The petitioners seek review of decisions of the Environmental Protection Agency ("EPA") implementing the 1990 Amendments to the Clean Air Act ("CAA"), 42 U.S.C. Secs. 7401 et seq. First, petitioner Natural Resources Defense Council, Inc. ("NRDC") challenges the EPA's policy of permitting conditional approval of committal state implementation plan ("SIP") submissions. In addition, all petitioners challenge substantive agency decisions in the EPA's final rule on vehicle inspection and maintenance programs ("I/M"); the NRDC challenges the rule's effective and implementation dates for enhanced I/M, its geographic scope for both basic and enhanced I/M, and several aspects of the performance standard adopted for I/M, while petitioners National Automobile Dealers Association, et al. ("NADA") challenge the EPA's use of notice and comment rulemaking to issue its I/M guidance and the rule's preference for centralized vehicle emission testing. We address the challenges separately.1I. CONDITIONAL APPROVAL OF SIPS

First, the NRDC challenges the EPA's use of a "conditional approval" procedure, under section 110(k)(4) of the 1990 Amendments, Pub.L. No. 101-549, tit. I, Sec. 110(k)(4), 104 Stat. 2399, 2407 (codified at 42 U.S.C. Sec. 7410(k)(4)), to permit states to comply with statutory SIP deadlines by submitting "committal" SIPs that contain no specific remedial measures but merely promise to adopt such measures within a year. The NRDC contends this procedure is contrary to congressional intent and has impermissibly postponed the statutory deadlines for the affected SIP submittals. To remedy the delay caused by the EPA's conditional approval procedure, the NRDC asks that we require prompt submission and review of all overdue SIPs and immediate imposition of statutory sanctions on states that have not submitted adequate SIPs as of July 15, 1994. While we hold that the EPA misconstrued and misapplied section 110(k)(4), we nevertheless conclude that equity and practicality require that we approve the extensions in part and that we adopt more moderate remedial measures, as set out below.

Since it was amended in 1970, the CAA has required states to adopt, after reasonable notice and public hearings and approval by the EPA, SIPs designed to attain and maintain "national ambient air quality standards" ("NAAQS"). See 42 U.S.C. Sec. 7410(a)(2)(A) (added by Pub.L. No. 91-604, Sec. 4(a), 84 Stat. 1680 (1970)). Not until 1990, however, did Congress establish specific deadlines for submitting SIPs and SIP revisions. As amended in 1990, the CAA now requires states encompassing "nonattainment" areas2 to submit particular SIPs or SIP revisions by fixed deadlines. Two of the deadlines are significant here: (1) November 15, 1990, the date on which the Amendments took effect and "immediately after" which date states were required thereunder to submit "basic" I/M submissions, see 42 U.S.C. Sec. 7511a(a)(2)(B)(i), (b)(4), and (2) November 15, 1992, the date by which states were required to submit SIPS addressing enhanced I/M, see id. Sec. 7511a(c)(3), and application of "reasonably available control technology" ("RACT") to stationary emission sources of nitrogen oxides ("NO subx "), see id. Sec. 7511a(b)(2), (f).3

Under the 1990 Amendments, the "basic" I/M SIP submittals, due immediately after November 15, 1990 for all ozone nonattainment areas, from "marginal" to "extreme,"4 were to "include[ ] any provisions necessary to provide for a vehicle inspection and maintenance program of no less stringency than that of either the program defined in House Report Numbered 95-294, 95th Congress, 1st Session, 281-291 (1977) as interpreted in guidance of the Administrator issued pursuant to section 7502(b)(11)(B) of this title (as in effect immediately before November 15, 1990) or the program already included in the plan, whichever is more stringent." 42 U.S.C. Sec. 7511a(a)(2)(B)(i), (b)(4), (c), (d), (e).

The "enhanced" I/M submittals, due by November 15, 1992 for "serious," "severe," and "extreme" ozone nonattainment areas and for certain metropolitan areas in "Ozone Transport Regions,"5 id. Secs. 7511a(c)(3)(A), (d), (e), 7511c(b)(1)(A), were to "comply in all respects with guidance published in the Federal Register" by the EPA "[w]ithin 12 months after November 15, 1990," that is, by November 15, 1991, one year before the enhanced I/M SIP deadline. Id. Sec. 7511a(c)(3)(A)-(B), (a)(2)(B)(ii).6 That guidance was to include a "performance standard achievable by a program combining emission testing, including on-road emission testing, with inspection to detect tampering with emission control devices and misfueling for all light-duty vehicles and all light-duty trucks" and "program administration features necessary to reasonably assure that adequate management resources, tools, and practices are in place to attain and maintain the performance standard." Id. Sec. 7511a(c)(3)(B).

The NO subx RACT submittals, also due by November 15, 1992,7 but for "moderate" to "extreme" areas only, were to include "provisions to require the implementation of reasonably available control technology" for "major stationary sources" of NO subx emissions except for certain sources regarding which the EPA determines, at the time of SIP approval, that emission reductions will harm, or at least not improve, air quality. Id. Sec. 7511a(b)(2), (f).

The 1990 CAA Amendments also established statutory teeth to enforce the new SIP deadlines. Initially, section 110(k)(1)(B) requires the EPA to make a finding that the submittal is complete or incomplete within 2 months of submission or 6 months of the submission deadline. Id. Sec. 7410(k)(1)(B). If the EPA determines a submittal is incomplete, it is deemed not to have been made. Id. Sec. 7410(k)(1)(C). If the EPA finds the submittal complete, the agency has 12 months to approve, disapprove (in part or whole) or conditionally approve the submittal. Id. Sec. 7410(k)(2)-(4).8 At any time after the EPA makes a whole or partial disapproval or an incompleteness finding, it "may apply" the two sanctions set out in 42 U.S.C. Sec. 7509(b), namely, withholding approval of highway projects and grants and imposing stricter permit requirements for new emission sources. Id. Sec. 7410(m). If a state fails to correct a deficiency within 18 months after a disapproval or finding of failure to submit, the EPA is required to impose one of the statutory sanctions. 42 U.S.C. Sec. 7509(a). Further, unless a complete SIP is approved within 2 years after such disapproval or finding, the state will be subject to an EPA-promulgated Federal Implementation Plan. 42 U.S.C. Sec. 7410(c)(1).

Although the statutory scheme set out above provides the EPA with clear submission and sanction schedules, the agency nevertheless strayed from Congress's directions in two relevant respects. First, the EPA waited until November 5, 1992 to publish the enhanced I/M guidance required under the Amendments to be provided no later than November 15, 1991. Second, in part to compensate for its dilatory promulgation of that guidance, the EPA decided to invoke the CAA conditional approval provision to afford states additional time, beyond the statutory submission deadlines, to develop and submit some of the required SIPs, notably here, those relating to I/M and NO subx RACT.

The EPA first indicated its intent to permit conditional approval of the I/M submittals, both basic and enhanced, in its April 16, 1992 "General Preamble for the Implementation of Title I of the Clean Air Act Amendments of 1990" ("General Preamble"), 57 Fed.Reg. 13,498 (April 16, 1992). See 57 Fed.Reg. at 13,514 ("The EPA will use its authority under the new section 110(k)(4) to conditionally approve basic I/M programs in the case of moderate ozone areas that were newly subject to this requirement at the time of enactment, based upon the State's commitment to develop such a program within 1 year from conditional plan approval, or by the date established EPA's guidance [sic], whichever is sooner."); id. at 13,519 ("In the event that EPA's enhanced I/M performance standard is not finalized soon enough to provide sufficient time for full SIP development, EPA will use its authority under section 110(k)(4) to conditionally approve SIP submittals committing to adopt enforceable enhanced I/M programs consistent with EPA's guidance."); cf. id. at 13,523 (announcing intent to conditionally approve transportation control SIPs). The EPA later confirmed its intent to conditionally approve the I/M, as well as other, submittals in a memorandum to regional EPA directors dated July 22, 1992 ("Shapiro Memorandum") and in an August 26, 1992 letter to the NRDC ("Rosenberg Letter"). Subsequently, in a final rule on "Inspection/Maintenance Program Requirements," 57 Fed.Reg. 52,950, 52,970-71, (Nov. 5, 1992), the EPA codified the conditional approval policy for enhanced I/M submissions. See also 40 C.F.R. Sec. 51.372(b).9 Finally, in a "Nitrogen Oxides Supplement to the General Preamble" ("NO subx Supplement"), 57 Fed.Reg. 55,620, 55,622-23 (Nov. 25, 1992), the EPA authorized conditional approval of the NO subx RACT submittals required under 42 U.S.C. Sec. 7511a(b)(2) and (f). The NRDC filed three petitions challenging the EPA's conditional approval policy as set out in the Shapiro Memorandum and the Rosenberg Letter and as specifically applied in the NO subx Supplement and the I/M Rule. Before we address the NRDC's challenge to that policy, however, we must first consider whether three of the cited decisions are final and ripe for review.

