A Tale of Two Theories: The Legal Basis for EPA's Proposed Revision to the Routine Maintenance, Repair, and Replacement Exception, and the Implications for Administrative Law

Matthew C. Stephenson

The author is a Clerk to the Honorable Stephen Williams, Senior Judge for the U.S. Court of Appeals for the District of Columbia Circuit, and a John M. Olin Fellow in Law, Economics, and Business at Harvard Law School. He may be contacted at mstephenson@post.harvard.edu.

[33 ELR 10789]

How many lawyers, regulators, engineers, and contractors does it take to change a light bulb? Lots, if you happen to be changing the light bulb at a "stationary source" of pollution, and the bulb change counts as a "modification" of the source under the Clean Air Act (CAA).¹ According to that statute, any physical or operational change that results in an increase in the source's emission of certain pollutants triggers an exacting and costly set of permitting requirements. Since emissions can increase if a source operates faster or more frequently than it did before, numerous physical or operational changes could potentially trigger an "emissions increase" within the meaning of the statute. Perhaps not changing a light bulb, but replacing old turbine blades, upgrading a computer system, fixing leaky pipes, repairing a broken gauge, or any of a host of other activities might increase total emissions (by causing a source to operate at a higher capacity), and, if these activities count as physical or operational changes, they will all have to undergo the rigors of the CAA's aggressive permitting requirements.

Something seems intuitively wrong with that, even from a committed environmentalist's perspective. Some activities, though technically physical or operational "changes," appear to be so much a part of the routine operation of power plants, incinerators, factories, or other facilities that it seems illogical to subject them to permitting requirements designed to regulate more significant construction or modification programs. And so, the U.S. Environmental Protection Agency (EPA) has quite sensibly chosen to exempt certain types of activities, known as "routine maintenance, repair, and replacement" (RMRR) activities, from the stringent permitting requirements that the CAA would otherwise impose. So, you can change your light bulb or fix your leaky pipe without having to worry, but if you decide to spend a million dollars completely redesigning and renovating your main generator, you'll need to go get your permit.

That's all well and good, and fairly uncontroversial. But the hard questions that arise with respect to RMRR—as is often the case with these sorts of legal and regulatory provisions—concern the scope, form, and legal justification for the exemption.

First, with respect to scope, which activities should be considered RMRR, and which should not? The leaky pipe and the million-dollar renovation are easy cases; what about the harder intermediate cases, like the turbine replacement or computer upgrade? Should the RMRR category be broad or narrow?

Second, with respect to form, should EPA try to apply the RMRR exemption through something that looks more like a bright-line rule or something that looks more like an open-ended standard? Should EPA and state enforcement authorities try to make these determinations on a case-by-case basis? Or, should EPA give effect to its substantive policy judgments through broad, general, rule-like categories—rules that, while perhaps overinclusive or under-inclusive, provide clarity and certainty?

Third, with respect to legal justification, on what grounds does EPA have the authority to create and delimit the RMRR exception? Are there legal constraints on EPA's choices regarding scope or form? What must EPA show in order to establish that its choices in these matters are supported by congressional authority?

Currently, the scope of the RMRR exception is relatively narrow, its form is that of an open-ended, multi-factor standard, and its legal justification is not entirely clear, though courts have accepted (without much discussion) EPA's authority to promulgate the exception as it currently exists. But on December 31, 2002, EPA proposed a substantial change in the RMRR rule. The new proposal would broaden the scope of the RMRR exception and would put it in the form of a rule rather than a standard. The legal basis for this change, however, is unclear. This is in part because the courts have never articulated clearly the legal basis for EPA's authority to create the RMRR exception, as distinct from the Agency's authority to limit the application of that exception. Given that EPA's new rule, if and when it is published in final form, is almost certain to be challenged in court, understanding the legal basis for the RMRR exception has become more important.

The goal of this Article is threefold: to explain more fully the role of the RMRR exception in the CAA's regulatory scheme; to describe the distinct and potentially competing legal bases for the current rule and the proposed rule change; and to argue for what I view as the most theoretically and practically satisfying view of the proper legal basis for the RMRR exception. The Article addresses each of these objectives in turn.

Part I lays out the statutory and regulatory framework of the CAA and its various permitting requirements, describes the history and significance of the RMRR exemption to those requirements, and explains how the newly proposed rule would change the current regime.

Part II scrutinizes the legal basis for the RMRR exemption, demonstrating that there are two distinct potential legal theories for this exemption. Both theories, however, are potentially [33 ELR 10790] problematic as justifications for the proposed rule change. On the one hand, EPA could argue that the CAA is sufficiently ambiguous that the Agency can exempt certain activities as RMRR as an exercise of its authority, under the deferential standard of Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc.,² to interpret the statutes it administers when the U.S. Congress has not spoken clearly. On the other hand, EPA might justify the categorical definition of certain activities as RMRR by relying on its recognized—but limited—authority to create so-called de minimis³ exceptions to the requirements of the CAA even when the text of the statute is not ambiguous. In the official preamble to the proposed rule, EPA tried to paper over the distinction between these two approaches, but prior drafts of the preamble indicate that the issue of the proper legal basis for the rule change was the subject of considerable discussion within EPA and during interagency review. As the preamble was drafted and re-drafted, EPA moved away from a de minimis theory and toward a Chevron theory, though the published preamble remains somewhat ambiguous on this point.

The different theories turn out to have quite different implications for the scope and limits of EPA's authority under the CAA, and for broader administrative law issues as well. Thus, Part III considers the question of how EPA, and the court that reviews the challenge to the final version of the rule, ought to choose between these theories. I argue that the de minimis theory is the more satisfying legal justification, despite EPA's apparent reluctance to defend its rule change on that basis. From EPA's self-interested perspective, relying on a Chevron theory is risky because of the apparently clear language of the statute, whereas the safeguards built into the proposed rule put EPA in the position to make a good case that the new RMRR proposal is justified even under a more constrained de minimis approach. From the point of view of the court, and in the interests of administrative law development more generally, upholding the RMRR rule on a Chevron basis threatens to gut the "congressional intent" prong of Chevron. Relying on the de minimis theory, in contrast, strikes the appropriate balance between congressional, judicial, and agency power in determining the scope of this and other exemptions.

Moreover, invoking the de minimis exemption in this case would also enable the reviewing court to resolve the currently open question of how much deference is due an agency's specification of a de minimis exemption. I argue that the appropriate standard of judicial review for this particular kind of agency determination is not the extremely deferential Chevron Step Two test but rather the more intermediate, standard-like approach of Skidmore v. Swift & Co.⁴ A court applying the Skidmore standard to the proposed RMRR rule change would probably uphold the rule as a reasonable exercise of EPA's power to create de minimis exceptions, but it would give challengers a fair chance to show that the scope and form of the proposed rule sweep in too many activities that cannot be justified on a de minimis theory.

I. The Statutory and Regulatory Framework

A. The CAA's Permitting Requirements: New Source Performance Standards (NSPS), New Source Review (NSR), and Prevention of Significant Deterioration (PSD)

Under the CAA Amendments of 1970,⁵ the construction or modification of a "stationary source" of certain "criteria pollutants" must meet NSPS promulgated by EPA. The NSPS require sources to install the best available control technology (BACT) at new or "modified" sources. In 1977, Congress-frustrated at the failure to meet the CAA's initial clean air goals by the original statutory deadlines—enacted new amendments to the CAA, several of which imposed additional requirements for new or modified stationary sources. Most importantly, the 1977 Amendments established the NSR program,⁶ under which new or modified major stationary sources in "nonattainment areas"—that is, geographic regions where concentrations of criteria pollutants exceed the national ambient air quality standards (NAAQS)—are subjected to additional preconstruction permitting requirements more stringent than those already imposed by the NSPS.⁷ Most significantly, NSR requires projects to employ technologies that guarantee the lowest achievable emissions rate (LAER)⁸ —a requirement even stricter than BACT⁹ —and to offset any new emissions from the new or modified source by equivalent reductions in emissions from some other source.¹⁰

The 1977 Amendments also established a PSD program.¹¹ Unlike the NSR program, which applies only in nonattainment areas, PSD applies only in attainment areas—the goal of the program is to prevent especially clean areas from becoming any dirtier.¹² The PSD program imposes [33 ELR 10791] three requirements on new or modified¹³ sources in attainment areas. First, these sources must employ BACT.¹⁴ While this is the same standard that would otherwise be imposed by NSPS, the PSD program requires that the BACT determination be made on a case-by-case basis, rather than through an industrywide determination.¹⁵ Second, additional technological and other requirements apply if the source is located in certain designated areas, such as those near natural parks or other pristine wilderness regions.¹⁶ Third, no individual source is allowed to degrade more than a certain percentage of the existing clean air in an attainment region.¹⁷

From the perspective of both environmentalists and regulated industries, a great deal turns on the question of when the NSR and PSD requirements apply. This question is especially difficult when considering some change to the operation of an existing source of pollution, as opposed to the construction of a new power plant or incinerator from scratch. The NSPS, NSR, and PSD programs can all be triggered by a "modification" of an existing source—but what counts as a "modification?"

The statutory definition of "modification" for purposes of all three programs is "any physical change in, or change in the method of operation of, a stationary source which increases the amount of any air pollutant emitted by such source or which results in the emission of any air pollutant not previously emitted."¹⁸ The "modification" question thus involves a two-step inquiry. First, is the challenged activity a "physical change" or a "change in the method of operation" of the source? Second, if the answer to the first question is yes, does that physical or operational change increase emissions?¹⁹

The answer to the second question turns on the way EPA calculates and projects emissions. Though this subject is important and controversial—and is the main focus of another major clean air rule change recently promulgated by the Bush Administration EPA²⁰ —the focus of this Article is on the first question: what constitutes a physical or operational change for purposes of the CAA?

B. The RMRR Exception: The Current Approach

Though the definitions section of the CAA says that the permitting programs apply to "any physical change [ [ or change in the method of operation"²¹ that increases emissions, EPA realized early on that, given the way the Agency determines emissions increases, a literal reading of "any physical or operational change" would sweep within its scope many activities that the Agency did not think were appropriate subjects of NSR or PSD regulation, and which the Agency thought Congress could not reasonably have intended it to regulate. Thus, in its regulations implementing these provisions of the CAA, EPA explicitly provided that the statutory phrase "physical change or change in the method of operation" did not include "routine maintenance, repair, and replacement" activities.²²

EPA's regulations did not define further what activities counted as "routine maintenance, repair, and replacement." Rather, EPA has made "case-by-case determination[s] by weighing the nature, extent, purpose, frequency, and cost of the work, as well as other relevant factors, to arrive at a common-sense finding."²³ More recently, EPA has described this standard as a "four-factor test."²⁴ To aid parties in understanding how the standard applies in practice, EPA has made available a searchable database of its past applicability determinations.²⁵ EPA also encourages owners and operators who are unsure if a particular planned activity falls within the scope of the RMRR exception "to consult the appropriate reviewing authority for assistance."²⁶ Nonetheless, the precise definition of RMRR is not crystal-clear—it is a standard rather than a rule, though its substantive scope appears, from the available evidence, to be relatively narrow.²⁷

There has been some litigation over the appropriate scope of the RMRR exception, but, interestingly enough, the lawsuits have generally not involved challenges to EPA's authority to create the exception in the first place, nor claims [33 ELR 10792] by environmental groups or other injured parties that EPA has construed the exception too broadly.²⁸ Rather, the most prominent cases that have addressed the proper scope of the RMRR exception have involved assertions by owners or operators of sources that EPA should have exempted some activity from NSR under the RMRR exception, but improperly failed to do so. None of these challenges have been successful.

In the most well-known and most-cited case raising the RMRR issue, Wisconsin Electric Power Co. (WEPCO) v. Reilly,²⁹ the Wisconsin Electric Power Company (WEPCO) wanted to renovate the five coal-fired steam-generating units at its Port Washington power plant.³⁰ This renovation project included replacement of the air heaters and steam drums, as well as repair and replacement of turbine-generators, boilers, and other auxiliary systems.³¹ WEPCO thought that these activities ought to be considered RMRR under the CAA and EPA's regulations, but EPA disagreed, finding that the renovation program would have to comply with the NSPS and PSD permitting requirements.³²

WEPCO sued, claiming that EPA's failure to classify its renovation project as RMRR was inconsistent with the CAA. WEPCO asserted "that Congress did not intend for simple equipment replacement to constitute a physical change for purposes of the [CAA's] modification provisions,"³³ and relied for this interpretive claim on Webster's Dictionary definitions of "modify" as meaning "to change or alter" or "to make basic or fundamental changes in."³⁴ WEPCO asserted that, because its activities would not effect a "fundamental change" in its plants, EPA's decision that the renovation program was covered by NSPS and PSD permitting requirements was inconsistent with the CAA's plain meaning.³⁵

The U.S. Court of Appeals for the Seventh Circuit disagreed. At the outset, the court noted the potential scope of the term "modification" and explained the policy reasons why EPA "promulgated specific exceptions to the modification provisions" including RMRR.³⁶ Interestingly, the court never stopped to question whether EPA has the authority to carve out these exceptions, or explained the Agency's legal justification for doing so. The court went on to note that EPA was entitled to substantial—though not unlimited—discretion in interpreting the CAA.³⁷

The court then turned to WEPCO's "plain meaning" argument—its implicit suggestion that the Agency was not entitled to Chevron Step Two deference because its decision was inconsistent with the clear intent of Congress under Chevron Step One. The court concluded that "what WEPCO calls 'plain' [i.e., the meaning of "modification"] is anything but plain and takes the definition far beyond the words enacted by Congress."³⁸ Congress, the court correctly observed, said "any physical change" that increased emissions would count as a modification.³⁹ The court went further, noting that WEPCO's definition of physical change "would open vistas of indefinite immunity" from the CAA's permitting requirements, and that EPA's reading of that phrase was more consistent with the CAA's technology-forcing goals.⁴⁰

Finally, the court rejected WEPCO's alternative argument that EPA's determination was inconsistent with the Agency's own regulations, noting the "substantial deference due to an agency's interpretation of its own regulations, especially with respect to technical and complex matters."⁴¹ The court assessed the application of EPA's multi-factor balancing test in this case, and found it satisfactory.⁴²

Thus, the WEPCO decision upheld the legitimacy of EPA's relatively narrow, standard-like RMRR exception, but left two critical legal questions unanswered. First, the opinion did not clarify the legal basis for creating the exemption in the first place. Indeed, given the WEPCO court's stress on the plain meaning of Congress' definition of "modification,"⁴³ it would seem that the creation of any RMRR exemption would be legally problematic. But, the WEPCO court never deals with this problem, treating EPA's legal power to create some RMRR exception as self-evident. Second, and relatedly, the WEPCO decision offers little guidance as to what alternative versions of the RMRR exception would also be permissible. The exception that WEPCO and subsequent cases upheld was narrow in scope and standard-like in form. What about a substantively broader exception? Or an exception that eschewed case-by-case inquiry in favor of general rules?

