General Motors Corp. v. Environmental Protection Agency

No. 88-4257 (5th Cir. April 13, 1989)

The court holds that, under Clean Air Act § 113(d), the Environmental Protection Agency (EPA) has no authority to disapprove a state-issued delayed compliance order (DCO) that has expired by its own terms. In January 1987 Texas issued the DCO to allow General Motors to exceed air pollution limits until August 1987; EPA disapproved the DCO in February 1988 after proposing to disapprove it in July 1987. The court first holds that EPA may not enforce against violations that happened after 90 days had passed from when the DCO had issued and before EPA has acted on the DCO, because Clean Air Act § 113(d) requires EPA to either approve or disapprove state-issued DCOs for major stationary sources within 90 days. The court next holds that EPA lacks statutory authority to disapprove a state-issued DCO once the DCO has expired. If EPA disapproves a state-issued DCO, it must immediately begin enforcement action or issue a DCO of its own, and it cannot reasonably do either once the DCO has expired. Finally, the court holds that it has no jurisdiction over the case. Because EPA's disapproval of the expired DCO was unauthorized, EPA has taken no final action on the DCO. Clean Air Act § 307(b) limits judicial review to final actions.

Counsel for Petitioner
Robert T. Stewart, Michael Weinberg
Jones, Day, Reavis & Pogue
301 Congress Ave., Ste. 1200, Austin TX 78701
(512) 477-3939

Counsel for Respondent
David W. Zugschwerdt
Land and Natural Resources Division
U.S. Department of Justice, Washington DC 20530
(202) 633-2686

Before GEE, HIGGINBOTHAM, and DUHE, Circuit Judges.

[19 ELR 20841]

PATRICK E. HIGGINBOTHAM, Circuit Judge:

Texas issued a delayed compliance order temporarily authorizing General Motors to exceed relevant air pollution limits imposed by Texas state standards promulgated pursuant to the federal Clean Air Act, 42 U.S.C. §§ 7401 et seq. Several months after the delay order had expired by its own terms, the federal Environmental Protection Agency disapproved the order. General Motors appeals the disapproval. We find that under 42 U.S.C. § 7413(d)(2), the EPA has no authority to disapprove an expired delayed compliance order. We therefore hold that the EPA's delinquent action is ineffective to vitiate the force of the state order as a defense for General Motors in any enforcement proceeding, whether brought by the EPA or by a private citizen. We conclude that the order appealed from is not a final, reviewable action within the meaning of 42 U.S.C. § 7607(b), and dismiss this petition for want of jurisdiction.

The details of the dispute between Texas, the EPA, and General Motors are highly technical and quite intricate. We summarize only the gist of that controversy here. General Motors operates an automobile plant in Arlington, Texas. Included in the plant is a painting procedure which releases significant amounts of volatile organic compounds. The Clean Air Act regulates the permissible levels of such organic emissions. The particular standards applicable to the Arlington plant are those specified by the Texas state plan: under the Clean Air Act, each state must develop a State Implementation Plan — called a "SIP" by those fluent in administrative acronyms — subject to approval by the EPA. The Texas standard governing emission of volatile organic compounds became more stringent on the first day of 1987.

Anticipating that it would be unable to comply with the more stringent emission regulations, General Motors in late 1986 sought a delayed compliance order, or "DCO," from the Texas Air Control Board ("TACB"). A delay order permits a non-complying source to continue operating for a short period of time, during which the source must implement changes bringing its emissions inside the standards specified by the state plan. See 42 U.S.C. § 7413(d) (§ 113(d) of the Clean Air Act). Texas issued the delay order on January 16, 1987. The order was to expire by its own terms on August 28, 1987.

The Clean Air Act provides that "no such order issued by the State" with respect to a major stationary source "shall take effect until the Administrator determines that such order has been issued in accordance with the requirements of this chapter." 42 U.S.C. § 7413(d)(2). The Act also provides that the EPA "Administrator shall determine, not later than 90 days after receipt of notice of the issuance of [a delay order] whether or not [the order] is in accordance with the requirements of this chapter." The EPA received notice of the Texas delay order concerning the General Motors plant on January 26, 1987. The 90-day period lapsed in late April 1987. EPA had taken no action. Later, on July 31, 1987, the EPA proposed disapproval of the delay order. The delay order expired by its own terms less than one month afterward, on August 28. Finally, in February 1988, more than five months after the delay order expired, the EPA disapproved the order.