The EPA asserts that this court is without jurisdiction to review the decisions in the Shapiro Memorandum, the Rosenberg Letter, and the NO subx Supplement because they do not constitute final agency action or, alternatively, because they are not ripe for review. We disagree and hold that all three documents, insofar as they authorize conditional approval of "committal" submissions, appear to be both final and ripe.

"To determine finality, courts must decide 'whether the agency's position is definitive and whether it has a direct and immediate ... effect on the day-to-day business of the parties challenging the action.' " Her Majesty the Queen ex rel. Ontario v. EPA, 912 F.2d 1525, 1531 (D.C.Cir.1990) (quoting Ciba-Geigy Corp. v. EPA, 801 F.2d 430, 435-36 (D.C.Cir.1986)). "The inquiry seeks to distinguish a tentative agency position from the situation where 'the agency views its deliberative process as sufficiently final to demand compliance with its announced position.' " Id. (quoting Ciba-Geigy, 801 F.2d at 436). Thus, "the absence of a formal statement of the agency's position, as here, is not dispositive: An agency may not, for example, avoid judicial review 'merely by choosing the form of a letter to express its definitive position on a general question of statutory interpretation.' " Id. (quoting Ciba-Geigy, 801 F.2d at 438 n. 9). Accordingly, the fact that the EPA did not promulgate a formal rule codifying its conditional approval policy for SIP submittals other than those addressed in the I/M Rule is irrelevant to finality. Further, that the decision as set forth in the three documents constitutes final agency action is clear from the EPA's subsequent conditional approval of non-I/M committal SIPs (specifically NO subx RACT submittals) under authority of those documents. By granting such approval, the EPA has already caused the very effect that the NRDC claims is outside the agency's statutory authority--it has postponed the submission of substantive SIPs beyond the congressionally mandated deadline. Thus, the documents at issue reflect a final agency decision.

Similarly, there is little reason to question the ripeness of the EPA's decisions. "The ripeness doctrine generally prevents courts from becoming 'entangled' in 'abstract disagreements over agency policy' and from improperly interfering in the administrative decisionmaking process." Id. at 1532 (quoting Abbott Labs. v. Gardner, 387 U.S. 136, 148, 87 S. Ct. 1507, 1515, 18 L. Ed. 2d 681 (1967)). In determining ripeness, we address "both the fitness of the issues for judicial decision and the hardship to the parties of withholding court consideration." Abbott Labs., 387 U.S. at 149, 87 S.Ct. at 1515. Under the "first prong" of this test, fitness for review, courts consider "such factors as whether the issue presented is purely legal, whether consideration of the issue would benefit from a more concrete setting, and whether the agency's action is sufficiently final." Her Majesty the Queen, 912 F.2d at 1532. The question here is obviously a purely legal one (construction of section 110(k)(4)) and, as already noted, the agency's action is final and the setting, given the EPA's approval of non-I/M submissions, appears concrete. Further, under these circumstances, we need not reach the second prong of the ripeness test. "As we noted in National Recycling Coalition, Inc. v. Reilly, 884 F.2d 1431, 1434 (D.C.Cir.1989), '[w]here the first prong of the [ Abbott Laboratories ] ripeness test is met and Congress has emphatically declared a preference for immediate review ... no purpose is served by proceeding to the second prong.' " Id. at 1433. "Here, as in National Recycling, Congress has 'declared a preference' for prompt review by providing that:

[a]ny petition for review [of a final action of the Administrator] under this subsection shall be filed within sixty days from the date [of] notice of such promulgation....

42 U.S.C. Sec. 7607(b)(1)." Id. Thus, we conclude that the non-I/M decisions in the three challenged documents are ripe as well as final.

Next, we address the merits of the EPA's construction of the conditional approval provision. We review the agency's construction of the provision under the framework set forth in Chevron USA, Inc. v. NRDC, 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984). NRDC v. Reilly, 976 F.2d 36, 40-41 (D.C.Cir.1992). Given the express language of section 110(k)(4), the CAA's general SIP approval scheme and the legislative history, we conclude the EPA's construction of the conditional approval provision is contrary to Congress's unambiguous intent and must therefore be rejected. Cf. id. (denying deference to and rejecting EPA construction of CAA which "both the language and the purpose of the Act and the 1990 Amendments preclude.").

(4) Conditional approval

The Administrator may approve a plan revision based on a commitment of the State to adopt specific enforceable measures by a date certain, but not later than 1 year after the date of approval of the plan revision. Any such conditional approval shall be treated as a disapproval if the State fails to comply with such commitment.

42 U.S.C. Sec. 7410(k)(4). The EPA has construed this provision to authorize conditional approval of a committal SIP that contains no specific substantive measures so long as it is submitted by one of the statutory deadlines and includes (1) a promise to adopt specific enforceable measures within a year and (2) a schedule of "interim milestones" in this future adoption process. See, e.g., 40 C.F.R. Sec. 51.372(b)(1) ("States shall submit a SIP revision by November 15, 1992, which includes the schedule required in paragraph (a)(1) of this section10 and a formal commitment from the Governor to the adoption and implementation of an I/M program meeting all requirements of this subpart.").11 We believe the EPA's position reflects a strained construction of the statutory language. On its face, that language seems to authorize conditional approval of a substantive SIP or SIP revision which, although not approvable in its present form, can be made so by adopting specific EPA-required changes within the prescribed conditional period.

The statutory scheme confirms our interpretation of the statutory language. As already noted, the 1990 Amendments require that the EPA determine the completeness of each submission, that is, whether it contains "the information necessary to enable the Administrator to determine whether the plan submission complies with the provisions of [the CAA]," "[w]ithin 60 days of the Administrator's receipt of a plan or plan revision, but no later than 6 months after the date, if any, by which a State is required to submit the plan or revision." 42 U.S.C. Sec. 7410(k)(1). Such a determination cannot reasonably be made unless the conditionally approved submittal contains something more than a mere promise to take appropriate but unidentified measures in the future. In addition, the statute appears to require that the conditional approval decision, like any approval or disapproval determination, be made within 12 months after the completeness finding, see 42 U.S.C. Sec. 7410(k)(2), (3), (4), which, as we noted, must follow the submission. In short, the CAA contemplates the following schedule: (1) submission of a SIP, (2) followed by a finding, within two to 6 months, that the SIP is complete, and (3) approval (whether absolute, partial, or conditional) or disapproval within 12 months thereafter. To permit conditional approval at the time of initial submission, without any completeness determination or substantive review, would turn the statutory timetable on its head.

Finally, our construction of section 110(k)(4) is bolstered by the House Report on the 1990 Amendments, which explains that "[s]ection 110(k)(4) authorizes conditional approval of a SIP where the State commits to adopt such specific enforceable additional measures as EPA requires within one additional year," H.R.REP. No. 490, 101st Cong., 2d Sess., pt. 1 at 220 (1990) (emphasis added). That the contemplated specific and enforceable measures are to be "additional" presupposes that the SIP already contains some specific enforceable measures before it can be conditionally approved.

