. §§ 7407(d) & 7501(2). The degree of nonattainment is classified as marginal, moderate, serious, severe, or extreme. CAA § 181(a), 42 U.S.C. § 7511(a).

If a state has an area within it that EPA has classified as being in nonattainment with respect to ozone (or one of the five other regulated pollutants), the state must devise and implement a “state implementation plan” (“State Plan”). Section 110 governs the interplay between the states and EPA with respect to the formulation and approval of such State Plans. 42 U.S.C. § 7410. The basic procedure is that “each state determines an emission reduction program for its nonattainment areas, subject to EPA approval, within deadlines imposed by Congress.” Natural Resources Defense Council, Inc. v. Browner, 57 F.3d 1122, 1123 (D.C.Cir.1995).

Should a state fail to submit an implementation plan, or should its plan fail to provide the required reductions in air pollution, certain penalties — some mandatory, others at EPA’s discretion — may follow. See generally Commonwealth of Virginia v. United States, 74 F.3d 517, 520 (4th Cir.1996); Natural Resources Defense Council, 57 F.3d at 1123-25 & n. 5. The noncomplying state may, for instance, be prevented fipm spending federal highway money in nonattainment areas. See CAA §§ 110(m), 176(c) & 179(b)(1); 42 U.S.C. §§ 7410(m), 7506(c) & 7509(b)(1). This sanction becomes mandatory if the state fails to implement an adequate State Plan within 24 months of EPA’s finding that the state’s proposed plan is deficient. CAA § 179(b)(1), 42 U.S.C. § 7509(b)(1). At that same point, EPA must impose a “federal implementation plan” (“Federal Plan”) on those areas of the state in nonattainment. CAA § 110(c), 42 U.S.C. § 7410(c). The Federal Plan “provides an additional incentive for state compliance because it rescinds state authority to make the many sensitive and policy choices that a pollution control' regime demands.” Natural Resources Defense Council, 57 F.3d at 1124.

Section 110(a)(2), the' predecessor to the current section 110(k)(5), spelled out the extent of EPA’s authority over state implementation plans within this structure. 11 When enacted in 1970, the relevant portion of section 110(a)(2) read:

The Administrator shall, within four months after the date required for submission of a plan under paragraph (1), approve or disapprove such plan, or each portion thereof. The Administrator shall approve such plan, or any portion thereof, if he determines that it was adopted after reasonable notice and hearing and that—

* * *

(H) it provides for revision, after public hearings, of such plan (i) from time to time as may be necessary to take account of revisions of such national primary or secondary ambient air quality standard or the availability of improved or more expeditious methods of achieving such primary or secondary standard; or(ii) whenever the Administrator finds on the basis of information available to him that the plan is substantially inadequate to achieve the national ambient air quality primary or secondary standard which it implements.

42 U.S.C. § 1857e-5(a)(2)(H) (1970) (emphasis added). At the time, section 110(a)(3) required EPA to review a state’s revisions to its plan — which is what EPA is calling for here — by the same criteria EPA used to judge the original plan. 42 U.S.C. § 1857c-5(a)(3) (1970). In 1970 as now, if a state failed to submit a plan, or if the plan did not satisfy section 110(a)(2), EPA had a non-discretionaiy duty to implement the Act through a federal implementation plan. CAA § 110(c), 42 U.S.C. § 1857e-5(c) (1970).

In 1975, the Supreme Court analyzed section 110’s “division of responsibilities” between the states and the federal government. Train v. Natural Resources Defense Council, 421 U.S. 60, 79, 95 S.Ct. 1470, 1482, 43 L.Ed.2d 731 (1975). The narrow issue in Train was whether states could use section 110(a)(3) to seek “revisions” of State Plans in order to ease limits on a single pollution source, in essence circumventing section 110(f), which restricted the circumstances under which such a “variance” could be granted. A state could, the Court held, use section 110(a)(3) for such a purpose so long as the revised plan would still enable the state to attain and maintain national ambient standards. Train, 421 U.S. at 72, 95 S.Ct. at 1478.

A “broader issue,” the Court thought, was at stake in the case — namely, “whether Congress intended the States to retain any significant degree of control over the manner in which they attain and maintain national standards.” Id. at 78, 95 S.Ct. at 1481. The Act expressly gave the states initial responsibility for determining the manner in which air quality standards, were to be achieved. See id. at 64, 95 S.Ct. at 1474-75. Section 107(a) of the Act read then, as it does now: “Each State shall have the primary responsibility for assuring air quality within the entire geographic area comprising such State by submitting an implementation plan which will specify the manner in which national primary and secondary ambient air quality standards will be achieved and maintained within each air quality control region in such State.” CAA § 107(a), 42 U.S.C. § 7407(a) (emphasis added). In light of section 107(a), the Court construed section 110:

The Act gives the Agency no authority to question the wisdom of a State’s choices of emission limitations if 'they are part of a plan which satisfies the standards of § 110(a)(2), and the Agency may devise and promulgate a specific plan of its own only if a State fails to submit an implementation plan which satisfies those standards. § 110(e). Thus, so long as the ultimate effect of a State’s choice of emission limitations is compliance with the national standards for ambient air, the State is at liberty to adopt whatever mix of emission limitations it deems best suited to its particular situation.