The parties' persisting interest in the disapproval of an already expired order results from the order's possible impact upon an as yet unfiled EPA enforcement proceeding against General Motors. If the order were valid despite the EPA's disapproval of it, General Motors might assert the order as a defense against prosecutions for violations allegedly occurring during the lifespan of the order: a period no longer than the interval between January 16, 1987, to August 28, 1987.

In fact, the order's impact upon an enforcement proceeding would affect an even smaller time span. For various reasons, General Motors actually operated its Arlington plant for only a few weeks during the period covered by the order. The plant was working between March 2 and March 16, and again between March 31 and May 7. On all other days covered by the order, the plant was shut down.

The EPA issued a notice of violation to General Motors on May 12, after the end of the statutory approval period for the delay order and after the plant had ceased operating. The plant did not resume operations until September 1, 1987, after the delay order had expired. The disputed disapproval ruling nonetheless remains relevant to the conjectured enforcement proceeding, since the EPA may elect to prosecute General Motors for violations occurring prior to issuance of the notice: courts have held that, once a notice of violation issues, a subsequent, post-notice violation by the source will subject the source to prosecution for earlier, pre-notice violations. See 42 U.S.C. § 7413(b); see also United States v. Louisiana-Pacific Corp., 682 F. Supp. 1141, 1163 (D. Colo. 1988).

General Motors took a timely appeal to this court from the EPA's order disapproving [19 ELR 20842] the Texas delay order. The appeal presents three distinct sets of issues. The first set of issues deals with the sequence of orders in this case. We are called upon to decide whether the EPA, by missing the statutorily imposed deadline and waiting even beyond the delay order's expiration date, forfeits its power to disapprove the delay order, or to prosecute General Motors for operations authorized by the delay order.

The second set of issues concerns the procedural integrity of the EPA's disapproval order. General Motors contends that the EPA impermissibly commingled its rule-making and enforcement proceedings, and that the EPA is attempting to resolve enforcement issues on the basis of a record too barren for such purposes.

The third set of issues has to do with the merits of the EPA's disapproval order, and so with the substance of the Clean Air Act and the Texas state plan. The EPA disapproved the order on the ground that it failed to provide for eventual compliance, failed to specify adequate interim compliance deadlines, and failed to provide for appropriate record-keeping by the source.

We first consider whether we have jurisdiction to hear this petition. Although neither party questions our jurisdiction, the court is free to raise jurisdictional issues sua sponte. See, e.g., Wright, Law of Federal Courts 23 (4th ed. 1983). In today's case, the oddity of an EPA order purporting to disapprove a long-expired state order prompts us to question whether the controversy between these parties is justiciable.

Both General Motors and the EPA contend that we have jurisdiction under 42 U.S.C. § 7607(b)(1). That provision authorizes aggrieved parties to appeal directly to the federal courts of appeals from "the Administrator's action in approving or promulgating any . . . order . . . under section 7413(d) of this title," or from "any other final action of the Administrator under this chapter (including any denial or disapproval of the Administrator under subchapter I of this chapter)." Subchapter I includes § 7413(d). It would therefore appear that the EPA order disapproving the Texas delay order is appealable to this court under 42 U.S.C. § 7607(b)(1) so long as that disapproval is an "action . . . under" § 7413(d).

General Motors and the EPA apparently assume that the EPA order appealed from was indeed an action under § 7413(d), since that section clearly authorizes the EPA to disapprove state-issued delayed compliance orders. It is, however, not obvious that § 7413(d) authorizes the EPA to disapprove such orders after expiration of the statutorily-imposed ninety-day deadline, or after the state order has expired. If the section authorizes no such action, there is no evident basis for assuming that the order appealed from is a "final action of the Administrator" under the chapter. We must therefore explore the consequences of the EPA's procrastination.

We have elsewhere held that when Congress subjects the EPA to explicit statutory deadlines in the Clean Air Act, the EPA may not ignore those deadlines with impunity. American Cyanamid Co. v. U.S. EPA, 810 F.2d 493, 499-501 (5th Cir. 1987). In American Cyanamid, we reviewed an effort by the EPA to impose penalties against a Cyanamid plant in Louisiana. The Cyanamid plant had been operating in accordance with a revised state implementation plan proposed by Louisiana, but the EPA had not acted upon the proposed revision for several years. The EPA had in fact neither approved nor disapproved the proposed revision at the time Cyanamid was decided. The EPA's failure to act was in disregard of an express requirement in 42 U.S.C. § 7410 that the EPA approve or reject a revised state plan within four months after receiving notice of the revisions.