For the preceding reasons, we conclude that the conditional approval mechanism was intended to provide the EPA with an alternative to disapproving substantive, but not entirely satisfactory, SIPs submitted by the statutory deadlines and not, as the EPA has used it, a means of circumventing those deadlines. Accordingly, we hold that section 110(k)(4) does not authorize the EPA to use committal SIPs to postpone SIP deadlines. We must nevertheless decide whether the one-year extensions of the I/M and NO subx RACT submission deadlines, which the EPA's conditional approval policy effected, can be sustained on some other basis. We conclude that the extension of the enhanced I/M and the NO subx RACT deadlines can be upheld as necessary and appropriate under the particular circumstances here but that the basic I/M extension was entirely unjustified.

In the case of enhanced I/M SIPs, the EPA made its own procrustean bed. Under the 1990 CAA Amendments, Congress contemplated that states be given a one-year period after guidance promulgation to bring their SIPs into compliance with the enhanced I/M performance standard. The agency's failure to meet its November 15, 1991 deadline, however, made it impossible for states both to have the benefit of this lead time and to meet their November 15, 1992 enhanced I/M submission deadline. While the CAA is very specific about the consequences of a state's failure to meet the submittal deadline, the Act is silent on what should occur if the agency misses its guidance deadline. Under such circumstances, we must attempt to discern on our own what Congress would have intended. See Linemaster Switch Corp. v. EPA, 938 F.2d 1299, 1302-05 (D.C.Cir.1991). In undertaking this inquiry, we owe no deference to the agency's view of congressional intent. As we observed in Linemaster, "it would indeed be odd to conclude that Congress implicitly entrusted a laggard agency with the authority to devise a remedy for its own untimeliness." Id. at 1303. Instead we give considerable weight to the statute itself as Congress generally means what it says when imposing an explicit deadline. Judicial extension of its deadline is an extraordinary remedy not to be imposed as a matter of course. There are nonetheless circumstances in which an extension is warranted. Such is the case, for example, if Congress would have intended that the deadline be extended to provide a party the full statutory time for acting on a agency guidance.

In NRDC v. Thomas, 805 F.2d 410 (D.C.Cir.1986), we held that the EPA was required to postpone for one year the statutory deadline for automobile manufacturer compliance with EPA promulgated NO subx emission standards because Congress had expressly contemplated that manufacturers have four years to bring their vehicles into compliance and the EPA's tardy promulgation of the standards gave them only three years. The Thomas court explained: "Although fully cognizant of the frustration the drafters would have felt, could they have foreseen the course of events, we nonetheless find that they enacted a four year leadtime requirement and we have no alternative but to enforce it, unless or until Congress decrees otherwise." Id. at 435. We are under the same compulsion here.

Congress's carefully crafted statutory scheme provided the states a full year after the EPA guidance to develop and submit their enhanced I/M SIPs. Because the fundamental statutory requirement for those SIPs is that they "comply in all respects" with the EPA guidance, 42 U.S.C. Sec. 7511a(c)(3)(B), we can only infer that Congress believed a full year was necessary to complete complying SIPs. The inference is reinforced by the complexity of the performance standard that Congress envisioned, see id. Sec. 7511a(c)(3)(B)-(C), and that the EPA ultimately promulgated in its final I/M rule.

The EPA extended the NO subx RACT submission deadline because of a slightly different timing problem. Section 182(f)(1) of the 1990 Amendments exempts from its NO subx reduction requirements NO subx sources for which "the Administrator determines (when the Administrator approves a plan or plan revision) that net air quality benefits are greater in the absence of reductions of oxides of nitrogen from the sources concerned," "that additional reductions of oxides of nitrogen would not contribute to attainment of the national ambient air quality standard for ozone in the area," or "that additional reductions of oxides of nitrogen would not produce net ozone air quality benefits in such region." 42 U.S.C. Sec. 7511a(f)(1). The EPA determined in the NO subx Supplement, however, that "as a technical matter, photochemical grid modeling is the only reliable tool to justify an areawide exemption from the NOX requirements (or relaxation of otherwise required NOX reductions)" and that "for a variety of ozone nonattainment areas, ... photochemical grid modeling either has not been utilized previously or, if utilized, has not adequately considered the effects of NOX emissions reductions." The EPA therefore concluded that "the time needed to establish and implement a modeling protocol and to interpret the model results will, in a variety of cases, extend beyond the November 15, 1992 deadline for submission of NOX rules." 57 Fed.Reg. at 55,623. Accordingly, the EPA granted a narrow one-year extension for NO subx RACT submissions "limited to instances where the State documents that (1) credible photochemical grid modeling is not available or did not consider the effects of NOX reductions and (2) the State submits progress reports on the modeling showing the program is on schedule while the committal SIP is being reviewed by EPA." Id.

The CAA expressly gives the EPA 14-18 months after the submittal deadline to approve or disapprove the SIP and at the same time to determine whether a state qualifies for one of the section 182(f)(1) exemptions. Because only a single NO subx RACT submission is required under the statute, it is logical to infer that Congress intended data supporting exemptions to be included in that submittal and that the EPA have the full 14-18 months to review them and to make an exemption determination. If the EPA is correct about the grid modeling problem, however, that period would be considerably reduced in some cases. We believe, consistent with our conclusion on enhanced I/M submission lead time and the decision in Thomas, that had Congress foreseen the exemption timing problem, a matter outside the EPA's control, it would have elected to accord the EPA the full statutory review time.

Finally, we agree with the NRDC that the extension for basic I/M submittals was impermissible. The 1990 Amendments expressly direct that a basic I/M SIP be submitted "immediately after November 15, 1990" and require only that it comply with the provisions of the state's previous SIP or with EPA guidance "as in effect immediately before November 15, 1990." 42 U.S.C. Sec. 7511a(a)(2)(B)(i), (b)(4). Given the clear statutory submission deadline, which is in no way dependant on promulgation of additional EPA guidance,12 we see no justification for granting extensions for basic I/M submittals.

Having concluded that the EPA properly extended the statutory enhanced I/M and NO subx RACT deadlines, but not the basic I/M deadline, we must now consider what if any remedial measures to adopt. The NRDC asks that we take a strict approach and require the EPA to make immediate findings of noncompliance for all states that failed to meet the statutory deadlines, to approve or disapprove the submissions it has received within 120-180 days, and immediately thereafter to impose the statutory sanctions on those states that have not submitted fully approved SIPs. We believe the circumstances call for a less drastic remedy.

First, because the EPA, by ignoring the statutory deadline for promulgating guidance, was responsible for the tardy submission of enhanced I/M SIPs, we conclude that the EPA should be required to compensate for the delays it caused by accelerating its review of those submittals.13 It is for this reason that we have already ordered the EPA to propose either approval or disapproval of SIPs it has received no later than July 15, 1994 and to finally approve or disapprove them no later than September 15, 1994. See Amended Order, No. 92-1596 (filed April 22, 1994). On the other hand, we think it would be unfair to penalize states that reasonably relied on and complied with the EPA's extended deadlines by similarly accelerating the statutory sanction scheme. Accordingly, we direct that the sanction clock for I/M SIPs start, if necessary, from the time of SIP disapproval in accordance with the statutory scheme.

As for the NO subx RACT SIPs, we find it undesirable to shorten the SIP review time for two reasons. First, the reason for extending the NO subx RACT deadline was outside the EPA's control. As we explained above, some states simply needed more time for further grid modeling. Second, and more importantly, it is likely that the EPA will require the full statutory review period to make the necessary individual determinations whether each state granted a NO subx RACT extension qualifies for any of the section 182(f)(1) exemptions. Accordingly, we leave intact the NO subx RACT timetable, under which the EPA must approve or disapprove submittals it has received no later than May 15, 1995.

In summary, we hold that (1) the EPA's use of "committal" SIPs to extend the statutory submission deadlines is without statutory authority, (2) under the circumstances here the enhanced I/M and the NO subx RACT deadlines were properly extended to further the CAA's purposes, (3) July 15, 1994 should be treated as the deadline for EPA approval or disapproval of basic and enhanced I/M submittals, with the statutory sanction clock running from that time, and (4) the statutory period for reviewing NO subx RACT submittals should commence as of the extended submittal deadline of November 15, 1993, with the approval and sanction timetables to run from that time according to the statutory scheme.