WEPCO does not address these questions. Nor do most of the scattered subsequent cases that deal with the RMRR exception.⁴⁴ Yet they are about to become extremely important, given EPA's recently proposed revision to the RMRR exception.

[33 ELR 10793]

C. The RMRR Exception: The Proposed New Rule

The Bush Administration, of its own accord and at the urging of various regulated industries and other groups, is seeking to effect substantial changes in the government's approach to environmental regulation. With regard to the NSR and PSD programs, EPA is revising its regulatory strategy with respect to both prongs of the definition of "modification"—physical/operational change and emissions increase over a base rate. With regard to the latter, EPA promulgated a final rule on December 31, 2002, that significantly altered how emissions increases are calculated.⁴⁵ On the same day, EPA issued a proposed rule that would broaden and make more systematic the RMRR exemption, thereby altering the reach of the "physical or operational change" prong of the "modification" definition. EPA's proposed rule seeks to replace the relatively narrow, case-by-case approach to the RMRR exemption described above with a more rule-like, across-the-board exclusion for certain types of activities that would henceforth be considered RMRR per se.

What is the reason for this change? A cynic (or realist) might suggest that it is simply an effort by a conservative administration to make it easier for regulated industries to avoid stringent environmental regulation. But there is perhaps a more sympathetic interpretation. EPA, in the preamble to its proposed rule, identified three serious problems with the prevailing case-by-case approach to the RMRR exemption.

First, the present approach creates substantial uncertainty for owners or operators of affected facilities. While a potentially affected owner or operator could seek an applicability determination from EPA, the process can be expensive and time-consuming. If an owner or operator does not seek an applicability determination, the open-ended nature of the current RMRR standards makes it extremely difficult to figure out what is covered and what is not.⁴⁶

Second, because the case-by-case approach is imprecise and complicated, it imposes substantial burdens on state and local reviewing authorities, which often bear the burden of making these complex determinations and ensuring that they are consistent with what EPA or other reviewing authorities require.⁴⁷

Third, and perhaps most importantly, the risks and potential costs associated with NSR or PSD review may deter owners or operators from undertaking activities that should be considered RMRR, and that would, if pursued, improve reliability, efficiency, and safety.⁴⁸ Thus, the case-by-case approach may not only be costly, it may be counterproductive if it deters maintenance and repair activities that could improve energy efficiency and reduce air pollution.⁴⁹

These objections, taken together, are an archetypical example of the case against open-ended standards and in favor of clear, bright-line rules.⁵⁰ And, indeed, EPA's proposed change to the RMRR regulation endorses a rule-like approach.⁵¹ The proposed regulation contains two broad classes of activity that would, if the proposal were adopted, automatically be covered by the RMRR exemption.⁵² These two broad categories are, first, activities whose collective cost is less than a regulatorily determined annual maintenance, repair, and replacement allowance (AMRRA), and, second, activities that are considered pure "equipment replacement" activities that make no functional change in source operation.⁵³

Under the AMRRA provision, EPA would establish an annual allowance for each source that would be equal to the product of the source's replacement cost and an industry-specific percentage set by the Agency. The source would prepare an annual cost calculation, including its annual day-to-day maintenance and operations costs and the cost of each discrete activity carried out at the source in that year for which the source did not intend to obtain an NSR or PSD permit.⁵⁴ The source could then take the general maintenance [33 ELR 10794] costs and add in the costs of the discrete activities one at a time, starting with the cheapest activities and ascending in order of expense, up to the point where the addition of one more discrete activity would cause the total cost to exceed the annual allowance. Those projects whose summed cost is less than the annual allowance would automatically be considered RMRR.⁵⁵

The second proposed category of activity for automatic application of RMRR status is what EPA refers to as "equipment replacement." This categorical exemption from NSR would be limited to "replacement [of] existing equipment with equipment that serves the same function and that does not alter the basic design parameters of a unit … provided the cost of the replacement equipment does not exceed a certain percentage of the cost of the process unit to which the equipment belongs."⁵⁶ The rationale for this change is similar to that for the AMRRA provision, but focuses more on the nature of the activity than on the cost.

EPA in its proposed rule makes clear that it is soliciting comments not only on the merits and drawbacks of each approach, but also on whether they should be adopted together, or whether one is superior to the other.⁵⁷ Yet regardless of whether the final version of the new rule endorses the AMRRA approach, the equipment replacement approach, or both, the new rule will shift the scope and form of the RMRR exemption from a narrowly applicable standard to a broadly applicable rule. Whether this is a good idea or a bad one as a matter of policy is an important question, and one where environmental lobbyists and industry representatives take sharply divergent positions, but it is not the focus of this Article. Rather, the remainder of this Article is concerned with the legal justification for the rule change, and for the RMRR exemption more generally. Prior court decisions did not have to confront this problem directly.⁵⁸ But, with the new rule change, the legal basis question is likely to come to the fore.

II. The Legal Basis for the RMRR Exemption: Two Theories

As we have seen, the existing case law and EPA's own past explanations do not adequately address the question of the RMRR exemption's legal justification. I argue that there are two possible legal bases for the exemption, and, while these justifications are not necessarily mutually exclusive, they are distinct, independent, and, to a considerable extent, in tension.

First, the RMRR exemption might be justified on a "de minimis" theory. Here, the argument would be that, even though the relevant provisions of the CAA cover "any" physical or operational change, EPA may permissibly refuse to enforce the statute's literal provisions against activities that are so trivial, and where regulation would be of such minimal value, that they ought to be considered beyond the scope of the Act.⁵⁹

Second, the RMRR exemption might be justified as a permissible agency interpretation of the phrase "any physical or operational change" under Chevron Step Two. Here, the argument would be that the statutory phrase is sufficiently open-ended that courts can conclude Congress intended to delegate to EPA the authority to determine the meaning of these words. Under Chevron, this determination should be upheld so long as it is reasonable.

The fundamental conceptual difference between these two theories is that the de minimis theory would concede that RMRR activities are "physical or operational changes" but nonetheless exempt them on other grounds, whereas the Chevron theory would hold that RMRR activities are simply not "physical or operational changes" as the Agency defines that allegedly ambiguous phrase.

A close reading of the legal basis section of the preamble to the proposed rule reveals the tension between the theories, though EPA appears deliberately to have obscured rather than highlighted the distinct theoretical grounds for the exemption. Previous drafts of the rule's preamble present the difference between the de minimis and Chevron theories much more clearly. Examination of successive drafts shows how EPA progressively distanced itself from the de minimis theory in favor of something more closely resembling the Chevron theory, but at the same time—perhaps in recognition of the weakness of a pure Chevron approach—incorporated the principles of the de minimis exception without alluding to it explicitly.

A. The Published Legal Basis for the Proposed Change

The preamble to the proposed RMRR rule revision contains a two-paragraph section on the "Legal Basis for Recommended Approaches."⁶⁰ This section, however, is brief, opaque, and barely suggestive of the two quite distinct legal theories that might be advanced to defend the new rule. The first paragraph of the legal basis section merely restates the [33 ELR 10795] statutory definition of the term "modification" in the CAA—a physical change or change in operation that results in an emissions increase—and notes that this definition entails first determining whether a physical or operational change has occurred, and then assessing whether it increases emissions over baseline levels.⁶¹

The second paragraph of the legal basis section is more important, but it is here that a close reading reveals the fundamental ambiguity in the legal theory that EPA means to advance as justification for the RMRR rule revision. On the one hand, the first sentence of the paragraph reads: "The expression 'any physical change … or change in the method of operation' in [§] 111(a)(4) of the CAA is not defined."⁶² The implication of any suggestion by an administrative agency that a statutory term is not defined, or course, is that the agency believes that the courts ought to defer to the agency's interpretation of the statute under Chevron Step Two. But, the phrase "any physical change … or change in the method of operation" does not sound terribly vague or ambiguous. Indeed, this statutory definition is offered precisely to address the potential ambiguity that would otherwise inhere in the statutory term "modification."⁶³ So what is EPA's basis for claiming that physical and operational change are not defined?

The next sentence of the paragraph offers a possible answer, but one that suggests a quite different legal basis for the new rule—one more consistent with a de minimis theory. EPA notes that it has "recognized that Congress did not intend to make every activity at a source subject to the major NSR program."⁶⁴ Thus, EPA continues, it has previously adopted a number of exclusions from "what may constitute a 'physical or operational change,'" one of which is RMRR.⁶⁵ The proposed rule, EPA concludes, does nothing more than clarify this already existing exception.⁶⁶ But observe the subtle argumentative shift in this paragraph. EPA has adopted exclusions—specific types of activities that are not considered "modifications" for purposes of NSR or PSD, even though they would seem to be covered by a literal reading of the statutory definition of the term "modification." But to get to Chevron Step Two, you first have to get past Chevron Step One—whether Congress "has directly spoken to the precise question at issue."⁶⁷ How does EPA justify its exclusions in this paragraph? It invokes a policy consideration—the irrationality of subjecting every physical change, including "the most trivial activities [such as] the replacement of leaky pipes"⁶⁸ —to NSR if the change results in an emissions increase. EPA ascribes this policy consideration to Congress, notwithstanding the language Congress actually used.

Thus, by parsing the language of the legal basis section carefully, one can see the distinct potential theories that might offer legal justification for the RMRR rule revision. First, under a Chevron Step Two theory, EPA might argue that "physical or operational change," though perhaps clearer than "modification," is still a vague enough phrase that there are a range of permissible interpretations, and that among the reasonable constructions of those words is an interpretation that allows the per se exclusion of all the activities covered under the proposed RMRR rule. Second, EPA might claim, in keeping with a de minimis theory, that even though the phrase "physical or operational change" by its literal terms would cover activities that EPA wants to exclude from coverage under the new rule, another principle of statutory construction—one that grants agencies the discretion to refuse to interpret a statutory command literally in cases where it seems unreasonable given the purposes of the statute—trumps the literal meaning of the text.

B. Evidence From Prior Drafts

The best evidence that EPA entertained these two distinct legal theories for the RMRR rule comes from the Agency itself. EPA includes, as part of the publicly available supporting documentation for the proposed rule, several earlier drafts of the rule and its preamble. And, in the earliest publicly available draft of the preamble to the proposed RMRR rule, dated August 28, 2002, EPA cited both the de minimis theory and the Chevron Step Two theory as possible independent legal bases for the rule.⁶⁹ Not only did EPA not commit itself to one position or the other in this earlier draft, but it actually went so far as to "solicit[] comments on [the] two possible alternative legal bases" for the proposed change to the RMRR rule.⁷⁰

The August 28 Draft explains the de minimis justification for the RMRR rule as deriving from the "authority recognized …" by the U.S. Court of Appeals for the District of Columbia (D.C.) Circuit in Alabama Power Co. v. Costle⁷¹ to "decline to regulate activities fitting within statutory proscriptions when such regulation would be of trivial or no value."⁷² The Alabama Power decision's discussion of the de minims exemption did not mention the RMRR exemption specifically. Rather, Alabama Power offers a general discussion of EPA's power to create de minimis exemptions to the CAA, and stresses that "this exemption authority is narrow in reach and tightly bounded by the need to show that the situation is genuinely de minimis…."⁷³ The court admonished that the Agency must show that the regulation yields little or no benefits, not merely that "the acknowledged benefits are exceeded by the costs."⁷⁴ The August 28 Draft appears to accept this relatively narrow scope for the de minimis justification.⁷⁵

The August 28 Draft also includes, as a separate and distinct possible basis for the RMRR rule change, a discussion [33 ELR 10796] of "a Chevron approach" under which EPA would "construe" the relevant statutory provisions "as specifying the type of activities that constitute a 'physical change or change in the method of operation' that could trigger NSR, leaving EPA free to exercise its discretion to determine the type of activities that trigger NSR."⁷⁶ This is confusing wording, but the idea is clear enough: EPA, under Chevron, has the authority to construe "any physical change or change in the method of operation" as not extending to RMRR activities.⁷⁷

The August 28 Draft is so clear on the two alternative legal theories, and yet the published preamble to the proposed rule is so opaque. What happened? Again, earlier drafts included in EPA's public record of the rulemaking process prove to be a valuable resource for answering this question, especially since many of these earlier drafts include red-lining, marginal commentary, and e-mail notes discussing the reasons for changes. These earlier drafts are suggestive of the internal agency process that led to the decision—or lack of a decision—about which theory to rely on.