Train, 421 U.S. at 79, 95 S.Ct. at 1482. The Supreme Court repeated this interpretation in Union Electric Co. v. EPA, 427 U.S. 246, 96 S.Ct. 2518, 49 L.Ed.2d 474 (1976): section 110 left to the states “the power to determine which sources would be burdened by regulations and to what extent.” Id. at 269, 96 S.Ct. at 2531. To be sure, if EPA rejected a State Plan because it would not achieve or maintain ambient air quality standards, EPA could promulgate a federal implementation plan. But even then, as EPA admitted in confessing error in the Supreme Court, section 110 did not permit the agency to require the state to pass legislation or issue regulations containing control measures of EPA’s choosing. EPA v. Brown, 431 U.S. 99, 103, 97 S.Ct. 1635, 1636-37, 52 L.Ed.2d 166 (1977) (per curiam).

In 1977 Congress again amended the Act, changing some parts of section 110. 12 But these changes did not modify the “division of responsibilities” Train had discerned in the Act. EPA’s later administrative efforts to alter the balance were therefore firmly rebuffed. Bethlehem Steel Corp. v. Gorsuch, 742 F.2d 1028 (7th Cir.1984), held that EPA could not, under the guise of partially approving a state implementation plan, render the plan more stringent than the state intended. If EPA wanted to impose stricter regulations, the Clean Air Act gave it the option of implementing a federal plan. The court explained:

[T]he Clean Air Act creates a partnership between the states and the federal government. The state proposes, the EPA disposes. The federal government through the EPA determines the ends — the standards of air quality — but Congress has given the states the initiative and a broad responsibility regarding the means to achieve those ends through state implementation plans and timetables of compli-ance_ The Clean Air Act is an experiment in federalism, and the EPA may not run roughshod over the procedural prerogatives that the Act has reserved to the states, ... especially when, as in this case, the agency is overriding state policy.

Id. at 1036-37. Florida Power & Light Co. v. Costle, 650 F.2d 579 (5th Cir.1981), is to the same effect. EPA’s “attempting to require Florida to include” a particular provision in its State Plan was “clearly an abuse of discretion; it is agency action beyond the Congressional mandate,” action that would “usurp state initiative in the environmental realm,” and “disrupt the balance of state and federal responsibilities that undergird the efficacy of the Clean Air Act.” Id. at 587, 589.

Thus, as section 110 stood in 1975 when the Supreme Court decided Train and as it stood after the 1977 amendments, the provision did not confer upon EPA the authority to condition approval of Virginia’s implementation plan, or the plan of any other state, on the state’s adoption of a specific control measure. EPA “identifies the end to be achieved, while the states choose the particular means for realizing that end.” Air Pollution Control Dist. v. USEPA 739 F.2d 1071, 1075 (6th Cir.1984). The validity of EPA’s argument that it now has broader authority depends, therefore, on whether the 1990 Clean Air Act Amendments eliminated the “liberty” every state had under the earlier legislation “to adopt whatever mix of emission limitations it deems best suited to its particular situation.” Train, 421 U.S. at 79, 95 S.Ct. at 1482.

The 1990 amendments were the most comprehensive since 1970, but the changes to section 110, at least as they concern EPA’s approval of State Plans, were predominantly of syntax, not substance. Section 110(a)(2)(H)(ii), as amended, requires each State Plan to “provide for revision” not only when EPA finds the plan “substantially inadequate to attain” national ambient standards, but also when EPA finds that the plan does not “otherwise comply with any additional requirements established under this chapter.” CAA § 110(a)(2)(H)(ii), 42 U.S.C. § 7410(a)(2)(H)(ii).

Also in the 1990 law, Congress repealed section 110(a)(3)(A), a provision requiring the Administrator to approve a State Plan revision meeting the (a)(2) criteria. The language of the repealed subsection was moved to a new subsection (k), so that the new section 110(k)(3) imposed the same obligation upon the Administrator to approve plans satisfying the (a)(2) requirements: “In the case of any” State Plan submission or revision, “the Administrator shall approve such sub-mittal as a whole if it meets all of the requirements of this chapter.” Id. § 7410(k)(3).

The 1990 amendments added subsection (k)(5), on which EPA relies in this ease. Like the original section 110(a)(2)(H)(ii), this new section 110(k)(5) gave EPA authority to require states to revise their plans if they have failed to attain or maintain national ambient air quality standards. 42 U.S.C. § 1857c-5(a)(2)(H)(2) (1970); 42 U.S.C. § 7410(k)(5) (1994). So far as EPA’s authority is concerned, the new subsection merely duplicated the language in the 1970 version of section 110(a)(2)(H)(ii). The first two sentences of section 110(k)(5) now read:

Whenever the Administrator finds that the applicable implementation plan for any area is substantially inadequate to attain or maintain the relevant national air quality standard, to mitigate adequately the interstate pollution transport described in [CAA sections 176A or 184] or to otherwise

comply with any requirement of this Act, the Administrator shall require the State to revise the plan as necessary to correct such inadequacies. The Administrator shall notify the State of the inadequacies, and may establish reasonable deadlines (not to exceed' 18 months after the date of such notice) for the submission of such plan revisions.

