The Implications of Michigan v. EPA for Regulation of Hazardous Air Pollutants and Beyond

by Robert Glicksman

The following post is based on an article by Professor Glicksman on the George Washington Law Review website.[1]

In Michigan v. Environmental Protection Agency,[2] Justice Scalia, for a 5-4 majority, held that the Environmental Protection Agency (EPA)’s failure to consider cost at the initial stage of deciding whether to regulate emissions of hazardous air pollutants from electric generating units (EGUs or power plants) under § 112 of the Clean Air Act (CAA), even though it gave ample consideration to cost at multiple subsequent stages of the rulemaking, was unreasonable. The provision that EPA improperly interpreted is narrow in scope, applying only to EGUs. The decision remanding the case to the D.C. Circuit is unlikely to significantly affect EPA’s effort to regulate EGUs under § 112, unless the delay in the onset of regulation on remand stretches into a presidential administration that views this regulatory initiative less favorably than the Obama Administration’s EPA. Nevertheless, the case establishes principles and raises important questions of environmental and administrative law that are likely to arise in a host of future regulatory contexts.

In the face of EPA’s languid regulation of hazardous air pollutants, Congress in 1990 required EPA to regulate emissions of more than 180 hazardous air pollutants listed in the statute itself. It also required EPA, in § 112(n)(1) of the CAA, to perform a study of health hazards reasonably anticipated to occur as a result of EGU emissions of any of the listed pollutants, such as mercury. Section 112(n)(1) provides that EPA “shall regulate [EGUs] . . . if [it] finds such regulation is appropriate and necessary after considering the results of the study . . . .”[3] After conducting the study, EPA concluded that regulation of hazardous air pollutants from EGUs was “appropriate and necessary” and issued regulations to control emissions of those pollutants. Several states and industry and labor groups challenged the regulations in the D.C. Circuit, which upheld them by a 2-1 vote.

The issue before the Supreme Court was whether EPA violated § 112(n)(1) by determining that regulation of EGUs was “appropriate and necessary” without considering its cost. The majority and dissent differed sharply in their reading of the portions of the record bearing on the costs and benefits of regulating EGUs. According to the majority, the regulations would force power plants to incur costs of $9.6 billion per year, while generating quantifiable health and environmental benefits for mercury reduction alone were $4 to $6 million per year.[4] 

Thus, the direct costs were 1,600 to 2,400 times as great as the quantifiable benefits. Justice Kagan, joined by three other dissenting justices, concluded that the record showed that quantifiable benefits would exceed costs by as much as $80 billion each year. The discrepancy arose because the majority opinion omitted from its calculation the “co-benefits” (also referred to in the majority opinion as “ancillary benefits”) the regulation would provide when the technology installed to reduce hazardous air pollutants also reduced other harmful pollutants (primarily particulate matter), even though they are not covered by the rule. EPA’s regulatory preamble featured those co-benefits, which included the elimination of thousands of premature deaths that would result from emissions of those other pollutants in the absence of regulation.

Because the issue here turned on EPA’s interpretation of § 112(n)(1), which it construed as not requiring it to consider cost when deciding whether to regulate EGUs, the majority analyzed the case under the judicial review framework provided by Chevron.[5] As is his wont, Justice Scalia moved directly to Chevron’s second step, which requires a reviewing court to accept an agency’s reasonable interpretation of an ambiguous statute. The majority concluded that EPA “strayed far beyond” the bounds of reasonable interpretation by construing § 112(n)(1) “to mean that it could ignore cost when deciding whether to regulate power plants.”[6] As the Court has done in other cases, the majority equated Chevron step two with arbitrary and capricious review under § 706(2)(A) of the APA, citing the State Farm decision[7] as grounds for reversing decisions that do not rest on consideration of factors relevant under the agency’s organic statute.