1. Retreat From the Restrictive De Minimis Theory

The first development that emerges clearly from looking at the successive drafts is the downplaying, and ultimate abandonment, of the de minimis theory. The August 28 Draft's presentation of that theory is striking in four respects. First, this early document made clear that the de minimis justification was not an "interpretation" of the statutory language as such, but rather a legally justifiable decision to refuse to regulate certain activities that fell within the literal meaning of that language. Invoking the de minimis exemption, according to the August 28 Draft, involved EPA's making a decision "to decline to regulate activities fitting within statutory proscriptions when such regulation would be of trivial or no value."⁷⁸ Second, as noted above, the August 28 Draft cited Alabama Power as the basis for the de minimis exclusion; Chevron was not mentioned at all in this context.⁷⁹ Third, the August 28 Draft notes that the scope of the de minimis exclusion is "not without limit," and "should be given a relatively narrow construction."⁸⁰ Fourth, the August 28 Draft suggests that the de minimis theory is the Agency's preferred justification for the exemption, observing that EPA has "advanced the view that the routine maintenance exemption is justified on de minimis grounds" and in the process endorsed the view that this exception should be construed narrowly.⁸¹

For reasons that may seem obvious, EPA in subsequent drafts retreated from this restrictive justification for the RMRR exemption. A revised draft circulated on October 11, 2002, to legal counsel at EPA and the Office of Management and Budget (OMB)⁸² includes red-lining that "depicts all changes made since the start of interagency review."⁸³ Tellingly, the accompanying message advises the recipients to "note especially that [EPA's Office of General Counsel] is still reviewing the revised legal rationale."⁸⁴ Indeed, the revisions to the legal basis section that had already been made by the October 11 Draft are particularly notable in their changes to the exposition of the de minimis exemption.

First, the explicit statement that the de minimis exemption excludes activities that would otherwise fall within the statutory definition disappears. Whereas the August 22 Draft said that under a de minimis approach EPA, relying on Alabama Power, could "decline to regulate activities fitting within statutory proscriptions when such regulation would be of trivial or no value,"⁸⁵ the October 11 Draft, while also citing Alabama Power, says that that case held EPA has the authority to "decline to regulate routine maintenance activities on the grounds that regulation of such activities under the NSR program would be of trivial or no value."⁸⁶ This revision is striking for at least three reasons. First, as rewritten it is a gross mischaracterization of Alabama Power, which says nothing specific about routine maintenance activities. Second, the subtle shift from "when" to "on the grounds that" appears to be a way of implying that the courts should be highly deferential to EPA on the question of whether a de minimis exemption applies.⁸⁷ Although the October 11 Draft does not explicitly invoke Chevron in the context of the de minimis exemption, the shift in language is a sign of things to come. Third, the October 11 Draft deletes the language making clear that the activities covered by the de minimis exemption would otherwise "fit[] within statutory proscriptions."⁸⁸

Next, the October 11 Draft pointedly excludes the limiting language contained in the earlier draft—gone is the statement that the de minimis exemption's scope "is not without limit" and "should be given a narrow construction."⁸⁹ And, in the October 11 Draft, EPA carefully distanced itself from its past reliance on the de minimis exemption. While the August 22 Draft, as discussed above, stated that "EPA has advanced the view that the routine maintenance exemption is justified on de minimis grounds," citing [33 ELR 10797] EPA's brief in Tennessee Valley Authority (TVA) v. Whitman,⁹⁰ the October 11 Draft deletes the reference to TVA and qualifies the more general claim about EPA's endorsement of the de minimis theory by saying that "in the past, EPA has advanced" the de minimis theory.⁹¹

The next draft, dated October 20, 2002,⁹² evinces an even more dramatic shift away from the limited understanding of the de minimis exemption that appeared in the earlier versions. Indeed, the two theories are no longer portrayed as separate. Rather, the decision to exclude RMRR activities from the scope of the statute is described as an exercise of EPA's authority under Chevron "to adopt a reasonable construction of the phrase 'physical change' that is consistent with the policies of the Act and of the NSR program."⁹³ The draft goes on to describe EPA's current view—that the RMRR exclusion "is best viewed as limited to a small category of activities with low cost that are frequently undertaken at a particular source"—as one that "EPA has viewed … as counseled in part by certain statements in Alabama Power."⁹⁴ Thus, by the October 20 Draft EPA no longer articulated the de minimis theory as a separate and distinct legal theory, nor endorsed the restrictions imposed on that theory by Alabama Power and EPA's own past arguments. Rather, Alabama Power's recognition of a de minimis exemption is implicitly folded into the Agency's ability to define ambiguous statutory terms.

2. Formulation and Re-Formulation of the Chevron Theory

Though the legal basis section of the preamble went through several more rounds of modification, the de minimis theory never resurfaced as an explicit justification for the RMRR exception or the proposed rule change. Instead, EPA, in this series of successive drafts, moved toward a greater emphasis on a Chevron theory. But the Chevron argument is tricky, given the apparently clear language of the statute—any physical or operational change counts—and the reinforcing language of judicial decisions that stress the plain meaning of the congressional definition of "modify." EPA's struggles to find a satisfactory legal explanation for why it should be entitled to Chevron deference when defining the scope of the RMRR exception are evident from the earlier drafts.

As noted above, the August 22 Draft does not so much explain the basis for the Chevron theory as assert it.⁹⁵ The October 11 Draft at least tries to clarify the argument by suggesting that not all activities that take place at a source are physical or operational changes, and since "Congress has not specified the types of activities that constitute a physical change or change in the method of operation, EPA is free to exercise its discretion to construe these terms, as long as that construction is reasonable under Chevron."⁹⁶ Of course, this is only a marginal explanatory improvement on the August 22 Draft, since the argument that the RMRR exemption is simply a matter of standard Chevron-style interpretive authority is still just an assertion, without any supporting evidence from case law or practice.

The October 22 Draft tries to remedy this problem. First, it seeks to give the Chevron theory for the RMRR exemption a more historical pedigree, claiming that "ever since promulgation of its original [PSD] regulations in 1980, EPA has defined 'physical change' as not including 'routine maintenance, repair and replacement.'"⁹⁷ While this statement is true in one sense, it is clearly meant to imply that the original basis for the RMRR exemption was the Chevron theory, under which the RMRR exemption arises from an interpretation of the meaning of "physical or operational change." But, as discussed above, EPA's own earlier drafts of the preamble acknowledged that the Agency had in the past relied on a de minimis theory, under which RMRR activities were covered by the literal terms of the statute, but were nonetheless exempt because of their trivial impact.⁹⁸

Second, as noted earlier, EPA in the October 11 Draft attempts to fold the de minimis theory and its associated arguments into the Agency's preferred Chevron theory. While suggesting that its earlier approach to the RMRR exemption—which seemed more like a classic de minimis approach—was "counseled" by "certain statements" in Alabama Power,⁹⁹ EPA stressed in the October 20 Draft that the Agency "does not believe (and has not argued) that its current view is compelled by the language or policies of the Act."¹⁰⁰

Third, the first version of the October 20 Draft also tries to ground the claim that the phrase "physical or operational change" is ambiguous in some existing case law. Thus, this version included the sentence: "As the court stated in WEPCO, the meaning of the phrase 'any physical change' is 'anything but plain.'"¹⁰¹ However, as we have seen,¹⁰² the "anything but plain" language in WEPCO applied to the term "modification," not to "physical change"—and, in fact, the WEPCO court, though itself not entirely clear on this point, seemed to think that the "any physical change" language was a fairly direct and unambiguous congressional instruction. Perhaps because of this problem, the e-mail note accompanying the October 20 Draft from Lee Otis at the U.S. Department of Energy, who wrote the legal basis section of that version, acknowledges that "what I have over-reads the sentence in WEPCO …. I should modify the legal discussion not to rely on it and instead get to Chevron Step [Two] a different way."¹⁰³

For this reason, in the supplement to the October 20 Draft, the reference to WEPCO is deleted and replaced with a different and more subtle argument as to why Chevron deference ought to apply to EPA's RMRR exception. The new version explains:

It is consistent with the plain meaning of the phrase "any physical change" to understand it not to cover [RMRR] activities: they do not change a source's original design, and their function is to maintain the source in its pre-existing [33 ELR 10798] condition rather than to modify it. Moreover interpreting "physical change" to encompass every activity undertaken at a source to maintain it would effectively nullify the distinction Congress drew with respect to coverage between new and existing sources.¹⁰⁴

This argument is interesting in part because, while it is presented as a rationale for Chevron Step Two deference, it is really an argument—note the allusions to "plain meaning" and nullification of congressional intent—that Chevron Step One demands the RMRR exemption. This, however, might be a hard sell after WEPCO and Alabama Power, which seem to treat the RMRR regulations, and de minimis exemptions more generally, as discretionary rather than mandatory. Note also the subtle appeal, recast in definitional terms, to the principles that served to justify the de minimis exemption under Alabama Power—the worry that minor changes that do not alter the basic characteristics of sources are so trivial that they should not be covered by NSR or PSD.

However, this extended explanation and defense of the Chevron theory was quickly abandoned. In the interagency mark-up draft that circulated four days later, the entire legal basis section was struck out.¹⁰⁵ The deleted section was replaced with the brief, two-paragraph legal basis section that ultimately appeared in the published version of the preamble described above.¹⁰⁶ That section, as we have seen,¹⁰⁷ implicitly endorses a Chevron theory, but also incorporates some of the policy considerations associated with the narrower de minimis theory.

C. Are the Theories Really Separable? The Interaction of the Chevron Framework and the Absurdity Canon

The foregoing discussion, and much of what follows, is premised on the notion that there is a legally salient conceptual difference between, on the one hand, interpreting an ambiguous statute and, on the other hand, creating a de minimis exemption to an otherwise unambiguous statute. But this is a contestable proposition. Indeed, although in prior drafts of the RMRR rule's preamble EPA treated the Chevron and de minimis theories as distinct, the Agency might now take the plausible position that the question whether a statute permits a given de minimis exemption is simply a question about statutory clarity. If a statute does not unambiguously preclude the creation of particular de minimis exemptions, the argument would go, then the statute is ambiguous on whether such an exemption is permitted,¹⁰⁸ and that ambiguity is sufficient for the Agency's creation of a de minimis exemption to survive Chevron Step One's inquiry into statutory clarity.¹⁰⁹

A proponent of this argument could further point out that the creation of a de minimis exemption is essentially an application of the "absurdity" canon of statutory construction,¹¹⁰ a canon meant to give effect to the presumed purpose of the legislation (or intent of the legislator)¹¹¹ even if the text seems to point to a different result.¹¹² Thus, if it is not clear whether or how the absurdity canon should apply to the statute at issue, one cannot say that the court, "employing traditional tools of statutory construction, [can] ascertain[] that Congress had an intention on the precise question at issue,"¹¹³ as is required for a finding that a statute is unambiguous for purposes of Chevron Step One.

This argument is compelling in some respects, but it is inadequate as a justification for applying Chevron Step Two deference to any statutory phrase that admits of some possible de minimis exemption. After all, most statutes are not so [33 ELR 10799] "extraordinarily rigid" that they overcome the general presumption in favor of permitting de minimis exemptions,¹¹⁴ but this does not render all such statutes "ambiguous" in the same sense that a statute is ambiguous if it uses vague or open-ended terms, or leaves a gap or conflict that admits of multiple resolutions. "There is a basic difference," the U.S. Supreme Court has observed in other contexts, "between filling a gap left by Congress' silence and rewriting rules that Congress has affirmatively and specifically enacted."¹¹⁵

The critical distinction here is that the creation of a de minimis exemption, unlike most other exercises of statutory construction, involves an explicitly contratextual interpretation of a statute.¹¹⁶ Whereas most canons of construction are designed to give a more determinate meaning to open-ended text, the absurdity canon—of which the creation of a de minimis exemption is an application—arises only when the statutory text is in fact relatively determinate. In the context of Chevron analysis, this distinction means that the debate about whether the absurdity canon justifies the exemption of particular activities from a statute's domain takes place at Chevron Step One.

If a reviewing court, applying the standard tools of statutory construction, were to conclude that the relevant statutory text is indeterminate, the Chevron Step One analysis is normally at an end and the agency is free to pick an interpretation that is reasonable under Chevron Step Two. But, if the relevant statutory text is determinate, then an agency must invoke the absurdity canon to avoid invalidation of an inconsistent rule. That is, before even getting to Chevron Step Two, the agency must first engage in an argument about the "plain meaning" of the statutory text—and this is a debate that occurs at Chevron Step One, where the court does not necessarily defer to agency interpretations.

The clearest authority for this understanding of the interaction of Chevron and the absurdity canon is the D.C. Circuit's opinion in Mova Pharmaceuticals v. Shalala,¹¹⁷ which held that an agency that "concludes that a literal reading of a statute would thwart the purposes of Congress [] may deviate no further from the statute than is needed to protect congressional intent…. Our review of the agency's deviation from the statutory text will occur under the first step of the Chevron analysis, in which we do not defer to the agency's interpretation of the statute."¹¹⁸ The view that the applicability of the absurdity canon is a matter of Chevron Step One analysis has been accepted even by commentators who endorse broad agency power to create de minimis exemptions from broadly worded statutes.¹¹⁹

Thus, despite some incautious language in earlier decisions suggesting that a statute with a literal meaning that might lead to some absurd results "has no plain meaning,"¹²⁰ the agency does not get a free pass from Chevron Step One simply by demonstrating that the statute does not unambiguously preclude any de minimis exemptions. The agency must still establish that the particular categories of activities it wants to exempt are not covered by the "plain meaning" of the statute, despite the literal language.¹²¹

Thus, there is a real distinction between the de minimis theory and the Chevron Step Two theory as justifications for the proposed RMRR rule modification. The Chevron Step Two theory involves a claim that textual analysis—informed by context, statutory structure, and (perhaps) legislative history—yields no determinate answer on whether the CAA definition includes the activities EPA wants to exclude as RMRR. The de minimis theory, in contrast, would apply if the textual analysis does yield a determinate answer, but the Agency wants to argue that this result, i.e., the failure to exclude the proposed RMRR activities from the CAA's permitting requirements, is so absurd that a literal interpretation of the statute would be inconsistent with the statute's plain meaning, as a matter of Chevron Step One analysis.

This is not necessarily to say that the court should treat an agency's creation of a de minimis exemption less deferentially than an agency's interpretation of an ambiguous statute. Although Mova indicated that agency invocation of the absurdity canon ought to receive no special deference,¹²² other decisions¹²³ and commentators¹²⁴ suggest that courts ought to treat agency authorization of de minimis exemptions with considerable deference. If the appropriate level of deference for agency-created de minimis exemptions is the same as that applied to agency interpretations of textually ambiguous terms (an argument I consider and reject below),¹²⁵ then the distinction between the Chevron Step Two theory and the de minimis theory would be a distinction without a difference. But if, as I argue, a different standard of review is appropriate, then the distinction is meaningful.