42 U.S.C. § 7410(k)(5).

The main differences between this portion of subsection (k)(5) and the original section 110(a)(2)(H)(ii) are that EPA may set the timetable for a state to revise its plan, and that a plan’s failure “to mitigate adequately the interstate pollution transport described in” sections 176A or 184 may trigger the need for revision. Neither of these differences, however, amounts to a new grant of authority to EPA to require states to insert in their plans control measures EPA has selected. The original act, interpreted in Train v. Natural Resources Defense Council, 421 U.S. 60, 95 S.Ct. 1470, 43 L.Ed.2d 731 (1975), permitted EPA to call upon a state to revise its plan whenever the Administrator found the plan substantially inadequate. Section 110(k)(5) does the same.

EPA apparently thinks the “as necessary” language in section 110(k)(5) altered the division of responsibilities between the states and the agency. We suppose the idea is that because section 110(k)(5) empowers EPA to “require the State to revise the plan as necessary to correct” inadequacies, it empowers EPA to require the state to include particular control measures in the revised plan. There is nothing to this.

The Supreme Court’s decision in Train provides the starting point, and is itself the endpoint for interpreting section 110, unless an “intervening development of the law has ‘removed or weakened the conceptual underpinnings from the prior decision, or ... the later law has rendered the decision irreconcilable with competing legal doctrines or policies.’ ” Neal v. United States, - U.S. -, -, 116 S.Ct. 763, 768-69, 133 L.Ed.2d 709 (1996) (quoting Patterson v. McLean Credit Union, 491 U.S. 164, 173, 109 S.Ct. 2363, 2370-71, 105 L.Ed.2d 132 (1989)). Congress’ insertion of “as necessary” in section 110(k)(5) does not affect the underpinnings of Train, nor is the Train interpretation irreconcilable with those two words. All this additional language does is keep EPA within bounds. If a state plan is inadequate — that is, if the state is not achieving an ambient air quality standard — EPA can call only for revisions “as necessary” to correct that problem. EPA cannot, in other words, require the state to go through a wholesale revision of its entire plan. In that respect, section 110(k)(5) complements section 110(a)(2)(H)(ii), the section requiring State Plans to contain procedures for revisions. We know, for the reasons spelled out above, that Congress did not give EPA authority to choose the control measures or mix of measures states would put in their implementation plans. And we can think of no reason why Congress, merely by inserting the words “as necessary,” would have meant to hand over that authority to EPA when it calls upon states to revise their implementation plans. 13 We would have to see much clearer language to believe a statute allowed a federal agency to intrude so deeply into state political processes. See Gregory v. Ashcroft, 501 U.S. 452, 463, 111 S.Ct. 2395, 2402, 115 L.Ed.2d 410 (1991).

We therefore hold that the 1990 amendments did not alter the division of responsibilities between EPA and the states in the section 110 process. It was with this understanding that we recently summarized the statutory system: “The states are responsible in the first instance for meeting the” national ambient standards “through state-designed plans that provide for attainment, maintenance, and enforcement of the” national standards “in each air quality control region. Thus, each state determines an emission reduction program for its nonattainment areas, subject to EPA approval, within deadlines imposed by Congress.” Natural Resources Defense Council, Inc. v. Browner, 57 F.3d 1122, 1123 (D.C.Cir.1995).

Because section 110 does not enable EPA to force particular control measures on the states, EPA’s authority to promulgate the rule under review must be derived from section 184 alone.

D

Does section 184 give EPA the authority to condition approval of a state’s plan on the state’s adoption of particular control measures? The language of this statute, unlike section 110, answers with an emphatic yes. By a majority vote, the Ozone Commission recommends “additional control measures” for all or part of the Region. CAA § 184(c)(1), 42 U.S.C. § 7511c(e)(l). If EPA approves the recommendation, section 184(c)(5) requires it to issue a “finding” that the implementation plans of the affected states are inadequate. “Such finding shall require each such State to revise its implementation plan to include the approved additional control measures within one year after the finding is issued.” Id. We think the only possible reading of section 184(c)(5) is that a state must revise its plan to include whatever “additional control measures” EPA has approved, or it must suffer the consequences of losing its federal highway funds, of having new factories and other “major stationary sources” of air pollution restricted, and of having the federal government impose an implementation plan in the state. See Commonwealth of Va. v. Browner, 80 F.3d 869, 881-83 (4th Cir.1996).

Virginia and the other petitioners do not quarrel with this reading of section 184. But, they say, section 184 is unconstitutional in light of the role of the Ozone Commission. We would not have to reach the constitutional questions if, as Virginia also argues, sections 202 and 177 of the Clean Air Act barred EPA from ordering states to implement the California car program. It is an old principle that if a case may be decided on statutory grounds “it will be unnecessary, and consequently improper, to pursue any inquiries, which would then be merely speculative, respecting the power of Congress in the case.” Cohens v. Virginia, 19 U.S. (6 Wheat.) 264, 441, 5 L.Ed. 257 (1821) (Marshall, C.J.); Ash- wander v. Tennessee Valley Auth., 297 U.S. 288, 347, 56 S.Ct. 466, 483, 80 L.Ed. 688 (1936) (Brandéis, J., concurring). And so we turn to those two sections.