According to Justice Scalia, EPA erred by “refus[ing] to consider whether the costs of its decision outweighed the benefits. The Agency gave cost no thought at all, because it considered cost irrelevant to its initial decision to regulate.”[8] The reference to the “initial” decision is significant because it signaled the majority’s exclusive focus on the threshold decision of whether to regulate EGUs at all under § 112. Anything EPA did at subsequent stages of the regulatory process was irrelevant because the question before the Court was whether EPA properly interpreted the “appropriate and necessary” language of § 112(n)(1) in deciding to regulate EGUs. The majority reasoned that, “[r]ead naturally,” that “capacious” phrase “requires at least some attention to cost.”[9] Further, by stating that “[n]o regulation is ‘appropriate’ if it does significantly more harm than good,”[10] the majority required EPA to compare costs and benefits. It recognized that “[t]here are undoubtedly settings in which the phrase ‘appropriate and necessary’ does not encompass cost,” but concluded that “this is not one of them.”[11] Given what the majority called the “established administrative practice” of treating cost as a centrally relevant factor and considering both the advantages and disadvantages of agency regulatory decisions, it was unreasonable for EPA to construe § 112(n)(1) as an invitation to ignore cost.

The dissent’s main point was that, after deciding whether to regulate, EPA had considered cost at multiple stages of the process of deciding how to regulate, including when it divided EGUs into categories, each of which would be regulated in different ways and to a different extent; when it adopted the regulatory standards for each category; and when it performed the cost-benefit analysis required by Executive Order 12866.[12] The only stage of the multi-step regulatory process at which EPA did not consider cost was the initial determination of whether to regulate at all. According to Justice Kagan, had the majority considered the threshold determination of whether to regulate in the context of this larger series of regulatory determinations, instead of engaging in its “peculiarly blinkered” analytical process, it could not have concluded that EPA ignored the cost of regulation.[13]

The majority’s analysis is not only blinkered, but nonsensical. As Justice Kagan noted, “EPA could not have measured costs at the process’s initial stage with any accuracy” because it did not yet know how it was going to divide the industry into categories, what pollution control technologies were already being used in those categories that could act as regulatory benchmarks, or what levels of control the final standards would reflect. “Simply put,” she stated, “calculating costs before starting to write a regulation would put the cart before the horse.”[14] Responding to Justice Scalia’s comparison of EPA’s approach to a person who decides to buy a Ferrari without inquiring into its cost, Justice Kagan concluded that EPA acted as a sensible car owner would in deciding without checking prices to replace her car’s worn out brake pads, knowing from prior experience that she could comparison shop to ensure that the purchase was within her budget.

Justice Scalia’s opinion masks the incongruity of calculating the costs of a regulation that hasn’t yet been written. He cited estimates of the rule’s costs and benefits provided in EPA’s regulatory impact analysis, presumably to show that consideration of cost was possible. But EPA prepared that analysis years after it made the threshold determination under § 112(n)(1), and only after it had already divided the industry into categories and set numerical emission caps for each category. The majority chastised EPA for ignoring an important aspect of the problem, which is one of the grounds for finding arbitrary and capricious decisionmaking under State Farm. But had EPA relied on a cost calculus it prepared at the threshold stage in deciding that regulation of EGUs was “appropriate and necessary,” it would have run the risk of reversal based on another State Farm factor – offering an explanation that runs counter to the (scant) evidence before the agency. The majority’s analysis puts EPA between a rock and a hard place – ignore cost and be reversed, or rely on speculative figures to surmise the cost of a non-existent rule and be reversed.

How important is Michigan to the fate of regulation of EGU hazardous air pollutant emissions? The Court remanded the case to the D.C. Circuit, so the answer turns partly on whether that court decides to vacate the EGU rule or to remand without vacating while EPA complies with the Court’s directive to consider cost. When EPA gets the case on remand from the D.C. Circuit, it presumably will rely on the cost-benefit analysis it has already prepared on the rule, perhaps with supplementary analysis, in deciding whether to regulate. But the process will take time, and could extend into the next presidential administration, whose environmental values and policies are, of course, unknown. In the meantime, a majority of EGU units affected by EPA’s rule have already installed the controls needed to comply, switched to fuels that generate fewer relevant emissions, or shut down due to anticipated compliance costs.[15] Few utility managers are likely to reverse those decisions, regardless of the ultimate fate of the EGU rule.