[33 ELR 10800]

III. Choosing a Theory: The Case for De Minimis Over Chevron Step Two

The preceding examination of the earlier drafts of the legal basis section of the RMRR rule's preamble indicates a retreat from the de minimis theory and an endorsement of the Chevron Step Two theory. However, while this legal strategy may appear tempting to an agency that hopes to avoid aggressive judicial review, it is sufficiently problematic from a doctrinal point of view that EPA ought to think twice. Moreover, if EPA were to prevail in upholding this proposed rule on Chevron Step Two grounds, the congressional and judicial constraints on agency action would be needlessly weakened.

The RMRR rule, or some version of it, should—and probably can—be justified as an exercise of EPA's power to create de minimis exemptions. The Agency's judgment on this point should receive Skidmore deference. The creation of such an exemption involves derogation of an explicit congressional command, but the exemption is justified on the theory that the Agency's expert assessment leads it to conclude that following the literal instructions of Congress would be absurd or otherwise inconsistent with the statute's purpose. This is therefore not a situation where we can reasonably conclude that Congress left the Agency a gap to fill, but it is a situation where the Agency's expert judgment ought to receive particular respect. Skidmore deference strikes the right balance, preserving agency flexibility and discretion, but also creating appropriate incentives for the Agency to tailor its rule narrowly and discouraging it from using its de minimis authority improperly to narrow the scope of the statute.

A. The Case Against the Chevron Theory

EPA appears partial to a Chevron theory, under which the Agency's creation and subsequent modification of the RMRR exemption is grounded in an interpretation of the purportedly ambiguous statutory phrase "any physical change or change in the method of operation." However, to get to Chevron Step Two—where the Agency's interpretive choice would be entitled to deference—EPA must first surmount Chevron Step One, which asks whether Congress has "directly spoken to the precise question at issue."¹²⁶ And it is here that EPA's Chevron theory runs into its greatest difficulties. After all, is it really plausible that "any physical change or change in the method of operation" is sufficiently ambiguous to justify the exclusion of RMRR on a Chevron Step Two basis? At least four considerations support a negative reply.

First, the use of the word "any" to modify "physical change or change in the method of operation" is strongly indicative of a congressional purpose that the statute should apply broadly. The Court held in United States v. Gonzales¹²⁷ that, "read naturally, the word 'any' [in a statute] has an expansive meaning, that is, 'one or some indiscriminately of whatever kind.'" And, as the U.S. Court of Appeals for the Eighth Circuit observed in Missouri Municipal League v. Federal Communications Commission,¹²⁸ "time and time again, the Court has held that the modifier 'any' prohibits a narrowing construction of a statute."¹²⁹

A second argument from "plain meaning"—though perhaps a weaker one—would draw on the Court's opinion in Microwave Communications Inc. (MCI) v. American Telephone & Telegraph Co. (AT&T)¹³⁰ There, the Court held that the term "modify" as used in the Federal Communications Act had a sufficiently plain meaning that the Federal Communications Commission's interpretation of that term was invalid under Chevron Step One. Justice Antonin Scalia's majority opinion rejected the argument that "modify" could mean "to change fundamentally"; he held that, as a matter of textual meaning, the word modify "connotes moderate change."¹³¹ Interestingly, the unsuccessful petitioners in MCI had relied on essentially the same Webster's Dictionary definition that WEPCO had advanced unsuccessfully before the Seventh Circuit.¹³² Justice Scalia characterized this definition of modify—as denoting a "fundamental" change—as "peculiar," perhaps reflecting "intentional distortions, or simply careless or ignorant misuse."¹³³ Given this definitional precedent, albeit in a different statutory context, the definitions section of the CAA might plausibly be read to broaden the definition of "modify" to include large changes as well as small ones, but it cannot be read to include only large changes. Such a definition would be flatly inconsistent with the plain meaning of "modify" as established in MCI.¹³⁴

[33 ELR 10801]

Third, as a matter of inference from context and statutory structure, it is significant that the allegedly ambiguous "physical or operational change" language appears in the "definitions" section of the CAA. Congress, apparently aware that the word "modification" in the operative section of the statute was potentially ambiguous, sought to define the word with greater precision. One would presume that the definition of a term is not meant to be as ambiguous as the term it is supposed to define. The more natural reading is that the phrase EPA claims is ambiguous in fact reflects an effort to make clear that the words "modify" and "modification," as used elsewhere in the statute, are not as ambiguous as they might otherwise appear. Refusing to read any additional precision into the definitional phrase than is inherent in the original term would render the congressional effort at definition an exercise in futility.

A fourth consideration also involves statutory context. As has already been noted, "physical or operational change" is only one part of the statutory definition of "modification." A "modification" also has to increase emissions, or result in the emission of new pollutants. Read in isolation, the "any physical or operational change" language tempts the interpreter to conclude that the language is ambiguous, because a literal reading seems to sweep so broadly as to be patently unreasonable.¹³⁵ But the necessary limiting principle is supplied by the emissions increase prong of the definition. Read in the context of that second prong, the first prong is understandably, and perhaps deliberately, quite broad.¹³⁶

For these reasons, the potential legal attack on Chevron deference to the RMRR exemption is stronger than is usually the case when an administrative agency interprets a regulatory statute. EPA might therefore want to be wary of relying exclusively on a Chevron Step Two theory. True, chances are that the new RMRR rule would probably be upheld, given the D.C. Circuit's tendency to uphold challenged agency action—and, perhaps, given the court's ideological sympathy with the new rule's objectives.¹³⁷ But, it is nonetheless a risky strategy, especially if EPA chooses to fold the de minimis argument for its rule change into its Chevron Step Two theory. Doing so might deprive the Agency of an independent and more defensible ground for upholding the new rule.

Stepping out of EPA's shoes and into the position of a court, or of someone interested more broadly in the development of environmental and administrative law doctrine, upholding the RMRR rule on the basis of a Chevron Step Two theory is a bad idea. With respect to environmental law in particular, upholding the RMRR rule modification on this basis would allow EPA to exclude all sorts of other activities from the definition of "change"—even if those activities significantly increased emissions—so long as the exclusions conform to some "reasonable" definition of change. And, if we are now in a world where "any change" actually means "any big change," it is hard to articulate a judicially enforceable limiting principle as to the acceptable location of the line between "big" and "not big." Granting EPA Chevron deference in its definition of "physical or operational change" is likely to lead, as the WEPCO court feared, to "vistas of indefinite immunity from the provisions" of the CAA.¹³⁸ Concluding that a statutory provision is ambiguous, for purposes of Chevron, whenever the statute does not expressly foreclose the creation of any de minimis exemption whatsoever¹³⁹ is likely to lead to similar results—unless courts become more willing to find such foreclosure, in which case the result would be an undesirable loss of agency flexibility.¹⁴⁰

Taking a more general view, Chevron already effects a considerable shift of power from Congress to administrative agencies.¹⁴¹ The only real limitations on agency interpretive power after Chevron are, first, the requirement that even under Chevron's deferential standard the agency's interpretation must be "reasonable," and, second, Chevron Step One's analysis of the language Congress actually used. That latter check is rendered nugatory in an interpretive regime where "any physical or operational change," as used in the context of the CAA, is sufficiently ambiguous that EPA can, subject only to "reasonability" review, define this phrase (virtually) any way it likes. But, as the Court has stressed in other contexts, EPA should not be allowed to "construe [the CAA] in a way that completely nullifies textually applicable provisions meant to limit its discretion."¹⁴² Upholding the RMRR rule on the basis of a Chevron Step Two theory would thus be a dangerous precedent for those who are concerned about the balance of power in the administrative state.¹⁴³

B. The Case for the De Minimis Theory

Reliance on the problematic Chevron Step Two theory may also be unnecessary, given the strong arguments that EPA can make for upholding the RMRR rule revision solely as an [33 ELR 10802] exercise of the Agency's authority, recognized in Alabama Power, to create de minimis exceptions to the CAA. As we have seen, EPA's early drafts of the preamble endorsed this theory as a possible basis for upholding the rule, but subsequent versions shied away. The reason for this shift was most likely the more limited nature of the de minimis exemption, and the greater possibility for judicial scrutiny. However, EPA may have a fairly strong case that the de minimis exception justifies the revised RMRR rule.

Certainly, the old version of the rule—in which EPA granted exemptions on a case-by-case basis, and interpreted the scope of the exemption narrowly—is defensible strictly on a de minimis basis. EPA's new RMRR proposal, however, has a more rule-like form, and is broader in scope. Do these changes make the rule harder to defend as an exercise of de minimis power? Not necessarily.

Though the extant law is not entirely clear on the point, the use of a rule rather than a standard should not render an exercise of de minimis authority invalid, even though all rules are acknowledged to be overinclusive and underinclusive. Indeed, several of the cases upholding administrative agency authority to create de minimis exemptions have involved rule-like categories or cutoff points.¹⁴⁴ So long as EPA defends the scope of its rule—and provides data and analysis sufficient to justify its choice—the fact that EPA uses a categorical rule rather than a standard applied on a case-by-case basis should not be a reason for invalidating the RMRR exemption.¹⁴⁵

Choosing a rule-like form does, however, make the question of scope more salient. EPA's proposed rule makes certain activities RMRR per se, so a defense of the rule on de minimis grounds is an implicit assertion that the overwhelming majority of the activities that escape regulation under the proposed rule would in fact be considered de minimis activities if they were examined on a case-by-case basis. Rule-like categorical exemptions for purportedly de minimis activities are most vulnerable to invalidation if they exempt projects that fall squarely into the set of activities that the statute is supposed to regulate.¹⁴⁶ Thus, to defend its RMRR proposals, EPA would have to show that its categorical exemptions do not sweep so broadly that they cannot be fairly classified as merely de minimis exemptions.

To satisfy this standard, EPA would need to make some evidentiary showing that the general rule is, to borrow a phrase from an unrelated body of doctrine, "congruent and proportional" to the set of activities that the Agency could permissibly refuse to regulate on de minimis grounds.¹⁴⁷ Circuit courts have already adopted something like this requirement, applying a kind of "narrow tailoring" analysis when an agency invokes the absurdity doctrine to justify a contratextual rule. The clearest example is Mova, in which the D.C. Circuit rejected the U.S. Food & Drug Administration's (FDA's) invocation of the absurdity doctrine to justify an agency policy on new drug approvals that contravened the relevant statutory text. The "fatal flaw in the FDA's 'absurd results' argument," the court explained, "is that the agency could have … created narrower exceptions" instead of the broad contratextual rule the agency had chosen.¹⁴⁸ The Mova court did not deny that agencies have the power to deviate from the literal language of the statute in order to avoid results that are truly absurd; indeed, the court invited the FDA to address the absurd results problem in "some narrower way, as long as that solution conforms to the statute."¹⁴⁹ What they cannot do, as the Mova court put it, is to "embark[] upon an adventurous transplant operation in response to blemishes in the statute that could have been alleviated with more modest corrective surgery."¹⁵⁰ An implicit adoption of something like this requirement in the context of de minimis exemptions from environmental regulation also appears in Ober v. Whitman,¹⁵¹ where the U.S. Court of Appeals for the Ninth Circuit held that it would only defer to EPA's selection of a de minimis threshold for particulate matter if the Agency "has provided a full explanation of its de minimis levels and its application of those levels to sources of pollution," supported by adequate data.

EPA could provide the requisite evidence of congruence and proportionality in a number of ways. The Agency could provide data on the approximate cost of various projects and their associated environmental impacts (or lack thereof), as well as data on the average annual maintenance costs for various kinds of sources. It could also use its past determination decisions to establish that the overwhelming majority of the activities denied RMRR exemptions under the new rule would also have been denied under the old standard. Inasmuch as a de minimis theory would require EPA to engage in this kind of study—and perhaps would encourage the Agency to narrow the scope of the exemption where it otherwise would sweep in too many non-RMRR activities—the use of this theory would have beneficial policy effects.

[33 ELR 10803]

C. The Appropriate Level of Deference for De Minimis Exemptions

A critical question will be the level of scrutiny to apply to EPA's decision as to where to draw the line separating per se de minimis activities from other activities. Does Chevron deference apply to this decision, even though, as we have seen, a de minimis exemption is not a traditional exercise of Chevron Step Two interpretive authority? Should the courts construe the Agency's authority to create de minimis exemptions narrowly, and therefore apply a much more rigorous standard of review? Or something in between? This question has attracted surprisingly little attention, even from sophisticated commentators.¹⁵²

1. The Impact of Chevron on Deference to Agency De Minimis Determinations

Alabama Power appeared to envision relatively aggressive judicial review of the scope of de minimis exemptions. While the court noted that, in most cases, agencies have the power to create these exemptions,¹⁵³ it put the burden squarely on the agency to "make[] the required showing" that the matters excluded are truly de minimis¹⁵⁴ and admonished EPA that its authority to create de minimis exemptions "is narrow in reach and tightly bounded" by the need to make this showing."¹⁵⁵ However, Alabama Power was decided before Chevron. To the extent that we think the latter decision reshaped administrative law with respect to judicial deference to agency statutory interpretation,¹⁵⁶ we might well inquire whether and to what extent courts ought to show greater deference to agency creations of de minimis exceptions than Alabama Power implied was appropriate. Here, it is useful and important to distinguish two ways the Chevron decision could affect the power of an administrative agency to create a de minimis exemption.

First, Chevron might mean that a particular statutory phrase—a phrase that, pre-Chevron, the courts would have interpreted themselves—is sufficiently ambiguous that the agency has the power to interpret it. Therefore, whereas pre-Chevron an agency may have had to justify its exclusion of certain matters as a de minimis exemption to the court's construction of the "real" meaning of the statute, post-Chevron the agency need only show that its interpretation of the statutory language to exclude those matters is reasonable. Indeed, there is no need for a special de minimis theory in such a case. Where the statutory language is ambiguous, the agency may give its own construction to that language so long as the construction is reasonable—this is standard Chevron analysis.