We held that "the EPA may not collect a penalty for the period between (1) four months after a state submits a proposed revision and (2) the date the EPA rejects the revision." 810 F.2d at 500. We grounded this holding upon a number of considerations. Foremost among these was the cooperative role between the state and federal government envisioned by the Clean Air Act. We said that the issue in Cyanamid was "not so much pollution by American Cyanamid but the default of the EPA in carrying out the congressional intent to work in close cooperation with the states in implementing standards and enforcing the Clean Air Act. We emphasize the importance of the EPA acting, not for the benefit of the American Cyanamid, but for the benefit of the State of Louisiana which is by statute designated to play a significant cooperating role with the EPA." 810 F.2d at 500. We also observed that permitting the EPA to escape statutory deadlines would expose "American Cyanamid and similarly situated individuals and companies to the unacceptable risk that, contrary to state policy decisions authorized under the Act, they will be fined because of delays within the EPA." 810 F.2d at 501.

In addition to concern for state policy-making authority and for the planning capacity of regulated parties, the Cyanamid court noted a third defect resulting from the EPA's unauthorized delay. If the EPA were permitted to penalize sources for adhering to state policies in the absence of statutorily required EPA action, the EPA would have an extra incentive to disapprove the pending revision. The EPA could collect penalties for past noncompliance if it disapproved the revision, but not if the revision were accepted. 810 F.2d at 499.

III

The Cyanamid appeal, unlike this one, arose in the context of an enforcement proceeding. 810 F.2d at 494. The EPA urges us to disregard the delinquency issues in the present appeal, contending that those issues would become relevant only in the event that the EPA initiated an enforcement action. See Bethlehem Steel v. U.S. Environmental Protection Agency, 638 F.2d 994, 1002 (7th Cir. 1980) (pretermitting delinquency issues on the ground that those issues might better be resolved in an enforcement proceeding). We reject this argument for two reasons. First, as already explained, we are concerned about the possibility of a jurisdictional defect, a problem not considered by the Seventh Circuit. Second, if the Cyanamid doctrine would bar an enforcement proceeding against General Motors for the period covered by the delay order, then this suit may well be moot even if the EPA had statutory authority to disapprove the delay order. This possibility again implicates a jurisdictional defect which we cannot ignore.

Nonetheless, Cyanamid cannot resolve directly the dispute now before us. Today's case involves a deadline in 42 U.S.C. § 7413(d) and a conjectured enforcement proceeding under 42 U.S.C. § 7413(b), while Cyanamid involved a deadline in 42 U.S.C. § 7410 and an enforcement proceeding under 42 U.S.C. § 7420. However, the § 7413(d) deadline applicable to delayed compliance orders is no less important than the § 7410 deadline governing revision to state plans. As the Seventh Circuit has recognized, when the EPA does not act upon a delay order until after the statutorily specified deadline and, indeed, after the delay order expires, "the Agency clearly violate[s] the letter and intent of § 113(d)(2)." Bethlehem Steel, 638 F.2d at 1001. The statute does not permit the Administrator a range of options, but instead clearly requires "enforceable and therefore final action" within the specified timeframe. Id. (Emphasis in original). Any other construction not only lacks "support in the statutory language, [but] also manifests poor policy considerations." Id.

When construing the Clean Air Act, we deal with a detailed legislative scheme, and we must take our rule from the express language of the statute. Our decision today in fact turns upon careful scrutiny of § 7413(d)(2). To make sense of that language, however, we first review briefly the place of delayed compliance orders within [19 ELR 20843] the broader context of the Act. The provisions authorizing issuance of delay orders were added to the Act by the Clean Air Amendments of 1977, P.L. 95-95, and had at least three purposes: to accomodate sources incapable of compliance; to provide regulators with another tool for inducing troublesome sources to reduce emissions; and to preserve the authority of the states as the primary regulators of air quality under the Clean Air Act.

The provisions of the statute clearly reflect all three purposes. Under 42 U.S.C. § 7413(d)(1), a source is eligible for a delay order only if it is "unable to comply" with some requirement of "an applicable implementation plan." The state, or, after thirty days notice to the state, the EPA may grant such an order. The order must include a schedule and timetable for compliance, require compliance with applicable interim requirements, and provide for final compliance with the requirements of the applicable implementation plan. 42 U.S.C. § 7413(d)(1)(B), (C), (D). In the case of an order pertaining to a major stationary source, the order does not take effect until the EPA approves it, but the EPA is required to act on the order within 90 days after receiving notice.