More broadly, the Michigan decision establishes some clear ground rules. The dissent acknowledged that “[c]ost is almost always a relevant – and usually, a highly important – factor in regulation. Unless Congress provides otherwise, an agency acts unreasonably in establishing a standard-setting process that ignore[s] economic considerations.”[16] All nine justices therefore agree that, in the absence of clear direction to ignore cost in regulating, agencies must consider it, creating a presumption that cost is relevant to regulatory determinations. That is no small point, because many regulatory statutes, including a host of programs administered by EPA, omit cost as a relevant consideration. It is not clear what kind of statutory language would suffice to rebut the presumption the Court reads into the law. The majority distinguished but did not overrule another CAA case, Whitman,[17] which held that EPA is precluded from considering cost when adopting national ambient air quality standards that are “requisite to protect the public health,” allowing an adequate margin of safety.[18] That provision does not on its face preclude consideration of cost, but the Court reasoned that it specifies the exclusive, relevant considerations (public health protection), and those do not include cost. How far does that precedent now extend? What other language, short of an explicit prohibition on cost consideration, would invoke Whitman instead of Michigan? We don’t know the answers yet, but the Michigan majority described Whitman as setting “the modest principle that where the [CAA] explicitly regulates on the basis of a factor that on its face does not include cost, the Act normally should not be read as implicitly allowing the Agency to consider cost anyway.”[19]

The case leaves open other administrative and environmental law questions. It remains to be seen whether the Court will move even further than the majority opinion does to merge Chevron step one and two analysis, and to merge step two analysis with decidedly non-deferential arbitrary and capricious review. As some scholars have suggested, judicial review of agency action often seems to be conducted under a reasonableness standard, notwithstanding nominally different tests for review of statutory interpretation, policy formulation, and statutory implementation.

Another question is how much discretion agencies like EPA will have to fashion their cost consideration processes. In noting that regulation would not be “appropriate” if it does significantly more harm than good, the majority seems to require, at least as a practical matter, some comparison of cost and benefit levels. Justice Scalia stated, however, that the Court did not have to and was not deciding “that the [CAA] unambiguously required” EPA, in making its threshold determination of whether to regulate, “to conduct a formal cost-benefit analysis in which each advantage is assigned a monetary value. It will be up to the Agency to decide (as always, within the limits of reasonable interpretation) how to account for cost.”[20] Justice Kagan agreed, though she, too, seemed to endorse cost-benefit analysis, at least for the limited purpose of ensuring that regulations do not impose massive costs far in excess of regulatory benefits. The majority and dissent clearly parted ways in the degree of discretion agencies have in deciding when to consider cost. Future litigation will surely feature battles over whether Michigan requires or merely suggests cost-benefit analysis, and if the former, what forms will be regarded as within or outside the bounds of agency discretion.

Another issue relates to the proper treatment of regulatory co-benefits of the kind described above. The majority refused to decide whether EPA may consider co-benefits (what the majority terms “ancillary” benefits) in deciding whether regulation is “appropriate and necessary” under § 112(n)(1). The answer is important, as reflected in the markedly different cost-benefit ratios enumerated by the majority and the dissent. The majority’s analysis is skewed, positing that “‘cost’ includes more than the expense of complying with regulations,” and that “any disadvantage could be termed a cost.”[21] At the same time, the benefits the majority cited in its calculations relate exclusively to mercury reductions, excluding ancillary benefits when concluding that the costs of regulation would be 1,600 to 2,400 times as great as the benefits. Justice Kagan’s approach is more symmetrical. Because the two opinions agreed that the discretion to fashion cost calculation methodology rests with the agency, Michigan does not compel the majority’s slanted approach.