To illustrate with a hypothetical example, imagine that the CAA did not define "modification." Suppose further that EPA did not want to subject the repair or replacement of old, leaky pipes to NSR permitting. Pre-Chevron, the court would have had to determine what the term "modification" means. If the court concluded that the best interpretation of "modification" includes pipe repair, then EPA would have had to justify its exclusion of pipe repair from NSR on a de minimis theory. But now suppose Chevron comes along. EPA interprets "modification" to mean "change affecting basic characteristics or operation," and concludes that pipe repair does not qualify. The reviewing court, applying Chevron, would say, "although we might have interpreted the meaning of 'modification' differently, the term is ambiguous, and the agency's interpretation is not unreasonable, so we defer."¹⁵⁷ Because of Chevron, the agency would never have to invoke the de minimis theory at all, and would have no need to cite Alabama Power.

The interesting question, for purposes of applying the de minimis theory, is how much deference an agency's creation of a de minimis exemption should receive when the statute is not sufficiently ambiguous to infer a delegation of statutory interpretation authority to the agency. Where the statutory text clearly covers certain activities—and, according to the arguments above, the first prong of the CAA's modification definition meets that standard¹⁵⁸ —when can an agency nonetheless decline to interpret the statute as reaching those activities? As we have seen, Alabama Power recognized that agencies have this power, aptly comparing it to the "absurdity" canon in judicial statutory construction,¹⁵⁹ though the scope of that power is tightly constrained.

Perhaps Chevron might expand this power by affecting the amount of deference an agency creation of a de minimis exemption is due, even though the statute is not sufficiently ambiguous to get to Chevron Step Two. This second argument as to how Chevron affects the judicial deference to agency de minimis exemptions is more indirect—the essential idea is that the Chevron decision embodies a particular deferential attitude toward administrative agencies, and this attitude ought to carry over into other forms of agency statutory interpretation, including the invocation of the absurdity doctrine to create a de minimis exemption.

The actual RMRR exemption—if one stipulates for the moment that I am correct in arguing that the statutory definition is sufficiently clear that we would never get to Chevron Step Two as a matter of conventional Chevron analysis—serves as an example here. Suppose the RMRR exemption were challenged pre-Chevron. The exemption would be [33 ELR 10804] defended on a de minimis theory, and the reviewing court, applying the Alabama Power standard, would require the agency to make a detailed showing that the matters it excludes are truly de minimis. Now suppose the challenge is made post-Chevron. One might argue that, even though we have stipulated that the court would hold, for purposes of Chevron Step One, that RMRR activities are covered by the language of the statute, EPA's creation of a de minimis exemption should receive the same level of deference that the Agency would receive if it were interpreting a textually ambiguous statute at Chevron Step Two. After all, the argument would go, these are both exercises of the Agency's statutory interpretation authority. At Chevron Step Two, the Agency interprets statutory language that is not plain, pursuant to an implicit delegation by Congress; when creating a de minimis exemption, an agency invokes an interpretive canon to give effect to the true purpose of the legislation, and this power also arises from implicit congressional delegation.

There appears to be something of a circuit split—or at least some circuit confusion—on this latter issue of whether the creation of a de minimis exemption to an apparently clear statutory command is entitled to a higher level of deference post-Chevron. The most extensive discussion of the issue comes in the D.C. Circuit's 1996 decision in Environmental Defense Fund v. U.S. Environmental Protection Agency.¹⁶⁰ There, in response to an environmental group's claim that EPA had not satisfied the level of justification demanded by Alabama Power for the creation of a de minimis exception, the court pointed out that it had

decided Alabama Power … before the Supreme Court's decision in Chevron, which clarified the degree to which a reviewing court should defer to an agency acting within the scope of its delegated authority, whether implicit or explicit. To the extent that both Chevron and Alabama Power address agency power inherent in a statutory scheme, the same deference due to an agency's reasonable interpretation of an ambiguous statute may also be due to an agency's creation of a de minimis exemption.¹⁶¹

Though this would seem like a fairly conclusive statement from the circuit most likely to review the RMRR rule that Chevron applies to the creation of de minimis exemptions, the Environmental Defense Fund decision does not necessarily settle the question. First, the opinion is quite explicit that while the creation of de minimis exemptions may get Chevron deference, the court in that case did not need to "resolve whether, under Chevron, an agency may create a de minimis exemption with a justification less rigorous than we indicated in Alabama Power, because [in this litigation] the EPA has adequately explained itself even by the standard of the latter case."¹⁶²

Also, the Environmental Defense Fund opinion appears to blend the two distinct arguments for why Chevron deference might attach to an agency's exemption of low-significance activities from the scope of a regulatory statute. The language just quoted suggests the second of the two arguments discussed above: that even where the language of the statute is clear enough to satisfy Chevron Step One, the creation of a de minimis exemption ought to receive the same level of deference that would obtain at Chevron Step Two. However, earlier in the Environmental Defense Fund opinion the court explicitly argues that the statutory phrase "any activity" at issue in that case,¹⁶³ when read in context, is sufficiently ambiguous that EPA could reasonably interpret it to mean "any activity that is likely to interfere with the attainment goals in a [state implementation plan]."¹⁶⁴ That implies an argument more along the lines of the first of the two Chevron-related claims discussed above—the statute is ambiguous enough that the agency is entitled to deference at Chevron Step Two, and so what looks like the creation of a de minimis exemption is properly understood as a straight-forward exercise of construing ambiguous language. Viewed in that light, the Environmental Defense Fund court's discussion of Alabama Power is unnecessary dicta.

It is also not clear how to square the Environmental Defense Fund court's discussion of Chevron's impact on Alabama Power with the D.C. Circuit's subsequent holding in Mova that, when an agency invokes the absurdity canon to escape the literal language of a statutory provision, judicial "review of the agency's deviation from the statutory text will occur under the first step of the Chevron analysis, in which [the courts] do not defer to the agency's interpretation of the statute."¹⁶⁵

In apparent contrast to the D.C. Circuit's Environmental Defense Fund approach, the Ninth Circuit, in Natural Resources Defense Council, Inc. v. U.S. Environmental Protection Agency,¹⁶⁶ rejected EPA's claim that it has the power to create a de minimis exemption to certain provisions of the Clean Water Act (CWA) for certain small sites. The court, applying the Alabama Power test, held that the de minimis exemption is only available where the regulation would "yield a gain of trivial or no value" and that EPA had not provided data sufficient to show that the de minimis exemption at issue would "indeed have only a de minimis effect."¹⁶⁷

The Ninth Circuit appeared to shift to a more deferential posture in the Ober¹⁶⁸ decision, though Ober tried to preserve the Natural Resources Defense Council court's emphasis on the agency's duty to provide sufficient data to justify its de minimis exemption. In Ober, the court stated that it would "defer to the agency's judgment only if EPA has provided a full explanation of its de minimis levels and its application of those levels to sources of pollution. Once we have received that explanation, however, we owe deference to the agency's decision if it is a permissible interpretation of the statute."¹⁶⁹ But, while Ober, like Environmental Defense Fund, endorses Chevron-style review when the statute [33 ELR 10805] is unclear, it does not address the question of whether Chevron means that the Alabama Power standards ought to be relaxed when the statute is clear enough, as a matter of text and context, that Chevron Step One would otherwise be dispositive.¹⁷⁰

Thus, while some—but not all—courts to have considered the question seem to conclude that Chevron affects the application of the de minimis exemption authority recognized by Alabama Power, these cases are not entirely clear about how or why this is so. The nature of the confusion seems to arise from the conflation of the two distinct lines of argument as to how Chevron might affect an agency's power to exclude certain low-impact activities from a regulatory statute's reach. The cases to date have considered situations where the creation of the so-called de minimis exclusions clearly takes place at Chevron Step Two, because an analysis of text and context yields no determinate result. The more rigorous Alabama Power standard of review has no applicability for these sorts of determinations. After all, it would be bizarre to subject EPA's decision to consider an entire power plant to be a "stationary source"¹⁷¹ —a decision with potentially enormous economic and environmental consequences—to a more deferential standard of review than an EPA decision to exclude activities with a trivial impact from the scope of the statute. The RMRR case, as we have seen, is different because of the need to invoke the de minimis exemption at Chevron Step One in order to get around the insurmountable obstacle that the text of the statute would otherwise pose.

2. The Case for Skidmore Deference for De Minimis Exemptions From Clear Statutory Language

What, then, is the appropriate level of deference in a case where Congress has "spoken clearly"—that is, used statutory language of seemingly sufficient clarity to satisfy Chevron Step One and preclude application of Chevron Step Two—but the agency nonetheless wants to create a de minimis exemption to the apparently clear statutory command? The best approach would be for the courts to review these types of agency determinations under the Skidmore¹⁷² standard. Skidmore is less deferential than Chevron but more deferential than Alabama Power, and strikes the appropriate balance between granting the agency appropriate discretion to administer a complex statute and constraining the agency's ability to circumvent statutory limitations on its power.

Skidmore deference—recently revived as an alternative to Chevron deference by the Court in the Christensen v. Harris County¹⁷³ and United States v. Mead Corp.¹⁷⁴ opinions—treats agency decisions not as "controlling" but as "entitled to respect" proportional to "the thoroughness evident in [the decision's] consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control."¹⁷⁵ Skidmore deference, according to these cases, attaches when the court concludes that Congress did not intend the type of agency determination in question to have "the force of law."¹⁷⁶ How the court is supposed to determine whether a particular agency decision should have the force of law is the subject of some confusion and considerable dispute.¹⁷⁷ The most important factor in establishing whether an agency action has the force of law, according to Mead and Christensen, is whether the agency decision was made pursuant to notice-and-comment rulemaking or formal adjudication.¹⁷⁸ But, those cases note, such formal procedures are not a necessary condition for Chevron deference.¹⁷⁹

Nor should such formal procedures always be sufficient. Even though the RMRR exemption is being promulgated according to notice-and-comment rulemaking, with all the attendant safeguards, the RMRR rule involves derogation from the literal meaning of the language used by Congress. It may be sensible to infer an implicit congressional delegation of interpretive authority to an agency, subject only to reasonability review, when Congress has been silent or unclear. Although it is also sensible to infer congressional delegation to the agency of authority to create de minimis exemptions,¹⁸⁰ it is much less plausible to suppose that Congress would want the courts to apply to those exemptions a standard of review so deferential as to "manufacture for an agency a revisory power inconsistent with the clear intent of the relevant statute."¹⁸¹ When an agency creates a de minimis exemption, it is not interpreting a statute that is silent or ambiguous—rather, the agency, as discussed above, is invoking the absurdity canon to escape an otherwise clear command. The invocation of that canon must be subject to stricter judicial review, or else the crucial distinction between [33 ELR 10806] Chevron Step One and Chevron Step Two dissolves, and Congress' power to control agency action through clear statutory language erodes.

But, because the creation of a de minimis exemption involves an exercise of the agency's expertise and judgment, Skidmore respect is more appropriate than pure de novo review.¹⁸² The fact that the creation or revision of the exemption is subject to notice-and-comment rulemaking, while not enough to entitle the agency to Chevron deference, is an additional reason to accord the determination particular weight. The notice-and-comment process gives interested parties an opportunity to attempt to demonstrate that the rule sweeps too broadly to be considered a de minimis exemption, and puts pressure on EPA to provide data and analysis to support its assertion that the rule is sufficiently tailored. As long as EPA can make that case persuasively, the court should be reluctant to second-guess the Agency's line-drawing choices. But, the heightened level of scrutiny provides a critical check on the Agency's power to scale back the scope of a broadly worded statute through the creation of categorical exemptions from the statute's reach.¹⁸³

Conclusion

The legal basis for EPA's proposed revisions to the rule exempting routine maintenance, repair, and replacement activities from the permitting requirements of the CAA is not clear. The exemption, and the proposed revision, might be upheld on one or the other of two legal theories. First, it may be that the creation of the exemption is a permissible exercise of the Agency's power to create "de minimis" exceptions to otherwise clear statutory commands. Alternatively, it might be argued that the exclusion of routine maintenance activities from the scope of the statute's permitting requirements is an exercise of the Agency's authority to interpret the purportedly ambiguous phrase "any physical change or change in the method of operation"—an interpretive decision which is allegedly entitled to Chevron deference. Earlier drafts of the proposed rule's preamble demonstrate that EPA understood these to be two distinct theories, but chose to blend them, de-emphasizing the de minimis theory as a distinct argument and instead stressing the applicability of the Chevron theory.

The emphasis on the Chevron theory at the expense of the de minimis theory is unwise and undesirable, given the strong arguments against the notion that the statutory provision at issue is sufficiently ambiguous to justify Chevron deference. Relying solely on a Chevron argument increases the risk that EPA will lose. More importantly, if EPA wins on the basis of such an argument, the decision would have the undesirable effect of eroding important checks on agency action built into current environmental and administrative law. The de minimis theory is more theoretically satisfying and avoids these substantive problems. EPA (and other agencies) should be allowed to create rule-like de minimis exemptions even to clear statutory commands, but the specific exemptions must be justified, by evidence and analysis, as sufficiently narrow and limited. EPA's showing on this score ought to be subject to intermediate Skidmore deference—a standard that properly balances the tension that arises, when carving out these sorts of exceptions, between the value of agency expertise and the need for congressional control.

[Author's Note: On August 27, 2003, as this Article was about to go to press, EPA issued a final rule on its modifications to the RMRR exemption. See http://www.epa.gov/nsr/ERP_merged_8-27bh.pdf. EPA took no action on its proposal to create an annual maintenance allowance, but left open the possibility that the agency might take some action on that proposal at a future date. EPA did, however, adopt a version of the equipment replacement approach, under which such replacement is automatically considered to be RMRR, and therefore exempt from NSR and PSD permitting requirements, so long as: (1) the replacement components are identical to, or functionally equivalent to, the components being replaced; (2) the cost of the equipment replacement is less than 20% of the value of the process unit of which the replacement component is a part; (3) the replacement does not alter the process unit's "basic design parameters"; and (4) the process unit continues to meet other enforceable emission and operations limitations.