The NRDC contends that the EPA's enhanced I/M rule impermissibly extends the November 15, 1992 statutory deadline on which enhanced I/M programs for serious, severe, or extreme ozone nonattainment areas must "take effect." See 57 Fed.Reg. at 53,003 (codified at 40 C.F.R. Sec. 51.373). Section 182(c)(3)(B) of the 1990 Amendments, 42 U.S.C. Sec. 7511a(c)(3)(B), mandates not only that states submit implementation plans for enhanced I/M programs by November 15, 1992, but also that the enhanced programs "take effect" on that date. Equating "take effect" with "be fully implemented," the NRDC asks us to invalidate the provision of EPA's final rule calling for full implementation of enhanced I/M programs by 1996.14 Because we think it clear that, contrary to the NRDC's argument, "take effect no later than 2 years from November 15, 1990" does not mean "be fully implemented" by that date, we reject the NRDC's contention and so deny the petition in this respect.

The crux of the parties' disagreement is the meaning to ascribe to the words "take effect." The NRDC argues that this question can be resolved as a matter of plain meaning under the first step of the Supreme Court's now-familiar Chevron analysis. See Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984). According to the NRDC, "a Clean Air Act program 'take[s] effect' when it cleans the air to the degree required in the underlying regulation." NRDC's Opening Brief, No. 92-1596, at 32. The EPA, for its part, notes that the statute nowhere defines "take effect," and maintains that the interpretation of "take effect" lies soundly within the realm of EPA discretion under the second step of Chevron. In the preamble to its final rule, the EPA concluded that the "take effect" requirement is satisfied when a state makes its enhanced I/M program effective as a matter of law by adopting all necessary statutory and regulatory authority. See 57 Fed.Reg. at 52,971.15 The EPA further argues that its schedule for full implementation of enhanced I/M programs, which allows states two to three years after the programs become legally operative, is in fact quite abbreviated given the substantial challenge of constructing stations and implementing new high-technology testing.

Chevron instructs that "[f]irst, always, is the question whether Congress has directly spoken to the precise question at issue." 467 U.S. at 842, 104 S.Ct. at 2781. The NRDC maintains that the words "take effect" reflect a deliberate congressional decision to embrace an early implementation date. "Take effect," the NRDC argues, means "produce action and results," that is, achieve the Clean Air Act's underlying goals. NRDC's Opening Brief, No. 92-1596, at 32. We, however, find the meaning of "take effect" to be significantly less plain than the NRDC suggests. "Effectiveness" language is frequently used by Congress to connote legal effectiveness, not results. Thus, the CAA Amendments themselves provide that "the amendments made by this Act ... shall be effective on the date of enactment of this Act." Pub.L. No. 101-549, Title VII, Sec. 711(b), 104 Stat. 2684 (1990) (emphasis added). Even the NRDC would likely concede that this language does not require implementation of all standards, construction of all machinery, and improvement of the air by November 15, 1990, the date on which the Amendments became law.16 Similarly, we noted in Public Citizen v. U.S. Trade Representative, 5 F.3d 549 (D.C.Cir.1993), cert. denied, --- U.S. ----, 114 S. Ct. 685, 126 L. Ed. 2d 652 (1994), that, "[i]f approved by Congress, NAFTA is scheduled to take effect on January 1, 1994." Id. at 550 (emphasis added). Obviously neither we nor Congress itself contemplated that the benefits inhering in a fully implemented NAFTA would accrue on that date. "Take effect" is a phrase whose meaning varies considerably with context.

Because the phrase "take effect" is itself ambiguous, its meaning must be discerned according to Chevron 's second step. Thus, "the question for the court is whether the agency's answer is based on a permissible construction of the statute." 467 U.S. at 843, 104 S.Ct. at 2782. We find the EPA's conclusion that the "take effect" language requires only legal effectiveness entirely reasonable in light of the statutory scheme. First and foremost, it strains credulity to assert that Congress expected states simultaneously to undertake the legal legwork necessary to implement a wholly new enhanced I/M program and to begin, much less complete, the task of getting such a program up and running within a year of promulgation of the EPA's guidance. The Act confers tremendous flexibility on state legislatures and regulators to decide how best to attain the performance standards promulgated by the EPA. In view of this breadth of state latitude, state authorities simply cannot guess prior to state legislative or administrative action what form the program will ultimately take or how to prepare for its instantaneous implementation. And it is also unreasonable to assume that Congress expected states to undertake legislative or administrative action within a few weeks or even months in order to devote the balance of the year to implementation. The Senate Report accompanying the 1990 Amendments states that "[e]xperience since passage of the Clean Air Amendments of 1970 has shown that nine months is not adequate time for States to prepare and submit implementation plans for new or revised air quality standards." S.REP. No. 101-228, 101st Cong., 2d Sess. 20, reprinted in 1990 U.S.C.C.A.N. 3385, 3406.

The NRDC suggests that, under the EPA's interpretation, the "take effect" provision is superfluous in light of Sec. 7410(a)(2), which already provides that SIPs must include "enforceable" emission reduction programs by the same date, November 15, 1992. We do not find this argument particularly compelling. There might well be a distinction between an "enforceable" program and one that has already taken legal effect. The word "enforceable" is susceptible of various connotations, and could reasonably be read to require only that relevant regulatory machinery, including a viable enforcement apparatus, be promulgated, rather than that the legal machinery be currently in effect or that implementation be underway. The "take effect" provision, as construed by the EPA, ensures that the state programs are legally effective pending the EPA's approval of SIPs, a process that could take over a year, and so cannot be considered superfluous. We perceive no want of logic in a construction of the statute that recognizes three distinct steps to compliance: the point at which a program is promulgated, the point at which it takes legal effect, and the point at which it is fully implemented. We note that in other sections of the Amendments, Congress itself appears to recognize this distinction between promulgating a standard--putting all the machinery in place--and making it legally effective. See 42 U.S.C. Sec. 7412(q)(1) ("any standard under this section which has been promulgated, but has not taken effect ... shall not be affected by [this Act]") (emphasis added); 42 U.S.C. Sec. 7412(q)(4) ("no standard promulgated under this section ... shall take effect for two years following November 15, 1990") (emphasis added). In sum, we find the EPA's interpretation of "take effect" reasonable and worthy of Chevron deference.

Although we reject the NRDC's argument that the "take effect" language requires revocation of the EPA's implementation schedule, we nonetheless must address the NRDC's separate charge that the EPA's selection of implementation dates is arbitrary and capricious. The NRDC contends that the EPA's implementation dates are inconsistent with the statutory scheme, because the final rule does not require that I/M programs be fully effective by the dates for meeting final and interim air pollution reduction targets. Section 7511(a)(1) of the Amendments sets primary standard attainment dates for marginal to extreme ozone nonattainment areas.17 Marginal areas must attain the primary ozone standard no later than November 15, 1993. 42 U.S.C. Sec. 7511(a)(1). Serious areas, for which enhanced I/M programs are necessary, are required to attain the primary standard no later than November 15, 1999. Id.

We perceive no inconsistency between the dates for primary standard attainment and the EPA's implementation schedule for I/M programs, and so reject the NRDC's argument.18 With respect to enhanced I/M programs, we note that the 1999 attainment date for serious areas occurs after, not before, full implementation of enhanced I/M in these regions. With respect to basic I/M programs, although the EPA's implementation date of mid-1994 might at first glance seem counterintuitive, given the requirement that marginal areas conform to the attainment standard by November 15, 1993, this inconsistency is by no means fatal to the EPA's implementation schedule. The statute provides for multiple methods of reducing emissions, and states may cut emissions elsewhere to meet the primary standard. Moreover, I/M programs are only required in marginal areas to the extent that they were required prior to the 1990 CAA Amendments. See 42 U.S.C. Sec. 7511a(a)(2)(B). Thus, a state may employ its existing apparatus toward attaining the primary standard, and the implementation schedule may have less relevance with respect to basic I/M programs in these regions. Finding no defect, we move on.