E

Does section 177, read together with section 202, forbid EPA from conditioning its approval of a state’s implementation plan on the state’s adoption of the California program to limit motor vehicle emissions? We hold that this provision did so bar EPA and that, in this respect, the final rule is invalid.

Section 202 of the Clean Air Act, codified at 42 U.S.C. § 7521, regulates emission standards for new motor vehicles or new motor vehicle engines. The standards, covering all air pollutants that may endanger public health or welfare, preempt the laws of every state except California. CAA § 209(a) & (b), 42 U.S.C. § 7543(a) & (b). Why Congress exempted California, and how, is explained in Motor Vehicle Manufacturers Ass’n v. New York State Department of Environmental Conservation, 17 F.3d 521, 524-28 (2d Cir.1994). We will not go into the subject here.

With respect to the federal “Phase I” standards applicable to all the states except California, subparagraph (b)(3)(C) of section 202 provides: “It is the intent of Congress that the numerical emission standards specified in subsections (a)(3)(B)(ii), (g), (h), and (i) of this section shall not be modified by the Administrator after November 15, 1990, for any model year before the model year 2004.” 14 CAA § 202(b)(3)(C), 42 U.S.C. § 7521(b)(3)(C). Section 202(b)(3)(C) thus prevents EPA from prescribing, before the model year 2004, emissions standards for motor vehicles more restrictive than those already written into the statute. Accord id. § 202(f)(1), (i)(3)(A)(ii), (i)(3)(B)(ii), (i)(3)(C), (i)(3)(E). The proviso just quoted resolved a debate in the Senate about whether there should be any future rounds of heightened emission standards. See, e.g., 136 Cong. Reg S250 (daily ed. Jan. 24, 1990) (statement of Sen. Baucus), reprinted in Legislative His-TOKY OF THE CLEAN AlR ÁCT AMENDMENTS OF 1990, at 4941 (1993) [hereinafter 1990 Legislative History]. Senator Wirth, one of the sponsors of the bill, explained: ‘We do not know if the second round is going to be needed. This compromise was reached. Let us have a jump ball on the situation. Defer the question until the late 1990’s; have EPA do a study. If that study points out a second round is needed” then “in effect we in the U.S. Congress can vote it up or down.” 136 Cong. Reo. S2739 (daily ed. Mar. 20, 1990), reprinted in 1990 Legislative History, supra, at 5864.

In the meantime, section 202 required EPA to conduct a study of the “need for, and cost effectiveness of, obtaining further reductions in emissions from [ears and certain trucks], taking into consideration alternative means of attaining or maintaining the national ambient air quality standards pursuant to State implementation plans.” CAA § 202(i)(2)(A)(ii), 42 U.S.C. § 7521(i)(2)(A)(ii). EPA must report to Congress on its study no later than June 1,1997. Based on the study, EPA must then determine by December 31, 1999, whether there is a need for new, stricter standards, and whether new standards would be cost effective. CAA §§ 202(i)(2)(B) & (i)(3), 317(a)(5),'42 U.S.C. §§ 7521(i)(2)(B) & (i)(3), 7617(a)(5).

To the extent EPA’s final rule can be viewed as setting emissions standards for new motor vehicles in the northeastern states stricter than those contained in section 202, the rule is inconsistent with the statute. 15 EPA nevertheless denies that it violated § 202. States may voluntarily choose to adopt the California standards, pursuant to section 177. 16 And so, EPA’s argument goes, it is the northeastern states, not EPA, who will be deciding to impose the California program on their citizens.

There is more than a bit of inconsistency in EPA’s stance. On the one hand, in responding to Virginia’s constitutional arguments, EPA denies that the Ozone Commission has any real power. EPA tells us that it and it alone has the final say under section 184. Any resulting orders of the sort section 184(c)(5) contemplates — orders, in the statute’s words, “requir[ing] each such State to revise its implementation plan to include the approved additional control measures” — are EPA’s, not the Commission’s. But when it comes to sections 202 and 177, EPA asks us to believe that the states who implement the California program in response to EPA’s section 184 order will be doing so voluntarily. Perhaps this is true with respect to states whose governors voted for the Commission’s recommendation, although the legislatures in those states might have something to say on the subject. But no one can suppose that Virginia, whose governor voted against the recommendation, will be freely enacting these restrictive emission standards.

It is time to examine section 177 more closely. It provides that “any State ... may adopt and enforce ... standards relating to control of emissions from new motor vehicles or new motor vehicle engines” if the state standards “are identical to the California standards” and if the standards are adopted “at least two years before” they take effect. CAA § 177, 42 U.S.C. § 7507. As EPA has acknowledged, section 177 gives states the discretion to pass laws containing certain emission standards, laws that would otherwise be preempted. Supplemental Notice, 59 Fed.Reg. at 48,761 col. 1. In other words, a state may decide to follow California’s lead, or it may not.

EPA’s rule takes this choice from the states. To say, as EPA has, that it is “requiring the states to exercise their own independent authority under section 177” is to admit that EPA is depriving the states of that authority. Final Rule, 60 Fed.Reg. at 4718 col. 1. Independence signifies freedom from the dictates of a federal agency. EPA’s rule does not respect the states’ independent authority; it removes it. At several points in its final rule EPA made no bones about this. EPA said its rule would “mandate state action that would otherwise be discretionary,” id. at 4718 col. 2, and that it was “requir[ing] all the northeastern states to adopt the California car program,” id. at 4713 col. 1. And if EPA has chosen those emission standards and mandated their implementation, as it admits it has, the standards are effectively federal standards, in violation of section 177 in light of section 202.

