In Michigan v. EPA (2015), one of Justice Scalia’s last major administrative law opinions, the Court said that the Environmental Protection Agency must consider costs, including compliance costs, when deciding whether it is “appropriate and necessary” to regulate mercury and other hazardous emissions from power plants, under the Clean Air Act. I’ve been seeing suggestions here and there that Michigan v. EPA should be read very broadly indeed — perhaps even as a kind of Lochner for administrative law. The suggestion seems to be that Michigan v. EPA requires agencies to consider costs, to compare costs and benefits, and even to use quantified cost-benefit analysis, whenever statutes use general phrases like “appropriate and necessary.”

Whatever the policy merits of cost-benefit analysis as a regulatory tool, this is a legally indefensible reading of the opinion, both taken on its own terms, and in the larger context of administrative law. The opinion itself merely required the agency to consider the disadvantages as well as the advantages of its decision, thereby precluding one-sided decisionmaking; and the opinion explicitly disavowed any requirement that costs and benefits must be quantified. More broadly, the Court’s caselaw, and background principles of the judicial role in administrative law, do not allow courts to read statutes to impose quantified cost-benefit analysis on agencies, unless Congress has clearly so mandated. To do so would entirely circumvent the logic of Vermont Yankee v. NRDC, which the Court emphatically reaffirmed in the very same term it decided Michigan v. EPA. Doubtless for that very reason, the Michigan v. EPA opinion did not say anything like the proponents of cost-benefit analysis attribute to it.

The doctrinal posture of Michigan v. EPA has confused some commentators, who variously describe it as a Chevron case or as a case of arbitrariness review under Section 706 of the Administrative Procedure Act. In fact it is both, because of the prong of arbitrariness review that asks whether agencies have “considered the relevant factors.” Those factors are of course factors made relevant by statute, and the Court has consistently said that Chevron deference extends to agency interpretation of statutes to ascertain which factors are relevant. So every “relevant factors” case is potentially also a Chevron case, and Michigan v. EPA follows this pattern.

The critical question, as the Court framed it, was whether the agency had engaged in reasoned decisionmaking by interpreting “appropriate and necessary” not to include consideration of cost as a relevant factor. (In fact, as Justice Kagan’s dissent pointed out, the agency only refused to consider cost at the initial stage of deciding whether to regulate. But this complication is tangential to the points I make here, and I will ignore it). The Court said that the agency’s refusal to consider costs was clearly mistaken; the “appropriate and necessary” language clearly made compliance cost a relevant consideration. As we will see, however, there is a world of difference between that holding and the broad reading, on which the Court was interpolating a mandate for quantified cost-benefit analysis into general, vague or ambiguous statutory language.

First: The broad reading seems to rest mainly on one salient passage, as follows:

[T]he phrase “appropriate and necessary” requires at least some attention to cost. One would not say that it is even rational, never mind ‘appropriate,’ to impose billions of dollars in economic costs in return for a few dollars in health or environmental benefits.

I understand the appeal of the passage to advocates, but in the context of the whole section, let alone the whole opinion, the passage doesn’t do anything like the work attributed to it. To understand why, we need to understand the basic peculiarity of the agency’s stance in the regulatory proceedings, at least as the Court saw that stance: EPA relied on regulatory benefits while refusing to consider costs. The agency cited the health and environmental benefits of regulating mercury emissions from power plants, while denying that costs were relevant to the determination whether to regulate.

A moment’s thought suggests that whatever “appropriate and necessary” means, it cannot be reasoned decisionmaking to perform half of a cost-benefit analysis. The Court’s main point, then, was simply that the agency would have to consider both benefits and costs, assuming it considered either. The Court was quite explicit about this point throughout:

Consideration of cost reflects the understanding that reasonable regulation ordinarily requires paying attention to the advantages and the disadvantages of agency decisions {emphasis in original} . . . . EPA has not explained why [a collateral statutory] reference to “environmental effects . . . and . . . costs” provides “direct evidence that Congress was concerned with environmental effects,” but not “direct evidence” that it was concerned with cost. Chevron allows agencies to choose among competing reasonable interpretations of a statute; it does not license interpretive gerrymanders under which an agency keeps the parts of the statutory context it likes while throwing away parts it does not.