The most important feature of the preamble to the August 27 final rule, as it relates to the arguments advanced in this Article, is the extensive "legal basis" section. In place of the two relatively opaque paragraphs that appeared in the December 2002, preamble to the proposed rule, the preamble to the final rule contains a lengthy discussion and analysis that explicitly grounds the creation and modification of the RMRR exception in EPA's power to interpret the allegedly ambiguous phrase "physical change or change in the method of operation." The preamble expressly relies on this Chevron theory, quoting extensively from the Chevron opinion, elaborating the policy rationales for allowing agencies to define (and redefine) ambiguous statutory terms, and purporting to demonstrate that "change" as used in the CAA is sufficiently ambiguous that Chevron deference ought to apply to EPA's construction. Moreover, the preamble to the final rule expressly rejects the de minimis theory as the basis for the RMRR exemption, in spite of the position EPA admits it had taken in prior litigation and enforcement actions.

Thus, EPA has come down conclusively on the side of a Chevron theory and against a de minimis theory. It is conceivable, of course, that specific features of EPA's final rule—which adopted a relatively constrained version of the equipment replacement approach—make the Chevron argument more plausible than would have been the case if EPA endorsed the annual maintenance allowance or a broader version of the equipment replacement approach. Indeed, the August 27 preamble appeared to acknowledge that EPA imposed the requirement that the equipment replacement not alter the "basic design parameters" of the process unit precisely because any activity that did change basic design parameters would necessarily be a "change" under the CAA. However, the rejection of the de minimis theory in favor of the Chevron theory remains problematic, for all the reasons discussed in this Article. The RMRR exemption, for better or worse, involves derogation from broad statutory language that cannot, despite EPA's protestations, fairly be characterized as ambiguous. The de minimis theory is an adequate and sensible basis for allowing EPA to exempt truly minor repair and replacement activities, while at the same time preserving adequate constraints on agency discretion by subjecting the creation of the exemption to more searching judicial review.]

1. 42 U.S.C. §§ 7401-7671q, ELR STAT. CAA §§ 101-618.

2. 467 U.S. 837, 14 ELR 20507 (1984).

3. The term "de minimis" is short for the phrase de minimis non curat lex ("The law does not concern itself with trifles.").

4. 323 U.S. 134, 139-40 (1944).

5. Pub. L. No. 91-604, 84 Stat. 1676.

6. 42 U.S.C. §§ 7501 et seq., ELR STAT. CAA §§ 171 et seq.

7. Id. §§ 7502(c)(5), 7503(a), ELR STAT. CAA §§ 172(c)(5), 173(a).

8. Id. § 7503(a)(2), ELR STAT. CAA § 173(a)(2).

9. According to the statute, LAER is

that rate of emissions which reflects (A) the most stringent emissions limitation which is contained in the implementation plan of any State for such class or category of source, unless the owner or operator of the proposed source demonstrates that such limitations are not achievable, or (B) the most stringent emission limitation which is achieved in practice by such class or category of source, whichever is more stringent.

Id. § 7501(3), ELR STAT. CAA § 171(3).

10. Id. § 7503(a)(1), ELR STAT. CAA § 173(a)(1). In addition to the offset and LAER requirements, NSR permits also require that the owner or operator of the proposed new or modified source show that all major stationary sources in the state that he/she owns or operates are in compliance with all applicable emissions limitations, id. § 7503(a)(3), ELR STAT. CAA § 173(a)(3); that the state's implementation plan is being adequately implemented, id. § 7503(a)(4), ELR STAT. CAA § 173(a)(4); and that a cost-benefit analysis shows the benefits of the new or proposed source "significantly outweigh" the environmental and social costs. Id. § 7503(a)(5), ELR STAT. CAA § 173(a)(5).

11. Id. §§ 7470 et seq., ELR STAT. CAA §§ 160 et seq. The 1977 Amendments affirmed and codified the principle first articulated in the controversial decision in Sierra Club v. Ruckelshaus, 344 F. Supp. 253, 2 ELR 20262 (D.D.C. 1972). For a comprehensive review of the CAA Amendments, see ARNOLD W. REITZE JR., AIR POLLUTION CONTROL LAW: COMPLIANCE AND ENFORCEMENT (Envtl. L. Inst. 2001).

12. See 42 U.S.C. § 7470, ELR STAT. CAA § 160.

13. While the substantive section of the statute imposes the PSD permitting requirements in case of "construction" of a new "facility," id. § 7475(a), ELR STAT. CAA § 165(a), the definitions section of the statute makes clear that, for purposes of the PSD program, "construction" includes "modification" as defined for purposes of NSPS review, id. § 7479(2)(C), ELR STAT. CAA § 169(2)(C), and "facility" means any of several dozen enumerated stationary sources that have the potential to emit more than 100 tons per year of any air pollutant, id. § 7479(1), ELR STAT. CAA § 169(1).

14. Id. § 7475(a)(4), ELR STAT. CAA § 165(a)(4).

15. Id. § 7475(a)(3), ELR STAT. CAA § 165(a)(3).

16. Id. § 7475(a)(5), (d), ELR STAT. CAA § 165(a)(5), (d).

17. Id. § 7479(3), ELR STAT. CAA § 169(3).

18. Id. § 7411(a)(4), ELR STAT. CAA § 111(a)(4). This definition of modify was originally established for the NSPS program, but it was explicitly incorporated by both the NSR program, id. § 7501(4), ELR STAT. CAA § 171(4), and by the PSD program, id. § 7479(C), ELR STAT. § 169(C).

19. See 57 Fed. Reg. 32314, 32316 (July 21, 1992).

20. 67 Fed. Reg. 80186 (Dec. 31, 2002).

21. 42 U.S.C. § 7411(a)(4), ELR STAT. CAA § 111(a)(4) (emphasis added).

22. 40 C.F.R. § 51.165(a)(1)(v)(C)(1); id. § 51.166(b)(2)(iii)(a); id. § 52.21(b)(2)(iii)(a); id. § 52.24(f)(5)(iii)(a); id. § 60.14(e)(1).

23. Wisconsin Elec. Power Co. (WEPCO) v. Reilly, 893 F.2d 901, 910, 20 ELR 20414, 20418 (7th Cir. 1990) (quoting memorandum from Don R. Clay, Acting Assistant Administrator for Air and Radiation, U.S. EPA, to David A. Kee, Director of Air and Radiation Division, U.S. EPA Region V (Sept. 9, 1988)).

24. See In re Tennessee Valley Auth., No. CAA-2000-04-008 at 65-69 (EAB 2000). The four factors are the same as those cited in WEPCO: (1) nature and extent; (2) purpose; (3) frequency; and (4) cost. The content of each of these four factors is further elaborated in U.S. EPA, APPLICABILITY DETERMINATION REGARDING THE PROPOSED REPLACEMENT AND RECONFIGURATION OF THE HIGH PRESSURE SECTION OF TWO STEAM TURBINES AT DETROIT EDISON'S MONROE POWER PLANT—DETROIT EDISON APPLICABILITY DETERMINATION DETAILED ANALYSIS 10 (2000), available at http://www.epa.gov/Region7/programs/artd/air/nsr/nsrpg.htm [hereinafter DETROIT EDISON APPLICABILITY DETERMINATION].

25. 67 Fed. Reg. at 80292 ("Applicability determinations are available electronically from Region 7 NSR Policy and Guidance Database (http://www.epa.gov/Region7/programs/artd/air/nsr/nsrpg.htm).").

26. 67 Fed. Reg. at 80293.

27. For some examples of EPA's use of the current RMRR standard in the context of specific applicability determinations, see, e.g., In re Tennessee Valley Auth. at 65; DETROIT EDISON APPLICABILITY DETERMINATION, supra note 24; U.S. EPA, OPINION ON WHETHER MAKING A BYPASS STACK FUNCTIONAL CONSTITUTES ROUTINE REPAIR AND MAINTENANCE (2002), available at http://www.epa.gov/Region7/programs/artd/air/nsr/nsrpg.htm; U.S. EPA, OPINION ON IF [SIC] PACKAGING CORPARATION [SIC] PROJECT IS ROUTINE MAINTENANCE (2001), available at http://www.epa.gov/Region7/programs/artd/air/nsr/nsrpg.htm. Notably, in all these cases EPA found the RMRR exception inapplicable.

28. This may be additional evidence that the scope of the exemption as currently applied is relatively narrow.

29. 893 F.2d 901, 20 ELR 20414 (7th Cir. 1990).

30. Id. at 905, 20 ELR at 20415.

31. Id. at 905-06, 20 ELR at 20416.

32. Id. at 906, 20 ELR at 20416.

33. Id. at 908, 20 ELR at 20417.

34. Id. (quoting Petitioner's Brief at 32-33).

35. Id.

36. Id. at 905, 20 ELR at 20415.

37. Id. at 906-07, 20 ELR at 20416 (citing Chevron, U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 14 ELR 20507 (1984).

38. Id. at 908, 20 ELR at 20417.

39. Id. (emphasis supplied by court).

40. Id. at 909, 20 ELR at 20417.

41. Id. at 910, 20 ELR at 20418 (citing Lyng v. Payne, 476 U.S. 926, 939 (1986)).

42. Id. at 910-13, 20 ELR at 20418-19.

43. See id. at 907-08, 20 ELR at 20416-17 ("Certainly, under the plain terms of the Act, WEPCO's replacement program constitutes a 'physical change.' … We follow Congress's definition of 'modification' … when interpreting this term within the context of the Clean Air Act.") (emphasis added).

44. See, e.g., United States v. Southern Ind. Gas & Elec. Co. (SIGECO), No. IP 01-1070-C M/S, 2003 WL 367901, at *11 (S.D. Ind. Feb. 18, 2003) (stating that the "routine maintenance exemption is found in the CAA regulations promulgated by the EPA pursuant to a Congressional grant of authority" but not explaining what congressional authority authorizes that particular exemption). See also In re Tennessee Valley Auth., No. CAA-2000-04-008, at 7, 48, 76 (EAB 2000) (like the Seventh Circuit in WEPCO, upholding EPA's power to deny application of the de minimis exception to a challenged activity, but not addressing directly EPA's legal authority to grant a de minimis exemption). There are some decisions that do uphold EPA's authority to create a routine maintenance exception against an environmentalist challenge. The most important of these, Environmental Defense Fund v. EPA, 82 F.3d 451, 26 ELR 20968 (D.C. Cir. 1996) and Ober v. Whitman, 243 F.3d 1190, 31 ELR 20550 (9th Cir. 2001), are discussed more extensively in Part III.C, infra.

45. 67 Fed. Reg. at 80186. As noted above, this rule change, while significant and obviously related to the RMRR rule, will not be discussed in this Article.

46. Id. at 80293.

47. Id.

48. Id. One might reasonably conclude that here EPA is smuggling into its critique of the form of the current RMRR exemption (a standard as opposed to a rule) an implicit criticism of the scope of that exemption under the current regime, which the Bush EPA appears to view as excessively narrow.

49. Id. EPA's rhetoric here may seem a bit confusing, or perhaps even disingenuous, since presumably a physical or operational change that reduced rather than increased emissions would not be considered a "modification" for purposes of NSR or PSD. Nonetheless, EPA's general point may still have merit, since environmental gains achieved through efficiency improvements may not get a source out of NSR requirements, especially if the efficiency gains at the target source mean that its own activity (and hence emissions) increase at the expense of some other, less-efficient source.

50. See Kathleen M. Sullivan, The Justices of Rules and Standards, 106 HARV. L. REV. 22, 62-67 (1992); Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 DUKE L.J. 557, 585-95 (1992).

51. Note that this was not the only approach available to EPA, even granting the flaws of the extremely open-ended case-by-case approach EPA had employed in the past. EPA could have adhered to a standard-based approach but articulated more clearly which factors were most important. It could have streamlined the process for applying for an administrative determination, or provided a special "fast-track" administrative determination process for proposed projects falling within certain clearly defined categories. Or, it could have promulgated the substance of its proposed regulation not in the form of a legally binding rule but rather in the form of an administrative guidance letter, stating that, in general, it did not intend to enforce the literal requirements of the Act against certain types of clearly defined modifications. See, e.g., Pacific Gas & Elec. v. Federal Power Comm'n, 506 F.2d 33, 37-40 (D.C. Cir. 1974). EPA's decision to promulgate a blanket exemption for certain categories of activities that would henceforth be considered automatically to be RMRR thus represents a distinctive policy choice that goes beyond the mere diagnosis that the old system worked poorly.

52. Activities falling outside either per se category could still be justified as RMRR under the old case-by-case approach. 67 Fed. Reg. at 80293-94. In that light, EPA's avowed preference for rules over standards is perhaps a bit one-sided.

53. Id. at 80293-96.

54. Certain activities would not be eligible for exclusion under the AMRRA approach. These include the construction or replacement of an entire "process unit" or any change that would increase "the source's maximum achievable hourly emissions rate of any regulated NSR pollutant, or in the emission of any regulated NSR pollutant not previously emitted by the stationary source." Id. at 80294-95. A "process unit" would be defined as "any collection of structures and/or equipment that processes, assembles, blends, or otherwise uses material inputs to produce or store a completed product. A single facility may contain more than one process unit." Id. at 80302. The exclusion of activities that increase the achievable hourly emissions rate appears to be another example of EPA's concern that the existing approach to the air pollution problem fails to distinguish adequately between emissions increases at a source that reflect an increased volume at that particular source—a change that might reflect greater efficiency and lower overall pollution—and emissions increases that reflect an increase in the rate at which a particular source pollutes. EPA apparently wants to make sure that low-cost changes that increase pollution rates are not considered RMRR, but changes that do not increase rates are not so considered, even if those changes do lead to emissions increases that would satisfy the second prong of the statutory definition of "modification."