By permitting states to push back compliance deadlines when a source agrees to make scheduled improvements in its pollution-control procedures, the 1977 amendments both recognized the difficulties imposed on some sources by inflexible compliance dates, and offered regulators a new lever by which to extract practically effective compliance strategies from particular sources. The division of authority between the state and federal government made clear that Congress expected that any request for a delay order would go first to the state regulator, but that the EPA would preserve an essential supervisory role.

The legislative history reflects the same set of purposes. The House report recognizes that EPA had adopted a practice of issuing orders authorizing continued operation of noncompliant sources. This practice had an uncertain statutory basis, and the Ford Administration sought Congressional enactment of more specific authority for such exemptions. See H.R.Rep. No. 294, 95th Cong., 1st Sess. 53-58, reprinted in 1977 U.S.Code Cong. & Admin.News, 1077, 1131-35. Citing the decision in Train v. Natural Resources Def. Council, 421 U.S. 60, 95 S. Ct. 1470, 43 L. Ed. 2d 731 (1975), the House observed that immediately enforceable deadlines, when coupled to discretionary exceptions for noncomplying sources that adhered to practically effective improvement schedules, were a particularly useful device for achieving air quality improvements. Such an approach is "salutary because it tend[s] to encourage sources (who would otherwise remain in violation of immediately effective requirements) to negotiate definitive compliance schedules with the States." H.R.Rep. No. 294, reprinted in 1977 U.S.Code Cong. & Admin.News at 1132. The House committee developed an approach which it said "reaffirms the proper role of the States as the primary enforcing authority of the State implementation plans." Id., 1977 U.S.Code Cong. & Admin.News at 1135.

The Senate counterpart to the House bill contained similar provisions, and the joint committee adopted, for the most part, the Senate language with respect to delayed compliance orders. As has already been seen, however, the resulting provisions continue to reflect the purposes set forth explicitly by the House report, manifested by the EPA's then-extant practice, and endorsed by the Ford Administration proposals.

Considered together, these purposes make prompt action by the EPA essential to effectuation of the statutory scheme. First, the delayed compliance order provisions become important only when a particular source is unable to comply with existing standards and deadlines. If the state issues a delay order and the EPA fails to act upon it, the source is left with a choice between ceasing operations and continuing without any protection against an enforcement action.

Second, the opportunity for a negotiated compliance schedule is lost. The EPA's inaction blocks negotiations without itself constituting a negotiating position for the federal government: absent an EPA determination of the delay order's validity, the source cannot know whether the EPA finds the proposed compliance schedule acceptable, or what might be done to meet the EPA's objections. Indeed, in this respect EPA inaction is more damaging to the non-complying source than is an express EPA disapproval of the proposed order. If the EPA were to disapprove the order quickly, the source might take a timely appeal from the EPA's decision, or might receive the benefit of a federally issued delay order. See 42 U.S.C. § 7413(d)(2).

Third, the state, the primary regulator of air quality under the Clean Air Act, is similarly without guidance or options. By issuing the delay order, the state declared its intent to accommodate the source to secure implementation of an agreed upon compliance schedule. The EPA's inaction leaves the state without means to achieve these policy goals.

Congress recognized that prompt EPA action was essential to effectuation of the policies underlying 42 U.S.C. § 7413(d). For that reason, Congress expressly required that the "Administrator shall determine, not later than 90 days after receipt of the notice of the issuance of an order under this subsection with respect to any major stationary source, whether or not any State order under this subsection is in accordance with the requirements of this chapter." 42 U.S.C. § 7413(d)(2). With respect to delay orders governing other sources, Congress placed an even heavier burden upon the EPA: the delay order is immediately effective, and remains so until the EPA affirmatively disapproves it.

Several factors suggest that prompt and timely action by the EPA is even more important with respect to delay orders than it is with respect to proposed revisions to state plans. First, the issuance of a delay order presumes that some source — a major stationary source if EPA approval is required — is incapable of complying with existing state standards. It follows that the source will face a choice between shutting down or violating the state standards if the EPA does not act on the delay order. Although a proposed revision to a state plan might be designed to accommodate a non-complying source, there is no statutory requirement that the revision benefit a source which is not complying — which, indeed, is incapable of complying — with the unrevised state plan. The revision might simply reflect a new, general strategy for reducing air pollution. The bubble concept at issue in Cyanamid is a good example: the concept may benefit some sources otherwise disadvantaged by the state plan, but, at least in the view of some commentators, the bubble concept is also a superior long-term strategy for inducing reductions to overall emissions. See, e.g., Chevron U.S.A., Inc. v. Natural Resources Defense, 467 U.S. 837, 104 S. Ct. 2778, 2792 n. 37, 81 L. Ed. 2d 694 (1984).