There is another way of approaching this subject. One might treat EPA’s action, not as a requirement that each northeastern state adopt the California standards, but merely as a condition on EPA’s approval of the state’s revised implementation plan. We do not think putting the matter this way changes the outcome. For one thing, the effect of EPA’s rule remains the same despite the difference in formulation. For another, the House Committee Report on section 177 stated: “Finally, the Committee intends these provisions as grants of authority to the States. They are not intended as requirements. Nor may the Administrator require the States to utilize the authorities contained in this section as a condition of approval of a State’s implementation plan.” 17 H.R.Rep. No. 95-294, at 311 (1977), reprinted in 1977 U.S.C.C.A.N. 1077, 1390; see also 123 Cong. Reo. 16,675 (1977) (statement of Rep. Rogers) (“No one will force the State to make a judgment. It is left up to the State. They can either do it or not do it.... If a State decides to make that change to clean up the air, clean up the automobile, it can adopt and enforce the California standards which are more strict than the Federal.”); id. at 16,677 (statements of Reps. Wirth and Waxman).

EPA would have us disregard this legislative history because it is pre-1990, when the Act was amended. True enough, the report is old, but so is section 177. The 1990 amendments did not affect the substance of section 177. Congress simply added language preventing the states from prohibiting the sale or manufacture of California-certified vehicles and from taking “any action of any kind to create, or have the effect of creating, a [different type of] motor vehicle.” CAA § 177, 42 U.S.C. § 7507. As explained by Senator Baucus, the author of the Senate bill: “The new language placed in Section 177 ... neither adds to nor detracts from the present authority of the states under Section 177 to adopt, administer and enforce motor vehicle standards that are identical to those adopted by the state of California. The new language simply codifies, in effect, Congress’ understanding of the authority originally” granted “to states by section 177 as expressed in the legislative history of section 177 when it was adopted in 1977.” 136 Cong. Reo. S16,976 (daily ed. Oct. 27, 1990), reprinted in 1990 Legislative HistoRY, supra, at 1021-22.

EPA has another defense against the charge that it has violated section 202 and section 177. The argument is that in enacting section 184 and giving the agency the authority to order states to include control measures in their plans, Congress did not “exclude motor vehicle standards from the measures available to achieve attainment,” Final Brief for Respondent at 41. But this proves much too much. The fact is that section 184 excludes no sort of control measures, of which there are many, some more effective than others. A state might ration gasoline; impose tougher emission standards for boilers, gas turbines, and large internal combustion engines, Supplemental Notice, 59 Fed.Reg. at 48,679 col. 2; encourage carpooling; develop “a comprehensive system of fees and incentives designed to affect driving habits and vehicle usage,” id. at 48,680 col. 2; require “California reformulated gasoline”, id.; enhance “vehicle inspection and maintenance” programs, CAA § 182(c)(3), 42 U.S.C. § 7511a(c)(3); encourage “the voluntary removal from use and the marketplace of pre-1980 model year light duty vehicles and pre-1980 model year light duty trucks,” CAA § 108(f)(l)(A)(xvi), 42 U.S.C. § 7408(f)(l)(A)(xvi).

But among the universe of possible control measures, Congress specifically barred EPA from imposing one particular type — stricter emissions limitations on motor vehicles than those set forth in section 202. And “when a conflict arises between specific and general provisions of the same legislation, the courts should give voice to Congress’s specific articulation of its policies and preferences.” Ohio Power Co. v. FERC, 954 F.2d 779, 784 (D.C.Cir.1992). Here Congress’s policy and preference is loud and clear. It “is the intent of Congress that” EPA not modify the “numerical emission standards” for “any model year before ... 2004.” CAA § 202(b)(1)(C), 42 U.S.C. § 7521(b)(1)(C). EPA therefore may not require, mandate, order, or impose conditions demanding that any state enact particular motor vehicle emission standards, even if those standards are identical to California’s. The time will come when EPA can make its case for tougher emission limitations on motor vehicles. But that time is years from now and, under section 202, the ease must be made to Congress.

Ill

This brings us to the issue we postponed at the beginning of part II of this opinion—whether the record supported EPA’s “SIP caU.” In its final rule EPA declared that each of the state’s implementation plans were “substantially inadequate to comply with the requirements of section 110(a)(2)(D) ... and to mitigate adequately the interstate poUution transport described in section 184 ... to the extent that they do not provide for emissions reductions from new motor vehicles in the amount that would be achieved by the” California.“low emission vehicle ... program.” Final Rule, 60 Fed. Reg. at 4736 col. 2 (to be codified at 40 C.F.R. § 51.120(a)).