55. Id. at 80294-95. EPA offers the following example to illustrate how the AMRRA system would work. Suppose a source has an AMRRA set at $ 2 million. In a given year, the source spends $ 1 million on general maintenance activities, and it also undertakes five discrete projects that cost $ 600,000, $ 350,000, $ 250,000, $ 200,000, and $ 150,000, respectively. The sum of all alleged maintenance costs is $ 2.56 million, which exceeds the AMRRA by $ 560,000. The cost of the most expensive discrete activity—the $ 600,000 project—must therefore be subtracted from the annual cost calculation. The maintenance costs and the other four projects then come to a total of $ 1.94 million, under the $ 2 million allowance, and so all those activities are automatically considered RMRR. The $ 600,000 project, however, must either obtain an NSR permit, demonstrate that it should be considered RMRR under the old case-by-case review standard, or otherwise show some exemption from the NSR requirements. Id. at 80295. The reason that EPA would require sources to start with the least expensive projects is so that sources would not be able to circumvent the process by counting their most expensive activities toward the AMRRA, and then subjecting their smaller activities to the case-by-case assessment that they are almost certain to pass. Id. In the previous example, for instance, EPA doesn't want the source to be able to count the $ 600,000 project toward the AMRRA and subject the three smaller projects ($ 150,000, $ 200,000, and $ 250,000) to NSR instead, because the $ 600,000 project is less likely to be a routine maintenance activity than any of the smaller projects.

56. Id. at 80295.

57. Id. at 80296.

58. See supra notes 29-44 and accompanying text.

59. See Alabama Power Co. v. Costle, 636 F.2d 323, 10 ELR 20001 (D.C. Cir. 1979) (recognizing EPA authority to create de minimis exemptions to provisions of the CAA); see also Monsanto v. Kennedy, 613 F.2d 947, 955 (D.C. Cir. 1979) (recognizing agency authority to authorize de miminis exemptions); Public Citizen v. Young, 831 F.2d 1108, 18 ELR 20173 (D.C. Cir. 1987) (describing the de minimis doctrine as an "escape hatch" from literalist interpretations that would lead to waste of agency resources or absurd results).

60. 67 Fed. Reg. at 80296.

61. Id.

62. Id.

63. See Wisconsin Elec. Power Co. (WEPCO) v. Reilly, 893 F.2d 901, 907-08, 20 ELR 20414, 20417 (7th Cir. 1990) (noting that Congress had defined "modification" for purposes of interpreting the CAA and finding that "under the plain terms of the Act" the appellant utility's replacement program fell within the scope of the Act).

64. 67 Fed. Reg. at 80296.

65. Id.

66. Id.

67. Chevron, 467 U.S. at 842, 14 ELR at 20509.

68. WEPCO, 893 F.2d at 905, 20 ELR at 20415.

69. August 28 Draft [EPA Docket No. OAR-2002-0068-0054] (2002) at 27-29.

70. Id. at 27.

71. 636 F.2d 323, 10 ELR 20001 (D.C. Cir. 1979).

72. August 28 Draft, supra note 69.

73. 636 F.2d at 361, 10 ELR at 20012.

74. Id. at 360-61, 10 ELR at 20012.

75. August 28 Draft, supra note 69, at 28 ("Under [the de minimis] approach, the exemption should be given a relatively narrow construction.") (citing Brief for Respondents, Tennessee Valley Auth. v. Whitman, No. 00-12310-E, Feb. 21, 2001, at 61).

76. Id. at 28-29.

77. For reasons that are perhaps self-evident, the August 28 Draft omits the word "any" before "physical change or change in the method of operation" when explicating the Chevron approach. See id.

78. Id. at 27 (emphasis added).

79. Id. at 27-28.

80. Id. at 28.

81. Id. (citing Brief for Respondents, Tennessee Valley Auth. v. Whitman, No. 00-12310-E, Feb. 21, 2001, at 61).

82. October 11 Draft [EPA Docket No. OAR-2002-0068-0059] (2002).

83. E-mail from Bill Wehrum, U.S. EPA, accompanying circulation of October 11 Draft [EPA Docket No. OAR-2002-0068-0059] (2002).

84. Id.

85. August 22 Draft, supra note 69, at 27.

86. October 11 Draft, supra note 82, at 29-30 (emphasis added).

87. Again, this formulation of the requirements for invoking the de minimis exemption appears to be inconsistent with Alabama Power itself, which holds that "the agency will bear the burden of making the required showing" and that "this exemption authority is narrow in reach and tightly bounded by the [agency's] need to show the situation is genuinely de minimis or one of administrative necessity." Alabama Power, 636 F.2d at 360-61, 10 ELR at 20012. However, the October 11 Draft may nonetheless be consistent with more current thinking. In Environmental Defense Fund v. EPA, 82 F.3d 451, 467, 26 ELR 20968, 20976 (D.C. Cir. 1996), the court noted that Alabama Power was decided before Chevron, and that therefore "the same deference due to an agency's reasonable interpretation of an ambiguous statute may also be due to an agency's creation of a de minimis exemption." However, the court explicitly declined to reach and resolve the issue of "whether, under Chevron, an agency may create a de minimis exemption with a justification less rigorous than we indicated in Alabama Power." Id.

88. Compare August 22 Draft, supra note 69, at 27 with October 11 Draft, supra note 82, at 30.

89. Compare August 22 Draft, supra note 69, at 28 with October 11 Draft, supra note 82, at 30.

90. No. 00-12310-E (11th Cir. Feb. 21, 2001).

91. October 11 Draft, supra note 82, at 30 (emphasis added).

92. October 20 Draft [EPA Docket No. OAR-2002-0068-0041] (2002).

93. Id. at 66.

94. Id. at 66-67.

95. See supra notes 76-77 and accompanying text.

96. October 11 Draft, supra note 82, at 31.

97. October 20 Draft, supra note 92, at 66.

98. See supra notes 78-81 and accompanying text.

99. See supra note 94 and accompanying text.

100. October 20 Draft, supra note 92, at 67.

101. Id. at 66.

102. See supra notes 38-39 and accompanying text.

103. E-mail from Lee Otis accompanying October 20 Draft [EPA Docket No. OAR-2002-0068-0041] (2002).

104. Supplement to October 20 Draft [EPA Docket No. OAR-2002-0068-0040] (2002), at 1.

105. October 24 Draft [EPA Docket No. OAR-2002-0068-0044] (2002), at 76-79.

106. See supra Part II.A. This version of the legal basis was apparently drafted in its final form by November 1. See E-mail from Bill Harnett, U.S. EPA, to Amy L. Farrell et al., "Legal Language for RMRR" [EPA Docket No. OAR-2002-0068-0047] (2002) (containing the legal basis language that appears in the published preamble).

107. See supra notes 62-68 and accompanying text.

108. Public Citizen v. Young, 831 F.2d 1108, 18 ELR 20173 (D.C. Cir. 1987), is an example of a case in which a court found that a statute precluded an agency from creating a de minimis exemption to a statutory command. However, the opinion implied that such a finding is exceptional; the canonical formulation of the de minimis exemption principle, the Public Citizen court noted, is "virtually a presumption in its favor." Id. at 1113, 18 ELR at 20175. Note also that, because the Public Citizen court decided that no de minimis exemption was permissible, the court did not have to reach the issue of how much deference was due the agency's determination of the particular de minimis threshold. See id. at 1112 n.4, 18 ELR at 20175 n.4 ("We do not … purport to decide the appropriate dividing point between de minimis and other risks.").

109. There is some authority for this proposition. See American Water Works Ass'n v. EPA, 40 F.3d 1266, 1271, 25 ELR 20335, 20336 (D.C. Cir. 1994) ("Where a literal reading of a statutory term would lead to absurd results, the term simply has no plain meaning … and is the proper subject of construction by EPA and the courts….") (internal quotation marks omitted) (citing Chemical Mfrs. Ass'n v. Natural Resources Defense Council, 470 U.S. 116, 15 ELR 20230 (1985)). See also Wassenaar v. Office of Personnel Management, 21 F.3d 1090, 1094 (Fed. Cir. 1994) ("Because a literal construction … results in absurd conclusion, we conclude that the plain language of the statute does not provide us with an 'unambiguously expressed intent of Congress.'") (quoting Chevron); Northrop Grumman Corp. v. United States, 47 Fed. Cl. 20, 37 (Fed. Cl. 2000) ("Step Two of Chevron is brought on … when the plain language of a statute leads to absurd consequences….").

110. See Alabama Power Co. v. Costle, 636 F.2d 323, 360 & n.89, 10 ELR 20001, 20011 & n.89 (D.C. Cir. 1979), which states:

The ability … to exempt de minimis situations from a statutory command is not an ability to depart from the statute, but rather a tool to be used in implementing the statutory design…. In this respect, the principle is a cousin of the doctrine that, notwithstanding the "plain meaning" of a statute, a court must look beyond the words to the purpose of the act where its literal terms lead to "absurd or futile results."

See also Public Citizen, 831 F.2d at 1112, 18 ELR at 20175.

111. Many scholars, influenced by modern work in social choice theory, have argued that the concept of "legislative intent," when applied to a collective decisionmaking body like Congress, is incoherent and useless as a guide for statutory construction. See, e.g., Frank H. Easterbrook, Statutes' Domains, 50 U. CHI. L. REV. 533, 547-48 (1983); Kenneth A. Shepsle, Congress Is a "They," Not an "It": Legislative Intent as Oxymoron, 12 INT'L REV. L. & ECON. 239 (1992). This general critique actually predates modern social choice theory by several decades. See Max Radin, Statutory Interpretation, 43 HARV. L. REV. 863, 871-72 (1930). Yet, for better or worse, the rhetoric of legislative intent persists despite its conceptual flaws.

112. See, e.g., Holy Trinity Church v. United States, 143 U.S. 457, 459 (1897):

It is a familiar rule that a thing may be within the letter of the statute and yet not within the statute, because not within its spirit nor within the intention of its makers…. The absurd results which follow from giving such broad meaning to the words, makes it unreasonable to believe that the legislator intended to include the particular act.

113. Chevron, 467 U.S. at 843 n.9, 8 ELR at 20508-09 n.9.

114. See Alabama Power, 636 F.2d at 360-61, 10 ELR at 20011; Public Citizen, 831 F.2d at 1113, 18 ELR at 20175.

115. United States v. Locke, 471 U.S. 84, 95 (1985) (refusing to depart from the literal text of the statute even when that text leads to a "harsh result") (quoting Mobil Oil Corp. v. Higginbotham, 436 U.S. 618, 625 (1978)). See also Brock v. Writers Guild of Am. W., 762 F.2d 1349, 1353 (1985) (rejecting Secretary of Labor's interpretation of the Labor-Management Reporting and Disclosure Act as extending the statute "beyond the point where Congress … would stop," stressing that there is a difference between "filling a gap left by Congress' silence" and rewriting the statute) (quoting 62 Cases of Jam v. United States, 340 U.S. 593, 600 (1951) and Higginbotham, 436 U.S. at 625).

116. For this reason, some committed textualists argue for abandoning the canon in its entirety. See John F. Manning, The Absurdity Doctrine, 116 HARV. L. REV. 2387. Most textualists, however, continue to recognize absurdity as an exception to the otherwise universal rule that the text controls, albeit an exception that ought to be construed as narrowly as possible. Id. at 2419-21.

117. 140 F.3d 1060 (D.C. Cir. 1998).

118. Id. at 1068.

119. See Cass R. Sunstein, Avoiding Absurdity? A New Canon in Regulatory Law, 32 ELR 11126 (Sept. 2002) (arguing that "courts should allow agencies to avoid absurdity, in the face of literal text, even in cases in which courts should not themselves exercise this power," but making clear that, as a matter of Chevron analysis, this issue is a matter of Chevron Step One).

120. American Water Works Ass'n v. EPA, 40 F.3d 1266, 1271, 25 ELR 20335, 20336 (D.C. Cir. 1994) (emphasis added).

121. In addition to these theoretical considerations, there is a good practical argument for adopting this understanding. Congress may want to give agencies the power to avoid truly absurd results, but reasonably fear that letting agencies carve out exemptions, subject only to very deferential judicial review, could give agencies too much power to curtail the scope of a deliberately broad statute. If the only way for Congress to constrain the agency's ability to narrow a statute's scope is to write a statute so rigid that no de minimis exemptions of any kind are allowed, Congress faces an unwelcome choice between giving agencies too little flexibility and giving them too much. If a statute can admit de minimis exemptions but still subject them to a higher standard of review, the problem is mitigated.

122. Mova, 140 F.3d at 1068.

123. See infra notes 160-70 and accompanying text.

124. See Sunstein, supra note 119.

125. See infra Part III.C.

126. Chevron, 467 U.S. at 842, 14 ELR at 20508. See also Microwave Communications Inc. (MCI) v. American Tel. & Tel. Co. (AT&T), 512 U.S. 218, 229 (1994) ("An agency's interpretation of a statute is not entitled to deference when it goes beyond the meaning that the statute will bear."); Atlantic City Elec. Co. v. Federal Energy Regulatory Comm'n, 295 F.3d 1, 11 (D.C. Cir. 2002) (holding that an agency "cannot rely on one of its own regulations to trump the plain meaning of a statute"); Hennepin County Med. Ctr. v. Shalala, 81 F.3d 743, 748 (8th Cir. 1996) ("The plain meaning of a statute controls, if there is one, regardless of an agency's interpretation.").

127. 520 U.S. 1, 5 (1997).

128. 299 F.3d 949 (8th Cir. 2002).

129. Id. at 954 (collecting sources); see also Merritt v. Dillard Paper Co., 120 F.3d 1181, 1186 (11th Cir. 1997) ("The adjective 'any' is not ambiguous; it has a well established meaning … 'any' means all."); G.M. Trading Corp. v. Commissioner of Internal Revenue, 121 F.3d 977, 981 (5th Cir. 1997) (relying on the "plain meaning" and "broad scope" of the word "any" in a statute). But see Bell Atl. Tel. Cos. v. Federal Communications Comm'n, 131 F.3d 1044, 1047 (D.C. Cir. 1997) ("The Supreme Court has specifically held that in context the word 'any' many be construed in a non-expansive fashion.") (citing O'Connor v. United States, 479 U.S. 27, 31 (1986)). However, the O'Connor opinion stressed that an expansive reading of "any" in a particular passage in the Panama Canal Treaty would require either "verbal distortions … to give plausible content" to other sections of the treaty, O'Connor, 479 U.S. at 33, or would render the entire treaty provision in question "utterly implausible." Id. at 31.