Second, a delay order is by its nature temporary. If the EPA dawdles not only beyond the ninety-day deadline but, as happened here, past the expiration date of the order, the EPA's review of the order necessarily takes on an abstract, almost counter-factual character. For example, one ground for the EPA's rejection of the Texas order issued in this case was that the order did not "demonstrate" that the General Motors factory would be in compliance as of the order's expiration date. But we do not understand the EPA to argue that the operations specified in the Texas order were necessarily incompatible with compliance as of the order's expiration date. The record before us affords no basis for concluding that General Motors was not in compliance as of the termination of the delay order, and thereafter. It would thus be possible, depending upon one's interpretation of the Texas state plan, to find that [19 ELR 20844] General Motors was in compliance with the state plan from August 28, 1987, onward; that General Motors was in compliance as a direct result of the changes implemented pursuant to the delay order; but that the delay order was defective because it failed to provide adequate assurances that General Motors would be in compliance. In other words, the EPA might disapprove the delay order, after its expiration, for the reason that the delay order might not have caused what it did in fact cause.

Another ground for the EPA's disapproval of the Texas delay order was that the order did not provide for appropriate recording of emissions during the order's lifetime. See TACB Order No. 87-01, Attachment A. Had the EPA determined in a timely fashion that such a deficiency existed, the EPA might have remedied the problem by issuing its own, modified delayed compliance order. See 42 U.S.C. § 7413(d)(2) (if the Administrator determines that a state-issued delay order is unacceptable, the EPA may issue its own delay order governing the source). Again, the record does not reveal whether it would have been difficult or easy for General Motors to substitute the EPA's recording system for the one specified by the order. But it is obviously impossible for General Motors to alter its recording methods for the period covered by the order when the EPA does not decide upon its objection to those methods until after the order has expired.

Third, the counterfactual character of the EPA's review has a secondary consequence: it skews the EPA's regulatory incentives. If the EPA rejects a delay order promptly, the EPA must assume certain costs. The negotiated compliance schedule is lost. The plant may have to shut down, and the EPA may have to take political responsibility for the closure. The EPA may have to invest resources in an enforcement proceeding. After the order expires, however, disapproval entails none of these costs. Instead, the EPA can only benefit: if the EPA chooses to prosecute the source for noncompliance, the prosecution will be more lucrative, since the EPA may collect penalties for the period when the source might have claimed protection from the delay order. Indeed, the prosecution may appear especially attractive, since the source, to secure the delay order, will have itself proven that it not only did not comply, but could not have complied, during the period of the delay order.

Fourth, the EPA's delinquency may effectively force the noncomplying source to choose among potential air quality violations. If the source conforms only to the delay order, and the EPA eventually disapproves the delay order, the source might be penalized for noncompliance with the underlying state plan. If, however, the source does nothing until the delay order is actually approved, the source might be penalized for noncompliance with the schedule of the delay order. The EPA has in fact encouraged the states to prosecute sources for noncompliance with delay orders during the period when EPA approval for the order is pending. See Memorandum from Marvin B. Durning, Assistant Administrator for Enforcement, to Regional Administrators, Regional Enforcement Directors (July 27, 1978) (hereafter "Durning Memorandum") at p. 10. The source might comply with both the underlying state plan and the delay order by shutting down while simultaneously making the improvements called for by the delay order. That option is, however, likely to be extremely expensive.

At least some of these policy considerations may also apply to the EPA's failure to observe statutory deadlines for approval of state plan revisions, since some revisions will be very similar to delay orders: a revision can be a short-term extension of existing time-limits which impose special burdens on a particular source. According to the EPA, an example of this sort of revision is TACB Order No. 87-04, which extends compliance deadlines for General Motors' Arlington plant on the basis of the "alternate methods of control" provisions in the Texas state plan. But see United States v. General Motors, 702 F. Supp. 133, 137 (N.D.Tex.1988) (rejecting, with respect to another TACB order, EPA's contention that an "alternative methods of control" order is in fact a SIP revision requiring EPA approval).

The array of statutory purposes underlying the delayed compliance order provisions justifies a limitation on the EPA's enforcement powers at least comparable to the restriction imposed by the Cyanamid decision. At a minimum, the EPA must forfeit its power to collect penalties for the period between April 27, when the deadline for EPA action passed, and August 28, when the delay order expired of its own terms. There remain only the questions whether the EPA also loses its power to penalize General Motors for noncompliance with the state plan during the period between the EPA's receipt of the order and the EPA's missed deadline for action upon the order, and whether the EPA in fact has any authority at all to disapprove a state-issued delay order after the order's expiration.