EPA’s decision to declare the state plans inadequate depended on its view that it could order the California remedy. Section 184 limits EPA to requiring the control measures the Ozone Commission proposes as “necessary” to cure the problem the Commission identifies. Upon finding the California program “necessary,” EPA became obligated to declare, pursuant to section 110(k)(5), that the states’ plans were inadequate under section 110(a)(2)(D). See CAA § 184(c)(5), 42 U.S.C. § 7511c(e)(5). Section 110(a)(2)(D) provides that state plans must prohibit “emissions activity”—such as driving cars—if this will produce air pollution in amounts that “contribute significantly to nonattainment” in any other state. CAA § 110(a)(2)(D), 42 U.S.C. § 7410(a)(2)(D). EPA’s reasoning was that if the California program is necessary, then the failure of a state to have that program equals a failure (under section 110(a)(2)(D)) to prohibit “significant contribution” to nonattainment in another state. See Final Rule, 60 Fed.Reg. at 4718 col. 3; see also Geoffrey L. Wilcox, New England and the Challenge of Interstate Ozone Pollution Under the Clean Air Act of 1990, 24 B.C. Envtl. Aff. L.Rev. 1, 49 (1996). Tins explains why EPA framed the SIP call as it did—the state implementation plans are inadequate “to the extent they do not provide for reductions from new motor vehicles in the amount that would be achieved” by the California program. 60 Fed.Reg. at 4736 col. 2 (to be codified at 40 C.F.R. § 51.120(a)); see also id. at 4718 col. 3, 4719 col. 1. In short, since the particular measure recommended to EPA is not one the agency can mandate, EPA’s finding of inadequacy—its SIP call— cannot survive.

Furthermore, throughout the rulemaking process, EPA operated on the belief that it had discretion to decide whether to issue a SIP call under section 110, but had no such discretion under section 184. Section 184 forced EPA to review the Ozone Commission’s recommendation, to decide if it was “necessary,” and if it was, to declare all the state plans inadequate. EPA had “no discretion”; it had to find each state plan lacking the measure deficient. But the agency .viewed section 110(k)(5) in a different light. As EPA put it, “if the Administrator were to determine ... that a particular measure is necessary to attain” the National Ambient Standards, “she might exercise her authority under section 110(k)(5) to find that a” State Plan “lacking such measure is substantially inadequate, but she might also refrain from doing so.” Final Brief for Respondent at 64. Thus, although EPA purported to make an independent finding under section 110(k)(5), it is doubtful whether the agency would have exercised the discretion it finds in that provision if section 184 had not been triggered and if the agency had realized that section 110 gave it no power to choose this particular control measure for the states.

Virginia is the only state that has challenged EPA’s rule and we have therefore considered whether to vacate the SIP call only with respect to Virginia. No one has argued for such a ruling. Furthermore, such limited relief would be inconsistent with EPA’s approach and would threaten to further complicate this proceeding. EPA’s modeling was based on an important and, it turns out, invalid assumption: that it could order each northeastern state to adopt the California program. Even if only Virginia’s SIP call were stricken, any findings under section 110 would still have to be revised. If the ozone problem does in fact snowball as it moves up the Eastern seaboard, Virginia’s lack of participation will necessarily affect all areas in the Region downwind of it, namely the rest of the entire Region. The SIP call requires states to “revise” their State Plans “as necessary to correct” them when the “Administrator finds that the applicable implementation plan ... is substantially inadequate to attain or maintain the relevant national ambient air quality standard,” or “to mitigate adequately the interstate pollutant transport described in” sections 176A or 184. CAA § 110(k)(5), 42 U.S.C. § 741000(6). In the absence of applicable modeling, no such agency finding could be made. Thus, there is no existing basis for these SIP calls.

Seven states plus the District of Columbia voted for the California program; only three states (New York, Massachusetts, and Connecticut) have enacted it. It is foreseeable that should the SIP call to other states remain in effect, some will be pressured to adopt the California program. They are, of course, free to do so on their own. But they may not be pushed into that step on the basis of an EPA order that is no longer factually supported. The automobile manufacturers and dealers, who would make and sell different and more expensive cars under the California program, would thus be harmed if we allowed the remaining SIP calls to remain in effect on what is an inadequate record.

Finally, we notice that EPA’s 1996 deadline for areas in moderate nonattainment has passed and that its 1999 deadline for areas in serious nonattainment is drawing near. See CAA § 181(a)(1) Table 1, 42 U.S.C. § 7611(a)(1) Table 1. If EPA considers it appropriate to reexamine state implementation plans in the Northeast, and if EPA requires the states to make revisions in order to bring about further reductions in ozone precursors, it must take these now-passed and rapidly approaching deadlines into account.

IV

In sum, we hold that EPA may not, under section 110, condition approval of a state’s implementation plan on the state’s adoption of a particular control measure. We also hold that EPA may not, under section 184, circumvent section 177, as we interpret it in light of section 202. For the reasons given, we hold that the SIP call EPA issued with respect to each state and the District of Columbia cannot stand.

The petitions for review are granted and the rule is vacated in its entirety.

So ordered.

1 . “Major sources of” volatile organic compounds "in the atmosphere include motor vehicle exhaust, emissions from the use of solvents, and emissions from the chemical and petroleum industries. In addition, there is now a heightened appreciation of the importance of reactive” volatile organic compounds “emitted by vegetation.” National Research Council, supra, at 1. Nitrogen oxides come “mainly from the combustion of fossil fuels; major sources include motor vehicles and electricity generating stations.” Id.