The NRDC next contends that the EPA's final rule impermissibly narrows the scope of enhanced I/M programs in the northeastern ozone transport region by exempting largely rural areas from the statute's requirements. The Amendments designate an ozone transport region, comprised of eleven northeastern states and the District of Columbia, in which special emissions rules apply regardless of ambient air quality. The statute requires that "each area in such State that is in an ozone transport region, and that is a metropolitan statistical area or part thereof with a population of 100,000 or more comply with the provisions of section 7511a(c)[ (3) ]19 ... (pertaining to enhanced [I/M] programs)." Section 7511a(c)(3), in turn, sets out requirements for enhanced I/M programs in serious areas, which are to be implemented "in each urbanized area [of a Metropolitan Statistical Area] ... with a 1980 population of 200,000 or more." Incorporating the cross-referenced section's urban limitation, the EPA's I/M rule for ozone transport regions exempts "largely rural counties." See 57 Fed.Reg. at 52,966. This, the NRDC argues, the EPA is not free to do in the face of the clear language of the statute.

The EPA counters that it is unclear whether Congress intended the ozone transport region provision to incorporate the urban limitations of the provision it references. The agency concedes that Congress sought to require enhanced I/M programs in the ozone transport region for areas with lower population than in the cross-referenced provision. Hence, the 100,000-person, rather than 200,000-person, threshold. However, the EPA argues that the very existence of a threshold signifies congressional recognition both of the difficulty of implementing the high-technology enhanced I/M programs in sparsely populated regions and of the trivial emissions reductions such implementation confers. The EPA maintains that its exemption--for counties with fewer than 200 persons per square mile--represents a reasonable interpretation of the statute.

Predictably, the parties' battle is again drawn on Chevron lines, the NRDC claiming the statute clearly resolves the issue (step one), and the EPA proclaiming the statute's ambiguity (step two). Perhaps because the statutory provision is so opaque,20 each side points to legislative history to shore up its defenses. The NRDC observes that Congress rejected a Senate measure that would have permitted the EPA to exempt areas in the transport region from I/M requirements if they did not "contribute" to the nonattainment problem. See 136 CONG.REC. S4363, S4377 (daily ed. Apr. 18, 1990). Accordingly, the NRDC maintains, Congress expressly rejected the idea that sparsely populated regions should escape enhanced I/M requirements. The EPA rejoins that when the House deleted that exemption authority, it also explicitly narrowed the scope of enhanced I/M in Sec. 7511a(c)(3), the cross-referenced provision, to urbanized areas of 200,000 or more, rejecting a Senate approach that would have expressly required the provision to apply to all vehicles in the covered area. Accordingly, the EPA argues, although Congress eliminated broad exemption authority in the transport region provisions, it simultaneously exempted all rural areas from enhanced I/M in the transport region through Sec. 7511a(c)(3). We believe that the legislative history battle ends in a draw, leaving intact the question whether Congress intended to incorporate the urban limitation into its ozone transport region provision.

Because it is not clear from the statute or its legislative history whether Congress sought to incorporate the urban limitation by reference, this question is best resolved under Chevron step two. In such cloudy circumstances, "it is enough that the Agency's construction is reasonable." Ohio v. EPA, 997 F.2d 1520, 1527 (D.C.Cir.1993). Moreover, "EPA need not establish that the statute compels its regulation." Id. (emphasis in original). Pursuant to this lenient standard, we find it clear enough that the agency's interpretation merits our deference. As the EPA observes, in serious, severe, and extreme nonattainment regions, Congress determined that enhanced I/M programs would be limited to urbanized areas. Despite the fact that these regions are the most polluted in the nation, Congress made a policy choice not to require enhanced I/M programs in their sparsely populated areas. It is neither illogical nor unreasonable for the EPA to conclude, in light of the urban limitation for these heavily polluted regions, that the high-technology enhanced I/M programs are similarly unwarranted in rural areas within the ozone transport region.

The NRDC also challenges the EPA's requirement that basic I/M programs extend only to "urbanized" portions of moderate and higher nonattainment areas. See 57 Fed.Reg. at 52,987 (codified at 40 C.F.R. Sec. 51.350(a)(4)). The CAA Amendments provide that "[e]ach State in which all or part of a Moderate Area is located" must submit a state implementation plan including a basic I/M program "[f]or all Moderate Areas." 42 U.S.C. Sec. 7511a(b)(4). Since the plain language of this provision makes no exception for rural regions within moderate areas, the NRDC contends that the EPA is powerless to create one. The NRDC notes, in this regard, that Congress expressly created a rural exception for enhanced I/M programs in the statute. See 42 U.S.C. Sec. 7511a(c)(3)(A). When Congress wanted to limit I/M programs, the NRDC argues, it did so explicitly. Moreover, the NRDC contends that the provision requiring state implementation plans for all moderate areas "without regard to whether or not the area was required by [the previous Act] to have included a specific schedule for implementation," 42 U.S.C. Sec. 7511a(b)(4), expressly disavows the agency's prior practice of exempting rural regions within nonattainment areas from I/M requirements. Although the NRDC's arguments are not insubstantial, we find ourselves ultimately unpersuaded.

In contrast to the NRDC, we find the operative statutory language equivocal and the legislative history unavailing. The statutory requirement that states submit implementation plans including basic I/M programs "[f]or all Moderate Areas" does not, in itself, mandate basic I/M throughout every square mile of moderate regions; it simply directs that each moderate area be covered by an implementation plan. Similarly, the requirement that states submit implementation plans for those moderate areas not covered in the previous statute does not by its terms affect the scope of I/M programs within those areas. Section 7511a(b)(4), which covers moderate areas, requires a basic I/M program "as described in subsection (a)(2)(B) of this section." Section 7511a(a)(2)(B), in turn, requires a basic I/M program "of no less stringency than ... the program defined in [the prior Act] as interpreted in the guidance of the Administrator." The prior EPA guidance required basic I/M only in urbanized areas within nonattainment areas. See 57 Fed.Reg. at 52,965-66. The question therefore pivots on the scope of the word "stringency." The NRDC contends that "stringency" refers to the substantive rigor of the program, not to its geographic scope. The EPA, in predictable contrast, maintains that "stringency" refers not only to testing standards, but also to the number of vehicles tested. We find the statutory language insufficiently precise to command either interpretation beyond doubt.

Admittedly, the NRDC makes a plausible point when it urges that Congress' creation of an express rural exclusion for enhanced I/M programs makes it less likely that Congress implicitly intended to maintain or desired to construct a rural exclusion with respect to basic I/M programs. See Hercules Inc. v. EPA, 938 F.2d 276, 280-81 (D.C.Cir.1991) (pertaining to CERCLA). Congress knew how to limit programs, the argument runs, and when it did not do so, such programs are perforce unlimited. However, in the context of this particular statute, in which Congress repeatedly cross-references internal sections, refers to provisions of the previous Act, and alludes to EPA guidances, we are not confident that the conventional inference trumps in all cases. The express creation of a rural exclusion for enhanced programs, which were newly-created by the 1990 Amendments, does not, in our view, compel the conclusion that Congress sought silently to alter any preexisting exclusions for basic I/M programs, particularly when Congress explicitly incorporated the preexisting guidance by reference.21

Resolution of this particular dispute, therefore, again takes us to Chevron 's second step, under which we defer to reasonable agency constructions of the "stringency" provision. Pursuant to this standard, we have no trouble deeming the EPA's interpretation a permissible view of the statute. Obviously, Congress can be presumed to have had knowledge of the contents of the prior EPA guidance whose standards it explicitly incorporated. See Lorillard v. Pons, 434 U.S. 575, 581, 98 S. Ct. 866, 870, 55 L. Ed. 2d 40 (1978); Dart v. United States, 848 F.2d 217, 229 (D.C.Cir.1988); National Soft Drink Ass'n v. Block, 721 F.2d 1348 (D.C.Cir.1983). Because the prior guidance excluded rural areas from basic I/M requirements, and because the "stringency" of a program may easily encompass its breadth, we have no cause to doubt the plausibility of the EPA's formulation, which we now uphold.