We have thus far mentioned pieces of 42 U.S.C. § 7413(d)(2), but the issues confronting us require that we now consider as a whole the various restrictions which that subsection imposes upon the EPA:

In the case of any major stationary source, no such order issued by the State shall take effect until the Administrator determines that such order has been issued in accordance with the requirements of this chapter. The Administrator shall determine, not later than 90 days after receipt of notice of the issuance of an order under this subsection with respect to any major stationary source, whether or not any State order under this subsection is in accordance with the requirements of this chapter. In the case of any source other than a major stationary source, such order issued by the State shall cease to be effective upon a determination by the Administrator that it was not issued in accordance with the requirements of this chapter. If the Administrator so objects, he shall simultaneously proceed to issue an enforcement order in accordance with subsection (a) of this section or an order under this subsection.

42 U.S.C. § 7413(d)(2).

We have already stressed that a ninety-day deadline applies in cases involving major stationary sources. That deadline is juxtaposed with a presumption, applicable in other cases, making a delay order immediately effective up until EPA disapproval. There is no comparable provision either making a delay order effective up until EPA disapproval once the ninety-day deadline has passed, or deeming EPA inaction after the deadline tantamount to approval. From the absence of any such provision we infer that the delay order does not "take effect" simply because the EPA misses the ninety-day deadline, and that the EPA retains power to approve or disapprove the order after passage of the deadline. The effect of the ninety-day deadline, considered by itself, is only to bar the imposition of penalties against the source for noncompliance with the underlying state plan during the period between the deadline and the EPA's eventual determination of the order's validity.

We also note that some of the hardships imposed by the EPA's delinquency will arise, though for a shorter period, even if the EPA acts in a timely fashion. During the ninety-day period when the EPA may lawfully consider the merits of a state-issued delay order, a major nonstationary source may be faced with dual compliance requirements: the state might require the source to comply with the delay order's schedule, but the source has no guarantee that it will be protected from prosecution pursuant to the underlying state plan.

The subsection imposes a second constraint on EPA approval or disapproval, however. If the Administrator objects to a delay order, "he shall simultaneously proceed [19 ELR 20845] to issue an enforcement order in accordance with subsection (a) of this section or an order under this subsection." This constraint forces the Administrator to issue either a notice of violation, see 42 U.S.C. § 7413(a)(1), or a federal delayed compliance order, see 42 U.S.C. § 7413(d)(1). The position of this second constraint within § 7413(d)(2) — immediately after the sentence dealing with sources other than major stationary sources — would permit an argument that the constraint applied only to disapproval of delay orders not dealing with major stationary sources. But there is no sound reason for so restricting the requirement, and the syntax of the subsection equally well allows distribution of the requirement over both sets of delay orders. We hold that the constraint does apply to the disapproval of delay orders dealing with major stationary sources.

Such a construction is entirely consistent with the policy purposes underlying the delayed compliance order provisions. In effect, the constraint forces the EPA to fish or cut bait. The EPA must accept the state's proposed enforcement mechanism — the compliance schedule set out in the delay order — or issue its own delay order, or begin enforcement proceedings. The last option is not an escape hatch. Section 7413(a) requires only a notice to the violator. But § 7413(b)(1) gives the Administrator no discretion if a major stationary source fails to reach compliance after receiving notice of a violation: the Administrator must commence a civil action against the source. In the case of a source which received a state-issued delayed compliance order, an EPA rejection of the state order and refusal to issue a federal modification of the order will necessarily result in shutdown or a civil action: the source received the delayed compliance order only because it was unable to comply with the state plan.

It is apparently impossible for the EPA to comply with this constraint once the period of the delay order has expired. The EPA cannot reasonably issue an alternative delay order to govern the period of the expired order, since the time has passed and the source could not conform its past actions to new restrictions. Nor can the EPA reasonably issue a simultaneous notice of violation under § 7413(a)(1). Were the EPA to disapprove the delay order during the term of the order, the EPA could assume that the source was in violation of applicable regulations, since the delay order is predicated on that assumption. There is no justification for such an inference after the order has expired, however.