130. 512 U.S. 218 (1994). But see Environmental Defense Fund v. EPA, 82 F.3d at 466, 26 ELR at 20975 (upholding a de minimis exemption to a portion of the CAA that applied to "any [federal] activity" that does not conform to an implementation plan, explaining that

although the terms of the statute do prohibit the federal government from engaging in "any activity" that is not in conformity, it seems eminently reasonable for the EPA to interpret this provision to refer to "any activity" that is likely to interfere with the attainment goals in [an implementation plan].

131. MCI, 512 U.S. at 228.

132. See supra note 34 and accompanying text.

133. MCI, 512 U.S. at 228.

134. But see Chemical Mfrs. Ass'n v. Natural Resources Defense Council, 470 U.S. 116, 118, 15 ELR 20230, 20233 (1985) (finding that "the word 'modify' … has no plain meaning as used in § 301(l) [of the CWA], and is the proper subject of construction by EPA and the courts"). However, not only was Chemical Manufacturers decided almost a decade before MCI, but in the former case a broad, literal reading of "modify" would involve a direct conflict with another specific command in the statutory text. See id. ("Reading ['modify' as used in § 301(l) broadly] would forbid what § 307(b)(2) expressly directs….").

135. Here, a good analogy might be the Sherman Act, 15 U.S.C. § 1, which on its face forbids "every contract … in restraint of trade." The Court, has sensibly recognized that, because a "literal reading of the statute would outlaw the entire body of private contract law," Business Elecs. Corp. v. Sharp Elecs. Corp., 485 U.S. 717, 736 (1988) (Stevens, J., dissenting), "a completely literal interpretation of the word 'every' cannot have been intended by Congress." City of Columbia v. Omni Outdoor Adver., 399 U.S. 365, 385 (1991) (Stevens, J., dissenting). Thus, instead of interpreting the statute literally, the Court interprets the Sherman Act as forbidding only "unreasonable" restraints of trade. See Standard Oil Co. v. United States, 221 U.S. 1, 60, 87 (1911).

136. Cf. Public Citizen v. Young, 831 F.2d 1108, 1112, 18 ELR 20173, 20175 (D.C. Cir. 1987).

137. Whether the political affiliation of the judges on the D.C. Circuit affects their disposition of environmental cases is a controversial question, and one on which, for purposes of this Article, I take no position. For a sample of the scholarly debate over this issue, see Christopher H. Schroeder & Robert L. Glicksman, Chevron, State Farm, and EPA in the Courts of Appeals During the 1990s, 31 ELR 10271 (Apr. 2001); Richard L. Revesz, Environmental Regulation, Ideology, and the D.C. Circuit, 83 VA. L. REV. 1717 (1997); Harry T. Edwards, Collegiality and Decisionmaking on the D.C. Circuit, 84 VA. L. REV. 1335 (1998); Richard L. Revesz, Ideology, Collegiality, and the D.C. Circuit: A Reply to Chief Judge Harry T. Edwards, 85 VA. L. REV. 805 (1999).

138. WEPCO, 893 F.3d at 909, 20 ELR at 20417.

139. See supra Part II.C.

140. See supra note 121.

141. See, e.g., Cass R. Sunstein, Law and Administrative Action After Chevron, 90 COLUM. L. REV. 2071 (1990); William N. Eskridge Jr. & John Ferejohn, The Article I, Section 7 Game, 80 GEO. L.J. 523, 547-51 (1992); Cynthia R. Farina, Statutory Interpretation and the Balance of Power in the Administrative State, 80 COLUM. L. REV. 452 (1989). Of course, Chevron also entails a more direct shift of power from courts to administrative agencies, but an indirect result of this shift, as these and other works demonstrate, is to increase the potential deviation of agency policy choices from congressionally preferred policy outcomes.

142. Whitman v. American Trucking Ass'n, 531 U.S. 457, 485, 31 ELR 20512, 20517 (2001).

143. Cf. Farina, supra note 141.

144. See, e.g., Ober v. Whitman, 243 F.3d 1190, 1195, 31 ELR 20550, 20551 (9th Cir. 2001); Environmental Defense Fund v. EPA, 82 F.3d 451, 465-67, 26 ELR 20968, 20975-76 (D.C. Cir. 1996); Ohio v. EPA, 997 F.2d 1520, 23 ELR 21157 (D.C. Cir. 1993). But see Environmental Defense Fund v. EPA, 636 F.2d 1267, 10 ELR 20972 (D.C. Cir. 1980) (denying a de minimis exemption from the Toxic Substances Control Act for polychlorinated biphenyls concentrations under 50 parts per million (ppm) where evidence demonstrated that exposure to concentrations below 50 ppm could have adverse environmental and health effects); Natural Resources Defense Council v. EPA, 966 F.2d 1292, 22 ELR 20950 (9th Cir. 1992) (invalidating as arbitrary and capricious a de minimis exemption for sites of less than five acres from the CWA where EPA admitted "that even small construction sites can have a significant impact on local water quality"). It is not clear whether these opinions object to any overinclusive de minimis exemption, or whether they stand only for the proposition that in those specific instances, EPA's rule was too overinclusive. The latter position would be entirely consistent with the approach urged here; the former, which implicitly demands case-by-case evaluations, would not be.

145. See Alabama Power Co. v. Costle, 636 F.2d 323, 361, 10 ELR 20001, 20012 (D.C. Cir. 1979) (noting that, if EPA sought to justify its "50-ton exemption" on de minimis grounds it would have to "show that the situation is genuinely de minimis" but not questioning the authority of the agency to promulgate a general rule); Environmental Defense Fund, 82 F.3d at 466-67, 26 ELR at 20975 ("We find nothing in the [CAA] to preclude the EPA's identification of categories of federal action that would produce either no or a trivial level of emissions….") (emphasis added). But see also Natural Resources Defense Council v. Costle, 568 F.2d 1369, 8 ELR 20028 (D.C. Cir. 1977) (invalidating a blanket exemption of certain categories point sources from the permitting requirements of the CWA, and requiring instead that EPA address pollution at these sources through the permitting process).

146. See, e.g., Natural Resources Defense Council, 966 F.2d at 1292, 22 ELR at 20950; Environmental Defense Fund, 636 F.2d at 1267, 10 ELR at 20972; Costle, 568 F.2d at 1377, 8 ELR at 20031.

147. Cf. City of Bourne v. Flores, 521 U.S. 507, 519-20 (1997).

148. Mova, 140 F.3d at 1071.

149. Id. at 1074.

150. Id. at 1069.

151. 243 F.3d 1190, 1195, 31 ELR 20550, 20551 (9th Cir. 2001).

152. Cass Sunstein, for example, argues: "Under [Chevron] step one, the relevant administration should … be authorized, most of the time, to accept or to reject literalism when that approach would lead to an absurd or patently unreasonable outcome." Sunstein, supra note 119, at 11131. But that formulation begs the standard-of-review question. How much of the time is "most of the time"? What sort of showing must the agency make to establish that failure to grant a particular exemption from the scope of the statute would be "absurd"—and how closely will the courts scrutinize this agency's explanation? Though Sunstein's discussion indicates that he favors a relatively deferential standard of review, he does not articulate any particular test.

153. Alabama Power, 636 F.2d at 360, 10 ELR at 20011.

154. Id.

155. Id. at 361, 10 ELR at 20012.

156. The degree to which Chevron actually effected a dramatic shift in the level of judicial deference to agency interpretations of law is a matter of some dispute. See Peter Schuck & E. Donald Elliott, To the Chevron Station: An Empirical Study of Federal Administrative Law, 1990 DUKE L.J. 984; Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 YALE L.J. 969 (1992); Orin S. Kerr, Shedding Light on Chevron: An Empirical Study of the Chevron Doctrine in the U.S. Courts of Appeals, 15 YALE J. ON REG. 1 (1998); Linda R. Cohen & Matthew L. Spitzer, Solving the Chevron Puzzle, 57 LAW & CONTEMP. PROBS. 65 (1994), Aaron P. Avila, Application of the Chevron Doctrine in the D.C. Circuit, 8 N.Y.U. ENVTL. L.J. 398 (2000).

157. Unless, of course, the court was aggressive in imputing a more specific meaning of "modify," as in the MCI decision. See supra notes 130-33 and accompanying text.

158. See supra notes 126-36 and accompanying text.

159. See supra note 110.

160. 82 F.3d 451, 26 ELR 20968 (D.C. Cir. 1996).

161. Id. at 467, 26 ELR at 20976.

162. Id.

163. 42 U.S.C. § 7506(c)(1), ELR STAT. CAA § 176(c)(1).

164. 82 F.3d at 466, 26 ELR at 20975.

165. Mova, 140 F.3d at 1068. Though the Mova court included Environmental Defense Fund in a string citation for the proposition that the intention of the drafters can sometimes trump the literal language of the statute, id., the court did not note the tension between Environmental Defense Fund's discussion of the standard of review for de minimis exemptions and its own application of more rigorous scrutiny to the FDA's invocation of the absurdity canon. Interestingly, Judge Laurence H. Silberman was on both the Mova and Environmental Defense Fund panels; apparently he perceived no tension between the decisions.

166. 966 F.2d 1292, 1306, 22 ELR 20950, 20957 (9th Cir. 1992).

167. Id. (citing Alabama Power, 636 F.2d at 361, 10 ELR at 20012) (internal quotation marks omitted).

168. 243 F.3d at 1190, 31 ELR at 20550.

169. Id. at 1195, 31 ELR at 20551 (citing Environmental Defense Fund, 82 F.3d at 467, 26 ELR at 20975).

170. In Ober, the statutory question that the court had to decide was whether EPA could create a de minimis exemption in a federal implementation plan where the statute specified that implementation plans had to include provisions for implementation of "reasonably" available control technology, and also specified that plans that would not bring an area into compliance with the NAAQS by a statutory deadline had to show that such attainment would be "impracticable." Ober, 243 F.3d at 1194-95, 31 ELR at 20550-51 (citing 42 U.S.C. §§ 7513a(a)(1)(B)-(C), ELR STAT. CAA §§ 189(a)(1)(B)-(C)). The question in Ober thus involved interpretation of the open-ended phrases "reasonable" and "impracticable" to determine whether creation of the de minimis exemption was permissible. The Ninth Circuit's application of the Chevron standard is therefore entirely consistent with the first Chevron de minimis theory, but says little or nothing about the second.

171. Chevron, 467 U.S. at 837, 14 ELR at 20508.

172. 323 U.S. at 139.

173. 529 U.S. 576, 587 (2000).

174. 533 U.S. 215, 227-28 (2001).

175. Skidmore, 323 U.S. at 139. The revival of Skidmore in the Christensen and Mead cases is controversial. Justice Scalia, for example, characterized this revival as "an avulsive change in judicial review of federal administrative action," Mead, 533 U.S. at 239 (Scalia, J., dissenting) and dismissed Skidmore as an "anachronism." Id. at 250.

176. Mead, 533 U.S. at 226-27.

177. For representative examples of the competing scholarly perspectives on this question, see Thomas W. Merrill & Kathryn Tongue Watts, Agency Rules With the Force of Law: The Original Convention, 166 HARV. L. REV. 467 (2002); David J. Baron & Elena Kagan, Chevron's Nondelegation Doctrine, 2001 SUP. CT. REV. 201.

178. Chirstensen, 529 U.S. at 587; Mead, 533 U.S. at 229-31.

179. Mead, 533 U.S. at 231.

180. See Ohio v. EPA, 997 F.2d 1520, 1535, 23 ELR 21157, 21164 (D.C. Cir. 1993) ("The literal meaning of a statute need not be followed where the precise terms lead to absurd or futile results, or where failure to allow a de minimis exception is contrary to the primary legislative goal…. [To require a different result] is to adjudge Congress incompetent to fashion a rational legislative design."); see also Alabama Power Co. v. Costle, 636 F.2d 323, 360, 10 ELR 20001, 20011 (D.C. Cir. 1979); Environmental Defense Fund v. EPA, 82 F.3d 451, 466, 26 ELR 20968, 20975 (D.C. Cir. 1996).

181. Natural Resources Defense Council v. Costle, 568 F.2d 1369, 1377, 8 ELR 20028, 20031 (D.C. Cir. 1977).

182. Thus, although I generally agree with the Mavo court's analysis of the interaction between Chevron and the absurdity canon, see supra notes 117 & 165 and accompanying text, I do not accept the implication that the agency's interpretation of the proper scope of a de minimis exemption gets no special consideration whatsoever. See also Sunstein, supra note 119, at 11131 (stressing agency accountability and expertise as reasons for allowing agencies a freer hand in applying the absurdity canon to regulatory statutes than would be appropriate for courts).

183. Sunstein, supra note 119, at 11132, considers the possibility that "agencies will seek to exempt risks not because those risks are genuinely trivial, but because of political pressures, imposed by regulated industries with self-serving agendas," but he is optimistic that "the underlying risks are minimal" because "the usual pattern is one in which sensible regulators are attempting to counteract excessive generality or palpable unreasonableness, neither intended by Congress." I am less sanguine about the risks posed by agency capture, or by well-intentioned but mistaken deviation from a congressional command, which is why I endorse a more rigorous standard of review than Sunstein implies. Given that the de minimis exemption is only supposed to apply in cases where application of the statute would produce results that are patently absurd, Skidmore ought to be an effective filter. If the proposed exemption really does involve avoidance of an absurd result, making the requisite showing to convince a court of the propriety of the exemption ought to be relatively painless for the agency. At the same time, Skidmore is more likely than Chevron to deter agencies from pushing the boundaries or trying to stretch the exemption beyond its intended purpose. Whether I am correct in this assessment is obviously an empirical question that is difficult to assess ex ante.