The NRDC next contends that the EPA employed unduly lenient testing methods in order to dilute the enhanced I/M performance standard it was required to promulgate under 42 U.S.C. Sec. 7511a(c)(3)(B)(i), and that this weak performance standard, in turn, was used to justify a shift away from the statutory presumption of annual testing to a presumption of cheaper, biennial testing.22 To water down the performance standard, according to the NRDC, the EPA exempted older automobiles from "high-tech" emissions tests, visual inspections to detect tampering, and evaporative system checks.23 See 57 Fed.Reg. at 52,956-57. The NRDC contends that the EPA conceded the sizeable benefits of subjecting older-model vehicles to these tests when it recommended that such testing comprise part of a biennial program as a feature that "exceed[s] the requirements" of the statute. Because the EPA's annual performance standard achieves smaller reductions than are technically feasible, the NRDC assails it as a violation of the statute. The NRDC also maintains that the EPA failed adequately to explain its decision not to promulgate a more stringent standard.

The EPA responds that it set a performance standard that provided states with "continued reasonable flexibility to fashion effective, reasonable, and fair programs," as required by Sec. 7511a(a)(2)(B)(ii) of the Amendments. The EPA argues that the statute cannot be read to require that it set the most austere performance standard possible, for otherwise there would be no way in which states could develop biennial programs with features exceeding the minimum requirements for annual programs or reductions that exceed those obtainable through annual inspections. See 42 U.S.C. Sec. 7511a(c)(3)(C)(v). The EPA maintains that the statute allowed it discretion in setting performance standards. As to the particular testing decisions, the EPA argues that it reasonably concluded that the marginal cost of high-technology IM-240 testing for pre-1986 models far exceeded the marginal benefits in reduction. Moreover, although the statute requires that the EPA set "a performance standard achievable by a program combining emission testing, including on-road emission testing, with inspection to detect tampering with emission control devices and misfueling for all light-duty vehicles," 42 U.S.C. Sec. 7511a(c)(3)(B)(i) (emphasis added), the EPA argues that the statute does not require that all vehicles be subjected both to emission testing and visual inspection. Accordingly, the EPA argues, it was free to exempt older models from visual inspection to detect tampering.

Each of the parties wins some and loses some on this issue. First of all, we think it clear that the statute does not mandate that the EPA set the most stringent possible annual performance standard. With its repeated emphasis on state flexibility, echoed in the legislative history, see S.REP. No. 101-228, 101st Cong., 2d Sess. 39, reprinted in 1990 U.S.C.C.A.N. 3425, the statute appears to place a premium on state ability to shuffle aspects of the program to meet the EPA's requirements and individual state needs. Moreover, the Act does seem to contemplate that the addition of extra features might enable biennial programs to "equal or exceed the reductions which can be obtained through ... annual programs." 42 U.S.C. Sec. 7511a(c)(3)(C)(v) (emphasis added). Implicitly, at least, Congress thus appears to have contemplated considerable EPA discretion in standard-setting.

However, Congress clearly did not intend that such discretion be boundless. In Sec. 7511a(c)(3)(B)(i), after instructing the EPA Administrator to promulgate a guidance setting forth the enhanced I/M performance standard, Congress clarifies that the standard must be "achievable by a program combining emission testing, including on-road emission testing, with inspection to detect tampering with emission control devices and misfueling for all light-duty vehicles and all light-duty trucks." The EPA contends that this provision simply specifies the ingredients of the program; it does not mandate that all the tests listed be performed on all covered vehicles. To support this proposition, the EPA cites to a statement in the House Report that "[o]n-road emission [testing] may not be practical ... for every vehicle, and is not required. However, it should play some role in the State program." H.R.REP. No. 490, Pt. 1, 101st Cong., 2d Sess at 239 (1990). The NRDC responds that, while on-road testing may not apply to every vehicle, inspection to detect tampering does apply to all vehicles under the clear terms of the statute.

We agree with the NRDC that the statutory provision provides a conjunctive requirement: the standard must be achievable by a program of emissions testing and visual inspection. Moreover, we find that, although the statute vests the EPA with considerable discretion--to choose, for example, what form of emissions testing to require and what level of volatile organic compounds and other particulate reductions ultimately to set--the statute does clearly require that the standard be the product of two different kinds of testing. In effect, the provision circumscribes the EPA's standard-setting discretion by specifying the two criteria on which it must base its decision. To the extent that the EPA develops target reduction rates by ignoring one or the other, its reduction forecasts are overly conservative and out of keeping with clear statutory requirements.24

Applied to the EPA's actual testing decisions, our assessment of the statute requires that we affirm in part, and remand in part. With respect to the EPA's decision not to subject certain older vehicles to high-technology IM-240 testing or evaporative system checks, we see no reason to undo the EPA's standard. The EPA indisputably has applied some form of emissions testing to all vehicles, in accordance with the plain terms of 42 U.S.C. Sec. 7511a(c)(3)(B)(i). The statute affords the EPA discretion in determining the type of emissions testing to apply, and under the second step of Chevron, we find no cause to doubt the reasonableness of the EPA's conclusion. The EPA considered evidence that idle tests suffice for older carbureted, noncomputerized models, but are insufficient to test emissions for newer cars, which come equipped with sensors and computers that continually adjust emissions. Moreover, the EPA demonstrated that the costs of IM-240 testing for older vehicles far exceed the costs of idle testing. See 57 Fed.Reg. at 52,951. Finally, we note that the EPA's detailed defense of this decision in the administrative record is sufficient to withstand the NRDC's inadequate explanation charge.

We are constrained, however, to grant the NRDC's petition with respect to the EPA's failure to subject older vehicles to visual inspections to detect tampering. As noted above, although the precise form of testing may be discretionary, Congress clearly instructed that the performance standard represent the product of some form of (1) emissions testing and (2) visual inspection to detect tampering for all vehicles. The Senate recognized the prominent success rate of visual inspections in detecting fuel switching and tampering, noting that in areas without I/M programs, fourteen percent of all cars have been subject to fuel switching, but that the figure drops to ten percent with an I/M program and to six percent with an I/M program combined with anti-tampering checks. See S.REP. No. 101-228, 101st Cong., 2d Sess. 39, reprinted in 1990 U.S.C.C.A.N. 3425. Because the EPA was without discretion to exempt pre-1984 models from visual inspection, we remand so that the EPA may recalibrate its standard after including the relevant tests.

III. NATIONAL AUTOMOBILE DEALERS ASSOCIATION'S CHALLENGES TO

Petitioners National Automobile Dealers Association, et al., (collectively, "NADA") present two main challenges to the EPA's final rule on state inspection and maintenance programs. First, NADA claims that the EPA exceeded its authority under section 182 of the Clean Air Act by issuing its model I/M program in the form of a final rule pursuant to the Administrative Procedure Act's ("APA's") notice and comment procedures, rather than an informal "guidance" as the agency has done in the past. Second, NADA advances a number of arguments against the final rule aimed at exposing what it views as an unlawful and arbitrary bias against decentralized "test-and-repair" networks.

We uphold the EPA's final rule against both of NADA's challenges. To the extent Congress in the 1990 Clean Air Act Amendments required certain components of the EPA's model I/M program to be binding upon the states, the agency was authorized, if not required, to engage in APA notice and comment rulemaking. Because its alleged injury stems from the mandatory requirements of the 1990 Amendments, rather than the procedures the EPA used to effect them, we find that NADA lacks standing to challenge the agency's use of rulemaking procedures to promulgate a performance standard for "enhanced" I/M programs. In addition, we find that the EPA properly issued the "basic" I/M components of the final rule (which Congress did not require to be binding upon the states) under the Administrator's general authority under section 301 of the Act "to issue regulations as are necessary to carry out his duties under the Act." Finally, we conclude that the EPA properly exercised its authority to determine the manner in which decentralized testing facilities would be deemed equivalent to centralized facilities.

As we discussed in reference to the NRDC's challenges to the final rule, see Part I, supra, the 1990 Amendments to the CAA require states to implement vehicle inspection and maintenance programs in designated regions within their borders that prior to 1990 had failed to meet the EPA's national ambient air quality standards. Depending upon the severity of the pollution in these so-called "nonattainment areas," states must adopt either a "basic" I/M program or a more stringent "enhanced" program.