In context, then, the Court’s point was simply that it is arbitrary, and (thus) inconsistent with any reasonable understanding of the relevant statutory factors, for agencies to do an incomplete cost-benefit analysis, justifying their action by pointing to benefits but not costs. If and when agencies don their green eyeshades, they may not look at only one side of the ledger while ignoring the other — hardly a surprising conclusion.

Second: Not a single word in Michigan v. EPA lays down any rules about what the agency may or may not count as a benefit or a cost, beyond specifying that cost includes the costs of compliance. There is above all no suggestion that the agency (1) may not cite unquantifiable benefits to justify its action; (2) must quantify its analysis of either costs and benefits; or (3) may only make its decision by comparing quantified costs and benefits. Despite a smattering of passages in which the Court illustrates its points by referring to dollar costs, doubtless for vividness and ease of exposition, the opinion goes well out of its way to explicitly disavow any attempt to require formal, monetized and quantified cost-benefit analysis:

The Agency must consider cost—including, most importantly, cost of compliance—before deciding whether regulation is appropriate and necessary. We need not and do not hold that the law unambiguously required the Agency, when making this preliminary estimate, to conduct a formal cost-benefit analysis in which each advantage and disadvantage 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.

That is why the Court, throughout, uses non-quantified “advantages and disadvantages” as a synonym for “costs and benefits.” The simple idea is that the agency must consider the “cons” as well as the “pros.” That is hardly a Lochnerian mandate for formalized and quantified cost-benefit analysis as a master principle, to be read into general or ambiguous statutes.

Third: A brief word about how Michigan v. EPA fits into the larger principles of administrative law. Cost-benefit analysis is, in the end, a particular method for making decisions — what its proponents call a “decision procedure.” And one of the master principles of administrative law, at least since Vermont Yankee v. NRDC, is that the choice of decision-making procedures is for agencies, unless Congress has specified otherwise. Now, of course, Michigan v. EPA did not involve the original Vermont Yankee situation in which judges attempt to impose procedures purely as a matter of common law; after all, the Court in Michigan v. EPA was interpreting a statutory provision. Yet Vermont Yankee‘s technical holding rests on a broader, principled commitment: that courts must not “stray beyond the judicial province to explore the procedural format or to impose upon the agency [the court’s] own notion of which procedures are ‘best’ or most likely to further some vague, undefined public good.” If mandatory decision-procedures, especially a particular, specialized decision-procedure like quantified cost-benefit analysis, could be implied into general, vague or ambiguous statutory texts like “appropriate and necessary,” or into the general “arbitrary and capricious” language of Section 706 of the APA, the result would be a serious circumvention of the principles underlying Vermont Yankee.

The Court in Michigan v. EPA, however, seemed unmistakably aware of this. Indeed that is quite likely why it went out of its away to reaffirm that the choice of decisionmaking methods, including the decision whether or not to quantify costs, lies within the reasonable discretion of the agency. The very same Term as Michigan v. EPA, six Justices joined in full an opinion for the Court in Perez v. Mortgage Bankers, which reaffirmed Vermont Yankee and its rationale in ringing, emphatic terms, re-declaring “the very basic tenet of administrative law that agencies should be free to fashion their own rules of procedure.” Even the three Justices (among them Justice Scalia) who concurred in the judgment expressed agreement with that part of the Court’s holding; their separate opinions were devoted to other issues. In that sense, Perez was unanimous. Any reading of Michigan v. EPA must take into account the Court’s near-simultaneous reaffirmation of Vermont Yankee. It is implausible to read the former in a way that allows an easy end-run around the latter. Happily, the better reading of Michigan v. EPA, taking the opinion in itself, is also the one that fits seamlessly with the broader fabric of administrative law.

Adrian Vermeule is the Ralph S. Tyler, Jr. Professor of Constitutional Law at Harvard Law School. Follow him on Twitter: @avermeule.