Justice Scalia’s statement that “any disadvantage could be termed a cost” raises another issue. One of the problems with EPA’s interpretation, according to the majority, is that it precluded EPA “from considering any type of cost – including, for instance, harms that regulation might do to human health.”[22] Industry has for years argued that agencies should be required to consider the adverse effects as well as the benefits of regulation. This argument is sometimes referred to as “risk-risk” analysis (regulation creates inadvertent risks that may offset intended regulatory benefits) or “richer is safer” analysis.[23] For the most part, the courts have not been receptive to the theory. The majority seemingly endorsed it here, but it is not clear which version it endorsed. One version would require agencies to consider the risk that regulation would force industry to replace banned products or activities with ones that create even greater risks than those targeted by regulation. This version is not terribly controversial. Another version, which many oppose because it is too speculative, would require agencies to consider whether regulation would reduce income (because, for example, regulated sources would lay off workers to offset the increased costs of regulation) and then calculate how such reductions would affect health (for example, if a laid-off worker spent less money on preventive health care or replaced healthy with less expensive junk food). The majority arguably referred to the first version in discussing the possibility “that the technologies needed to eliminate . . . emissions do even more damage to human health.”[24] Absent further elaboration, however, it is not clear what version, if any, of risk-risk analysis agencies will have to incorporate into their cost analyses. The more expansive, richer-is-safer version has the potential to yield cost-benefit ratios less favorable to decisions to regulate or that push toward less rigorous forms of regulation.

The Michigan decision involves interpretation of a narrowly applicable provision that Justice Scalia emphasized is sui generis. The decision nevertheless has much to say (and leaves much unresolved) about significant environmental and administrative law issues.


[1] Robert L. Glicksman, Response, Michigan v. Environmental Protection Agency, Geo. Wash. L. Rev. Docket (July 2, 2015),

[2] Michigan v. Environmental Protection Agency, ___ S. Ct. ___, 2015 WL 2473453 (2015).

[3] 42 U.S.C. § 7412(n)(1). An important reason for requiring this study was to allow EPA to determine whether other CAA regulatory programs, such as the one restricting power plant emissions of acid rain precursors, precluded the need for regulation under § 112.

[4] The majority acknowledged that EPA identified and relied on benefits that could not be quantified.

[5] Chevron USA Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837 (1984). The dissent also cited Chevron for the proposition that courts should review with caution agency interpretations based on experience and expertise. Justice Thomas wrote a solo concurrence attacking Chevron as a separation of powers violation by allowing agencies, through policymaking interpretations of ambiguous statutes, to appropriate legislative power vested in Congress under Article I. He also charged that deference under step two might abdicate the judicial duty to enforce the rule of law.

[6] Michigan, slip op., at 6.

[7] Motor Vehicle Mfrs. Ass’n of United States, Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29 (1983).

[8] Michigan, slip op., at 5.

[9] Id. at 6-7.

[10] Id. at 7.

[11] Id.

[12] Exec. Order No. 12866, 58 Fed. Reg. 51,735 (1993).

[13] Michigan, slip op., at 2 (Kagan, J., dissenting).

[14] Id. at 20.

[15] See Emily Holden et al., Mercury regs ruling emboldens Clean Power Plan critics, but few changes seen for utilities, Climate Wire, June 30, 2015. As many as 70 percent of affected sources may already be in compliance. See Jeremy Jacobs, Justices topple EPA mercury regs in ruling seen as limited, Greenwire, June 29, 2015.

[16] Michigan, slip op., at 6 (Kagan, J., dissenting).

[17] Whitman v. Am. Trucking Ass’ns, 531 U.S. 457 (2001).

[18] 42 U.S.C. § 7409(b)(1).

[19] Michigan, slip op., at 10.

[20] Id. at 14.

[21] Id. at 7.

[22] Id.

[23] Robert L. Glicksman et al., Environmental Protection: Law and Policy 797-98 (7th ed. 2015).

[24] Michigan, slip op., at 7.

© 2016 The Center for Progressive Reform