For both types of program, Congress directed the EPA to "review, revise, update, and republish in the Federal Register the guidance for the states for motor vehicle inspection and maintenance programs required by [the Act], taking into account the Administrator's investigations and audits of such program[s]." 42 U.S.C. Sec. 7511a(a)(2)(B)(ii) (emphasis added). Congress required that the EPA's "guidance" be incorporated into state implementation plans, and that it provide the states "reasonable flexibility to fashion effective, reasonable, and fair programs for the affected consumer." Id.

With respect to enhanced programs, the 1990 Amendments direct the states to "comply in all respects" with the EPA's guidance. 42 U.S.C. Sec. 7511a(c)(3)(B). They also mandate that state enhanced I/M programs contain certain minimum elements, including the requirement that "operation of the program [be] on a centralized basis, unless the State demonstrates to the satisfaction of the Administrator that a decentralized program will be equally effective." 42 U.S.C. Sec. 7511a(c)(3)(C)(vi). In addition, Congress required the EPA to include in its guidance a performance standard to be used in vehicle emissions testing, compliance with which "shall be determined using a method to be established by the Administrator." 42 U.S.C. Sec. 7511a(c)(3)(B).

For basic I/M programs, Congress directed the EPA to update its guidance to cover, at a minimum, certain features regarding the type and extent of the program. 42 U.S.C. Sec. 7511a(a). Unlike the requirements for enhanced I/M programs, however, Congress did not command that state basic programs "comply in all respects" with the EPA guidance. Nor did Congress require the EPA to promulgate a specific performance standard or to mandate the use of a centralized testing network.

In response to the 1990 Amendments, the EPA issued, after notice and comment, a final rule governing both enhanced and basic I/M programs. 57 Fed.Reg. 52,950. The final rule sets forth a model program incorporating the specific features mandated by Congress for enhanced programs and a performance standard that all basic and enhanced programs must meet. The enhanced program performance standard requires percentage reductions in three types of vehicle emission pollutants as measured in part by the "IM-240" exhaust test, a newly-developed test which the EPA found most accurately reflects vehicle emissions under actual driving conditions. The EPA asserted that the IM-240 test is more sophisticated than previous tests and has proven particularly effective in detecting pollutants emitted from newer technology vehicles. 57 Fed.Reg. at 52,953-54.

The EPA's enhanced model program also provides that vehicle testing must be performed at centralized testing facilities, a feature Congress required unless other methods were demonstrated to be equally effective. A state may adopt a decentralized network, but only if it can show that it would be as effective as a centralized network. In evaluating that showing, the EPA considers decentralized "test-only" networks (in which testing stations do not make repairs) presumptively equivalent to centralized networks. In contrast, a state choosing to adopt decentralized "test-and-repair" networks (in which testing facilities also perform repair work) must demonstrate equivalency. If it does not, the EPA reduces by 50 percent the emission credits earned by the state towards meeting the performance standard. In other words, absent evidence to the contrary, the EPA considers decentralized "test-and-repair" networks 50 percent less effective than centralized networks (and decentralized "test-only" networks) in reducing emission pollutants. The agency based this estimate largely on covert audits of existing decentralized networks which revealed "egregious levels of improper testing" at test-and-repair facilities. 57 Fed.Reg. at 52,959.

The EPA's basic model I/M program includes a performance standard that is applied in the same manner as that for enhanced programs: A state need not adopt the model program if it can demonstrate that its program is equally effective at meeting the performance standard. However, the actual performance standard is considerably less stringent than the enhanced program standard, and measurement of the standard does not require use of the IM-240 test. Centralized testing is also not required. According to the agency, the basic program performance standard can be met by using "a reasonably comprehensive, conventional [decentralized] test-and-repair system." 57 Fed.Reg. at 52,959.

B. Issuance of I/M Program "Guidance" Through APA Notice and Comment Rulemaking

As discussed above, section 182 of the 1990 Amendments directs the EPA to "review, revise, update, and republish in the Federal Register ... guidance" in the design of state I/M programs. 42 U.S.C. Sec. 7511a(a)(2)(B)(ii). NADA contends that Congress did not intend for such guidance to be issued in the form of a final rule promulgated pursuant to the APA's notice and comment procedures. It rests this position on four arguments: (1) that "guidance" is a term of art referring to "informational or interpretive policy adopted apart from notice and comment rulemaking lacking a fully binding legal effect," NADA's Opening Brief at 11; (2) that past EPA practice has been to issue informal, non-binding guidelines; (3) that Congress intentionally used the word "guidance" in reference to I/M programs while explicitly authorizing the agency to issue "rules and regulations" in other parts of section 182; and (4) that informal, non-binding guidance is consistent with the CAA's goal of allowing the states flexibility to design their own I/M programs.

For its part, the EPA concedes that it has in the past issued informal, non-binding guidance documents on state I/M programs, but it rejects the notion that section 182 prevents it from issuing guidance through notice and comment rulemaking. Rather, the EPA argues, language in section 182 and elsewhere in the Act authorizes, if not compels, binding rules. The agency locates the source of its authority in two places. First, with respect to enhanced I/M programs, the EPA notes that section 182(c)(3)(B) directs the states to "comply in all respects" with the EPA's guidance for enhanced I/M programs. 42 U.S.C. Sec. 7511a(c)(3)(B). The agency argues that this command, and not the agency's choice of procedures to effect it, creates a binding obligation on the states to adopt enhanced I/M programs that comport with the EPA's model program. Accordingly, the agency contends that NADA suffered no injury from the rulemaking and therefore lacks standing to challenge the EPA's choice of procedures as it relates to enhanced I/M programs. Second, with respect to basic I/M programs, the EPA points to section 301 of the Act, which grants the Administrator general authority "to prescribe such regulations as are necessary to carry out his functions under this chapter." CAA Sec. 301(a)(1); 42 U.S.C. Sec. 7601(a)(1). In the agency's view, binding basic I/M regulations were necessary "to insure that states will implement effective and cost-effective programs." 57 Fed.Reg. at 52,953.

We agree with the agency that Congress unambiguously intended the performance standard and other features of the enhanced model I/M program to be binding upon the states. See 42 U.S.C. Sec. 7511a(c)(3)(B) (directing states to "comply in all respects" with the EPA's guidance for enhanced I/M programs and mandating that the guidance include a performance standard and, inter alia, centralized testing). As a result, any rules setting forth the mandatory parameters of the states' obligations under the Act are legislative in character, and the EPA probably was required to promulgate such rules only through APA rulemaking procedures. See PPG Industries Inc. v. Costle, 659 F.2d 1239 (D.C.Cir.1981) (EPA violated the APA by mandating continuous sulfur dioxide compliance monitoring through informal rulemaking without notice and comment); Citizens to Save Spencer County v. EPA, 600 F.2d 844, 877-79 (D.C.Cir.1979) (EPA rules at issue were "legislative rather than interpretive and could only have been promulgated in accordance with the notice and comment procedures of Sec. 4 of the APA").

But NADA is not in a position to argue the point in any event. Congress' imposition of a binding performance standard upon the states makes it clear that NADA lacks standing to challenge the EPA's issuance of enhanced I/M program guidance in the form of a final rule. As far as we can tell, NADA fails to allege any injury resulting from the agency's pursuit of rulemaking apart from ones relating to the binding effect of the enhanced program guidance. See, e.g., NADA's Opening Brief at 30-31 (arguing that the rule allows the EPA to enforce provisions of the model program and impose sanctions against non-attaining states). It does not argue, for instance, that the substance of the EPA's enhanced model program would have been any different had it been issued as an informal guidance rather than a rule. NADA therefore is hard pressed to claim that it was injured by the rulemaking because, as discussed above, it is the statute, rather than the form of the guidance, which requires states to "comply in all respects" with the EPA's guidance and authorizes the Administrator to impose sanctions on nonattaining states.

It is most strange for a full participant in a rulemaking to challenge the extensive notice and comment procedures afforded it by the APA. Had the EPA imposed a legally binding performance standard through informal guidance, as NADA suggests it should have, NADA might well be in court arguing against the agency's decision not to provide notice and opportunity to comment. NADA's true concern, we think, is not so much the EPA's choice of procedures, but rather the binding nature of the performance standard and the EPA's authority to enforce it. Its claim is therefore a back-door challenge to the 1990 Amendments themselves, and that door cannot turn on the type of procedures used by the agency to implement Congress' statutory directive. Accordingly, we hold that NADA lacks standing to challenge the EPA's use of rulemaking procedures to issue enhanced I/M program guidance.