It follows that the EPA cannot comply with the second constraint in § 7413(d)(2) once the period of the delay order has expired. We therefore hold that the EPA lacks any statutory power to disapprove state-issued delayed compliance orders after the expiration of the effective period of those orders. The second constraint effectively imposes a second, easily ascertained deadline upon the EPA. See Sierra Club v. Thomas, 828 F.2d 783, 790 (D.C.Cir. 1987) (a duty of timeliness may be inferred from the Clean Air Act even when the deadline in question is not expressly set out).

The second constraint in § 7413(d)(2) does not, however, impose any limits upon EPA approval of delayed compliance orders, although the EPA may not be able to take any action upon an expired order which does not meet the standards specified in § 7413(d)(1). In any event, the EPA clearly has not approved this order. The EPA purports to have disapproved the order, but we rule that the EPA had no authority to do so because the order had expired. We therefore conclude that the EPA has not yet taken "any final action under" the Clean Air Act with respect to the delay order. Section 7607(b)'s grant of jurisdiction over "any denial or disapproval by the Administrator under Subchapter I of" the Clean Air Act provides us with no warrant to hear this petition.

Nor do we discern any other sense in which the EPA's supposed disapproval is a "final action" within the meaning of § 7607(b). We are cognizant that this phrase has been broadly construed, and is not limited by the example of the other statutory sections mentioned in § 7607(b). Harrison v. PPG Industries, Inc., 446 U.S. 578, 100 S. Ct. 1889, 1896, 64 L. Ed. 2d 525 (1980). One Justice has opined that by § 7607(b) Congress has undertaken "a massive expansion in the number of Agency actions directly reviewable by the courts of appeals," and has done so without any indication that "it had given thought to the consequences." 100 S. Ct. at 1899 (Blackmun, J., concurring). Justice Stevens, writing in dissent, expressed concern that the Harrison Court's majority had construed § 7607(b) in a way that would "distort the concept of final agency action by giving EPA virtually unlimited discretion to transform its informal advise into final agency action subject to court of appeals' review." 100 S. Ct. at 1904 (Stevens, J., dissenting).

This court has registered similar concerns, even when we have found that an order appealed from was a "final action" of the Administrator for purposes of the Clean Air Act. See City of Seabrook v. U.S. Environmental Protection Agency, 659 F.2d 1349, 1371 (5th Cir. 1981); see also State of Texas v. Environmental Protection Agency, 499 F.2d 289, 321 (5th Cir. 1974) (Clark, J., concurring) (commenting upon the "adverse effects flowing from the legislative mandate that judicial review proceedings be injected into the court system at the appellate level").

Nonetheless, despite this extraordinarily broad grant of appellate jurisdiction, we see no ground for considering the EPA order in question here a final agency action within the meaning of § 7607(b)(1). The EPA order does not even amount to "advice." It is entirely retrospective. It does not tell, or suggest, anything that General Motors would be well-counseled to do. Indeed, the order does not even "advise" General Motors that the EPA intends to pursue an enforcement proceeding: at oral argument, the EPA informed us that it was not sure whether any such action would be undertaken. The order purports to open up certain past actions to prosecution, but we have today held that the EPA is powerless to do that.

The purpose of § 7607(b) is "to provide prompt pre-enforcement review of EPA action." Harrison, 100 S. Ct. at 1897. That purpose would in no way be served by review of the present order. The EPA admits that no enforcement action is presently pending or planned. A ruling on the EPA's order would not tell General Motors how to avoid enforcement proceedings by altering its present behavior. Even if § 7607(b) has made the EPA's advice reviewable, we decline to find that the EPA's mere opinions, even when retrospective rather than advisory, are likewise reviewable. We are therefore without jurisdiction.

Although the jurisdictional defect in this case becomes evident only after thorough examination of the facts and the statutory scheme, our holding fits well with common sense, the language of the statute, and policies endorsed by Congress. We think it desirable to comment further upon the relation between today's holding and the more general dispute between the EPA and General Motors. We do so, however, to explain the limits upon our jurisdictional finding, rather than to resolve the substantive dispute that separates the parties. We are without jurisdiction to hear those substantive arguments today.

Common sense suggests that something is amiss when an agency purports to disapprove an expired order after the fact. We sit, however, as a court of law, and it is therefore crucial that the language of the text explicitly requires what common sense suggests. The statute imposes two constraints upon EPA evaluation of delayed compliance orders. The first constraint is a specific deadline. The second constraint is a responsibility the EPA must carry out if it abrogates the state regulator's chosen enforcement program. The EPA in today's case missed the deadline imposed by the [19 ELR 20846] first constraint and waited so long that it could not assume the responsibility imposed by the second constraint. By so doing, the EPA forfeited its opportunity to disapprove the order.