2 . 42 U.S.C. § 7511c(a). While Virginia is not among the specifically enumerated States in § 7511c(a), it is considered part of "the Consolidated Metropolitan Statistical Area that includes the District of Columbia.” Id. Accordingly, Virginia’s participation is limited to that portion of Virginia in the Washington, D.C., metropolitan area. See Supplemental Notice of Proposed Rulemaking, 59 Fed.Reg. 48,664, 48,685 col. 3 (1994) [hereinafter Supplemental Notice]. See generally U.S. Bureau of the Census, Population Estimates for Metropolitan Areas (MAs) Outside of New England: July 1, 1990 to July 1, 1995 (Mar. 8, 1996) <http://www.census.gov/esti-mates/metro-ci1y/metal95.txt> (listing counties comprising the Washington-Baltimore consolidated metropolitan statistical area).

3 . Another provision, also added in 1990, allows EPA (or the states) to establish interstate transport regions elsewhere in the country. CAA § 176A(a), 42 U.S.C. § 7506a(a). These regions may recommend strategies and plans to EPA and "may request” the agency to issue findings that a member state's implementation plan is substantially inadequate. Id. § 7506a(a) & (b)(2). An individual state, or the subdivision of a state, has the same ability to petition EPA with respect to the State Plan of a neighboring state. See CAA § 126(b), 42 U.S.C. § 7426(b). For that matter, any interested person has "the right to petition” EPA "for the issuance, amendment, or repeal” of any rule. 5 U.S.C. § 553(e). Section 184 is different. EPA must review the Ozone Commission’s recommendation, it must do so on an expedited timetable, and it must order the states to include the additional control measures in their plans if EPA approves the recommendation. CAA § 184(c)(5), 42 U.S.C. § 7511c(c)(5).

4 . EPA mentioned one other acceptable way for these states to reduce ozone precursors. The idea is to have automakers voluntarily opt into a program to meet higher emissions standards in the 49 states outside California (the "Equivalent Program”). These "national” standards would be less onerous than California’s and would enable automakers to enjoy economies of scale in their production processes. Final Rule, 60 Fed. Reg. at 4714 cols. 1-2. The program is voluntary because section 202 of the Clean Air Act forbids EPA from itself modifying motor vehicle emission standards "before the model year 2004." CAA § 202(b)(1)(C), 42 U.S.C. § 7521(b)(1)(C).

The northeastern states also would have to agree to opt into the Equivalent Program, and in so doing, to abandon the California program, or any such program an individual state had itself independently enacted. See Final Rule, 60 Fed. Reg. at 4724 col. 3; Notice of Proposed Rule-making, 60 Fed.Reg. 52,734, 52,740 col. 2 (1995).

Because the Equivalent Program is merely a proposal, we will not consider it further.

5 . Neither EPA’s final rule nor either of its notices of proposed rulemaking reach any conclusion about how EPA's ozone reduction rule will affect the price of new cars. Still, there are some estimates floating around. One is that ”[v]iewed from a cost-per-car perspective, implementing the [California] program in the Northeast would cost over $2,000 per new vehicle sold.... Viewed from an aggregate cost perspective, households in the Northeast can expect to pay $20 billion between now and the year 2010 if the [California] standards are required.” Gary Fauth, Regulating the Automobile: Learning from Cost-Effective Analysis, Bus. Econ., Oct. 1, 1994, at 23. A report prepared for EPA obtained estimates that ranged from $186.50 to $1,020 and more per car. See E.H. Pechan & Associates, Inc., Draft Report (Revised), Analysis of Costs, Benefits, and Feasibility Regarding Implementation of OTC Petition on California Low Emission Vehicles 40 (Dec. 5, 1994). The $186.50 figure is based on estimates from California regulators. See id. An independent study commissioned by the American Automobile Manufacturers Association estimated the increase at $1,020 per car; the automakers put the figure much higher. See id. at 33-34, 40.

6 . Mandatory Clean Air Act measures plus implementation of the "0.15 NOx standard for large boilers, gas turbines, and internal combustion engines would achieve approximately a 32% reduction in NOx from 1990 baseline levels.” Supplemental Notice, 59 Fed.Reg. at 48,682 col. 3. Add to that certain transportation control measures, which reduce nitrogen oxides by 2.5 percent, and the 1.4 percent reductions achievable with reformulated gasoline standards. Id. This yields a total 35.9 percent reduction.

7 . EPA asserts that states choosing the California program are expected to make up the shortfall, that the California program will be complemented by "other measures that may be promulgated thereafter." Final Brief for Respondent at 35. But this court has already rejected EPA’s "strained construction” of section 110(k)(4) to "authorize conditional approval of a” State Plan "that contains no specific substantive measures so long as it ... includes ... a promise to adopt specific enforceable measures within a year and ... a schedule of ‘interim milestones' in the future adoption process.” Natural Resources Defense Council, Inc. v. EPA, 22 F.3d 1125, 1133-34 (D.C.Cir.1994). State Plans must contain "something more than a mere promise to take appropriate but unidentified measures in the future.” Id. at 1134. Since EPA offers nothing more than wishful promises — and promises without deadlines, no less — we examine the reductions created by the measures actually undertaken by states choosing the California program.