Because similarly explicit congressional authorization of binding rules is absent with respect to basic I/M programs, we must decide separately whether the EPA properly issued the basic I/M components of the final rule through notice and comment rulemaking, thereby giving them binding force. As this inquiry involves an agency's interpretation of a statute it administers, the Chevron analysis discussed earlier again applies. See Part II-A, supra. Proceeding first under Chevron step one, we conclude that Congress did not unambiguously preclude the EPA from issuing basic I/M program guidance through notice and comment rulemaking. First, section 182 nowhere states that the EPA may not issue binding rules governing state I/M programs. Second, Congress' use of the term "guidance" is not inconsistent with a rule, like this one, that incorporates a binding performance standard but leaves the states considerable flexibility in fashioning programs that meet the standard. It is the performance standard, not the design of the EPA's model program, that is binding on the states. The model program itself remains essentially a "guidance," as that term is most commonly understood. Third, the EPA's deviation from its pre-1990 practice of issuing informal, nonbinding guidance documents is fully understandable in light of the ongoing failure of the states to attain the EPA's ambient air quality standards--a failure that gave rise to the 1990 Amendments. It appears, moreover, that the EPA's prior practice of issuing informal guidance to the states on the content of their SIPs is not as uniform as NADA alleges. In Citizens to Save Spencer County v. EPA, 600 F.2d 844, 877-79 (D.C.Cir.1979), for instance, this court upheld a binding EPA rule, issued after notice and comment, providing guidance to the states on how to incorporate into their SIPs certain regulations prescribed by the 1977 amendments to the Act.

Absent a clear statement by Congress in section 182 preventing binding basic I/M rules, we agree with the agency that section 301 of the Act, which authorizes the Administrator "to prescribe such regulations as are necessary to carry out his functions," 42 U.S.C. Sec. 7601, is an adequate source of authority for promulgation of the basic I/M components of the final rule. Although section 301 does not provide the Administrator "carte blanche authority to promulgate any rules, on any matter relating to the Clean Air Act, in any manner that the Administrator wishes," Spencer County, 600 F.2d at 873, it is sufficiently broad to allow the promulgation of rules that are necessary and reasonable to effect the purposes of the Act. Where, as here, Congress has erected no clear impediment to the issuance of binding rules, section 301 takes the agency as far as the second step of Chevron. Once there, the EPA provided a reasoned explanation for resorting to rulemaking--namely, that the agency's "experience over the last 15 years has shown that the lack of federal minimum requirements has led to less than effective I/M programs." 57 Fed.Reg. at 52,953 (citing EPA Inspector General and GAO reports). In light of the states' longstanding failure to meet ambient air quality standards, the agency reasonably concluded that a binding basic I/M performance standard was necessary "to insure that states will implement effective and cost-effective programs." Id.

NADA advances several related arguments in an attempt to establish that the EPA's enhanced model I/M program reflects an unlawful and arbitrary bias against decentralized "test-and-repair" networks. It first questions the EPA's authority to impose the 50 percent emission credit penalty on test-and-repair systems that do not establish equivalency with centralized systems. Failing that, it claims that the EPA failed to justify its methodology for arriving at the 50 percent effectiveness estimate. Next, it argues that the EPA unlawfully failed to establish criteria under which "enhanced" test-and-repair systems--decentralized I/M facilities that use newly-developed testing technology--could be considered equivalent to centralized systems. Finally, NADA contends that the EPA arbitrarily included the IM-240 test in its enhanced model program over other, allegedly less costly and more effective tests.

Each of these arguments is unavailing. In requiring states to adopt centralized testing networks unless decentralized networks were shown to be equally effective, the EPA acted with both Congress' blessing and good reason. The 1990 Amendments require all enhanced I/M programs to contain the following design element:

Operation of the program on a centralized basis, unless the State demonstrates to the satisfaction of the Administrator that a decentralized program will be equally effective. An electronically connected testing system, a licensing system, or other measures (or any combination thereof) may be considered, in accordance with criteria established by the Administrator, as equally effective for such purposes.

42 U.S.C. Sec. 7511a(c)(3)(C)(vi) (emphasis added). The highlighted language makes clear that Congress vested the EPA with broad discretion to establish appropriate criteria for demonstrations of equivalency. In this case, the agency determined that decentralized "test-only" systems were presumptively equivalent to centralized networks, but that states adopting "test-and-repair" systems had to make individualized showings of equivalency. If they cannot, the EPA imposes a 50 percent emission credit penalty towards meeting the air quality standard.

NADA can point to nothing in the statute that prevents the EPA from imposing the penalty if equivalency is not shown. As the EPA has authority to require a demonstration of equivalency between centralized and decentralized networks, it follows naturally that the agency must be allowed to establish criteria for making the comparison and a means by which to account for the relative ineffectiveness of the decentralized system (the rate of exchange, in a sense). The 50 percent emissions credit penalty is simply the agency's expression of the relative effectiveness of the two network types.

Furthermore, ample evidence in the record supports the EPA's imposition of the 50 percent penalty. The final rule cites several studies resulting from both overt and covert audits of testing facilities that demonstrate considerable levels of improper testing and emission control tampering in states with decentralized test-and-repair networks. In Missouri, for instance, an audit of test-and-repair facilities revealed that "84% of the stations falsely passed vehicles that were set to fail the test." 57 Fed.Reg. at 52,974. Studies conducted in the two most populous states--California and New York--tell similar stories. Id. The fundamental problems with these networks appear to be the incompetence of unlicensed and ill-trained testers and the inherent incentive on the part of testers simply to pass cars along for a quick fee (or the customer's satisfaction) instead of engaging in time-consuming repairs. The audits also revealed that testers frequently pass the vehicles of familiar customers, and that customers sometimes "shop" test-and-repair stations in search of a free pass. While NADA may be correct that other reports failed to demonstrate such rampant deficiencies, the agency surely did not create the 50 percent estimate out of whole cloth. Under Chevron step two, it did not act unreasonably.

Seizing on the second sentence of the statutory language quoted above, NADA makes a somewhat better case that the EPA failed to establish criteria under which certain types of decentralized systems--electronically connected systems, licensing systems, or others--may be considered equivalent to centralized networks. We find, however, that the agency has effectively established such criteria by concluding that any type of decentralized system is presumptively equivalent so long as the test and repair functions are separated. The agency reasonably based this conclusion on audits of existing test-and-repair networks, and it was under no statutory obligation to hypothesize as to the effectiveness of what NADA calls "enhanced decentralized networks"--essentially test-and-repair networks utilizing experimental testing procedures. NADA simply mischaracterizes the final rule in suggesting that the EPA refused to treat any decentralized networks as equally effective; in fact, the EPA deems decentralized test-only networks equally effective to centralized systems. Moreover, NADA largely neglects that a state may adopt test-and-repair networks (presumably of the "enhanced" variety) and nonetheless avoid the 50 percent penalty by making an individualized demonstration of program equivalency. See 40 C.F.R. Sec. 51.353(b). These features of the final rule give teeth to Congress' clear preference for centralized testing, and at the same time provide the states ample flexibility to adopt equally-effective decentralized networks. Accordingly, the final rule withstands this aspect of NADA's challenge.

Finally, we conclude that the EPA presented sufficient support for the inclusion of the IM-240 emissions test in its model program. It responded adequately to comments from the test-and-repair industry regarding the relative merits of NADA's suggested alternative, the BAR90 test in "steady state loaded mode," see 57 Fed.Reg. at 52,955-52,956, and it furthered the Act's goal of state flexibility by not requiring states to adopt the IM-240 test. Rather, states are free to choose any equally effective alternative test to attain the emission reductions required by the performance standard.

As set out above, we grant the petitions for review insofar as they challenge the propriety of the EPA's committal SIP policy and the exemption of pre-1984 vehicles from visual inspection. The petitions are in all other respects denied.

So Ordered.