The two constraints not only are explicit in the Act but also make sense in light of the policies which led Congress to approve the delayed compliance order provisions. As we have already noted, delay orders are time sensitive responses to situations in which a particular source is unable to comply with a state plan's deadline. Such situations call for hard choices, choices which the state regulator must make in the first instance. The delay orders expand the array of enforcement options available to the state. They provide a useful opportunity for effective compromise; indeed, they are so useful that the EPA itself had developed a practical equivalent to the orders before Congress provided any clear authorization for such action. Part of the Congressional purpose in codifying the practice was to return primary authority to the state regulator. Had the statute not included limits upon the EPA's ability to hold orders without acting upon them, the EPA, by the sort of procrastination it displayed in this case, could frustrate Congress' clear intent.

Our ruling does not in any way provide noncomplying sources with a windfall that permits them to escape regulation via administrative maneuvering. On the contrary, the source can have no delayed compliance order without invoking the state regulator's enforcement authority. See American Cyanamid, 810 F.2d at 499 (source does not benefit "undeservedly" where the state regulator undertook to authorize its operations). A source wishing to obtain a delay order must, moreover, first tell the state and the EPA that the source is not, and cannot be, in compliance with applicable state requirements. Finally, the EPA retains authority to disapprove the order up until the time the order expires.

Indeed, it is unclear whether our jurisdictional ruling provides General Motors with any practical advantages in its dispute with the EPA. The Cyanamid rule, apart from our more special finding today, would appear to bar enforcement of the underlying state plan against General Motors from the period beginning April 27, 1987, and running until the expiration of the delay order. The only period left in issue, then, is the period between issuance of the order by Texas and the EPA's missed ninety-day deadline. The Act, however, only recognizes a delay order as a bar to enforcement of an underlying state plan during "the period the order is in effect." 42 U.S.C. § 7413(d)(10). See also 40 C.F.R. § 65.03(a)(2). A delay order governing a major stationary source does not "take effect" until approved by the EPA. 42 U.S.C. § 7413(d)(2). It is not apparent to us whether a delay order that is lawfully and timely approved by the EPA is "in effect" as of its approval date, or retroactively as of its issuance date. The difference seems important. The EPA has taken the position that a delay order approved by the EPA will not insulate a party from penalties for noncompliance with the state plan prior to the effective period of the delay order. See Durning Memorandum at p. 9 (July 27, 1978) ("EPA will generally pursue an action for penalties covering the pre-delayed compliance order period where the State has exercised its discretion differently than EPA would have by issuing a delayed compliance order to a source which EPA . . . has determined to be an appropriate candidate for judicial action"). It is clear that the EPA could have waited lawfully until April 27, 1987, and then approved Texas' delayed compliance order. If that order would not have barred the EPA from levying penalties against General Motors for noncompliance prior to April 27, we do not see why the absence of approval or disapproval — or, indeed, a delinquent approval in 1989 — would do so. If the delayed compliance order is indeed irrelevant to the pre-deadline period, then even our jurisdictional ruling today is effectively moot with regard to the dispute between the parties. The EPA's briefs to us, however, seem to suppose that the delayed compliance order has some relevance to enforcement actions governing the period prior to April 27. The issue has not been briefed, we do not have jurisdiction to decide it, and we do not decide it. As already noted, we mention these points only to clarify the very limited scope of our jurisdictional ruling.

In sum, judicial review of the EPA's order disapproving the expired state order seems likely to produce an essentially advisory opinion, even by the lenient standards of the Clean Air Act. We have at once too much and too little before us. Too much, because the parties press us to decide broad issues about federal-state relations under the Clean Air Act, and about the substantive meaning of the Texas state plan, in order to resolve the status of a long-expired order whose remaining practical significance is diaphanous at best. Too little, because the essence of the dispute between these two parties is elsewhere. The focal point of that dispute is obviously an as yet unfiled, potential enforcement proceeding. Both sides tell us that, should we reach the merits of this dispute, our ruling will likely decide issues crucial to such an enforcement proceeding. The EPA also tells us that our ruling in this suit may affect its action upon TACB Order No. 87-04, issued in March 1987 and altering deadlines applicable to the General Motors plant in 1987. There is at least one other suit between the parties. United States v. General Motors, 702 F. Supp. 133.

We cannot today resolve these matters. Finding that the EPA has issued no final, reviewable order within the scope of its authority, we determine that we are without subject-matter jurisdiction to hear General Motors's petition. The petition is therefore

DISMISSED.