8 . The disparity is greater still if instead of a 50 percent minimum total reduction, we use the 65 percent minimum for nitrogen oxides, the percentage EPA reported as necessary “for attainment in the serious and severe nonattainment areas of the region." Supplemental Notice, 59 Fed.Reg. at 48,682 col. 3. Using this figure, a state choosing a Substitute Program would be required to legislate solutions 7.3 times as great (29.1%/4%) for nitrogen oxides as would states implementing the California program.

9 . The Appointments Clause provides, in relevant part: The President "shall nominate, and by and with the advice and consent of the Senate, shall appoint ... all other officers of the United States, whose appointments are not herein otherwise provided for, and which shall be established by law: but the Congress may by law vest the appointment of such inferior officers, as they think proper, in the President alone, in the courts of law, or in the heads of departments.” U.S. Const, art. II, § 2, cl. 2.

10 .The Joinder Clause provides that "no new state shall be formed ... by the junction of two or more states, or parts of states, without the consent of the legislatures of the states concerned as well as of the Congress.” U.S. Const. art. IV, § 3, cl. 1. The Compact Clause forbids a state from, “without the consent of Congress,” entering "into any agreement or compact with another state.” U.S. Const, art. I, § 10, cl. 3.

11 . Sections 110(a)(2)(A) to (G) required each State Plan to include deadlines for attaining national ambient standards, id. § 110(a)(2)(A); "emission limitations,” zd. § 110(a)(2)(B); ample monitoring and analysis, id. § 110(a)(2)(C); review provisions, id. § 110(a)(2)(D); "provisions for intergovernmental cooperation” to prevent one state’s pollution from interfering with another state’s compliance with the national ambient standards, id. § 110(a)(2)(E); sufficient funding, id. § 110(a)(2)(F); and provisions for inspections of motor vehicles "to enforce compliance • with applicable emission standards,” id. § 110(a)(2)(G). 42 U.S.C. § 1857c-5(a)(2)(A) to (a)(2)(G) (1970).

12 . The 1977 amendments added section 110(a)(2)(E), which set forth a new condition for EPA’s approval of a State Plan, namely that the plan contain "adequate provisions" "prohibiting any stationary source within the State from emitting any air pollutant in amounts which will ... prevent attainment or maintenance by any other State of any such national primary or secondary ambient air quality standard.” Pub.L. No. 95-95, § 108, 91 Stat. 685, 693 (1977) (codified at 42 U.S.C. § 7410(a)(2)(E) (1982)).

The 1977 legislation also amended section 110(a)(2)(I) to require that states submit State Plans as a "precondition for permitting major new stationary sources to locate in a nonattainment area.” New England Legal Found. v. Costle, 475 F.Supp. 425, 430 (D.Conn.1979) (Jon O. Newman, J., sitting by designation), aff'd in part, 632 F.2d 936 (2d Cir.1980), and affd. in part, 666 F.2d 30 (2d Cir.1981).

13 . If EPA could dictate the control measures, why did Congress require in section 110(0 that each state hold a public hearing on proposed revisions before the state adopts them and submits them to EPA for approval?

14 . The provisions cited in subparagraph (b)(3)(C) refer, respectively to: the ceiling on nitrogen oxide emissions for heavy duty trucks, model years 1998 and thereafter, CAA § 202(a)(3)(B)(ii); Phase I tailpipe emissions standards in Table G for light-duty trucks and cars, model years 1994 and beyond, CAA § 202(g); the same, in Table H, for light duty trucks greater than 6000 lbs., CAA § 202(h); and the requirement that stricter Phase II standards for light-duty vehicles and light-duty trucks (to replace the existing Phase I standards) begin, if at all, no earlier than 2004, CAA § 202(i).

15 . EPA ordered the states to enact California’s standards contained in 13 Cal. Admin.Code § 1960.1(g)(1), which sets emissions requirements for each class of California-certified cars; and § 1960.1(g)(2), which prescribes fleet averages for hydrocarbon emissions that must be met by selling a mix of the classes of cars certified in subpart (g)(1). Together these California provi sions impose emissions limitations significantly more restrictive for non-methane hydrocarbon emissions — of which volatile organic compounds, ethane, and some exotic hydrocarbons are the constituents, see 40 C.F.R. § 51.100(s) — and nitrogen oxide emissions than those under existing federal Phase I standards. Compare 13 Cal. Admin.Code § 1960.1(g)(1) & (g)(2) with CAA § 202(g), Table G. See generally 60 Fed.Reg. at 4736 col. 3 (implementing the California provisions as part of 40 C.F.R. § 51.120(c)(l)(iii) and (c)(l)(iv)).

For example, under the California regulations, model year 2001 cars may emit, as a fleet average, 0.070 grams of non-methane compounds per mile driven, 13 Cal. Admin.Code § 1960.1(g)(2). Under federal Phase I standards, cars may not emit more than 0.25 grams per mile driven, 42 U.S.C. § 7521(g), Table G.

16 . New York, Massachusetts, and Connecticut did so before EPA issued its rule.

17 . The Senate bill did not contain any provision comparable to section 177, and the Conference Committee adopted the House provisions. See H.R. Conf. Rep. No. 95-564, at 158 (1977), reprinted in 1977 U.S.C.C.A.N. 1502, 1538.