Today’s decision in Utility Air Regulatory Group v. EPA partially sustained and partially invalidated a major greenhouse gas regulation. In doing so, the Court passed up an opportunity to clarify the famous and deceptively familiar deference inquiry established in Chevron, U.S.A. Inc. v. Natural Resources Defense Council, Inc. The question is whether Chevron (i) requires, (ii) forbids, or (iii) permits judicial decisions that uphold agency interpretations as not just reasonable, but mandatory. This issue was once thought to have been answered by Chevron Step 1, which seemed to require judicial review for mandatoriness. In UARG, however, the familiar Chevron Step 1 is mostly absent from the scene — even though the Court expressly considered the possibility that the agency’s reading was “compelled,” or mandatory.
The best explanation for UARG is that the Court views the traditional Chevron Step 1 inquiry into mandatoriness as optional. In many cases, the only relevant Chevron question is whether the agency has acted reasonably. That is particularly true when the agency reading is invalidated as unreasonable. When upholding agency interpretations, however, it sometimes makes sense for a court to go further and opine that the agency’s reading is not just reasonable, but mandatory. The upshot is that what used to be called “Chevron Step 1″ has effectively become an optional additional step. In this respect, Chevron resembles modern qualified immunity doctrine, which always asks whether the challenged governmental action was reasonable, but also gives courts discretion to reach the merits.
In the decision below, the D.C. Circuit ruled based on “the familiar Chevron two-step.” Quoting Chevron itself, the D.C. Circuit divided the inquiry into two parts. “First . . . if the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” But “if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” While the D.C. Circuit doesn’t always frame the Chevron inquiry in this way, the foregoing statement probably deserves to be called the traditional view of Chevron: two steps, both concerned with statutory interpretation. Applying that approach, the D.C. Circuit appeared to rest on Step 1 by holding that the statute was “unambiguous” and that it “compelled” the agency’s reading.” Yet that traditional view of the “familiar Chevron two-step” has been much debated.
One increasingly popular alternative approach, set out in an important paper by Professors Stephenson and Vermeule, is that both steps are really asking the same thing and so should be reduced to a single step. Step 1 asks if the statute “clear[ly]” forecloses the agency’s view, and Step 2 asks if the agency has adopted a “permissible” construction. Under either phrasing, the test is whether the agency has occupied a statutory ambiguity. Put yet another way, the sole question is whether the agency’s reading is reasonable. Whether the agency’s explanation was arbitrary would be a separate inquiry undertaken under the APA. Both before and after Stephenson and Vermeule’s paper, Justice Scalia (UARG’s author) wrote opinions that substantially supported their point of view (more on that below). And, when granting cert in UARG, the Court adopted a question presented that seemed to frame the issue as having only one step: “Whether EPA permissibly determined that its regulation of greenhouse gas emissions from new motor vehicles triggered permitting requirements under the Clean Air Act for stationary sources.”
Another salient view of Chevron differs from both the traditional picture and the one-step version of Stephenson/Vermeule. On this third approach, Step 1 goes to statutory interpretation, whereas Step 2 asks whether the agency’s reasoning was arbitrary and capricious. Professors Bamberger and Strauss defended two-step Chevron largely on this ground. But while some cases adopt that understanding of Chevron, many other decisions, including UARG, don’t mention arbitrariness review. Further, arbitrariness review is separately provided for under the APA, so (as Stephenson/Vermeule argued) its additional inclusion within the Chevron rubric would be superfluous. Bamberger/Strauss also differed from Stephenson/Vermeule in placing greater emphasis on the fact that Step 1 allows courts to find statutory meaning to be not just reasonable but unambiguous. This point has important implications, since a lot can turn on whether an agency view is reasonable (and therefore revisable) or mandatory.
Against this background, consider UARG’s first Chevron ruling — the one against the agency. As noted, the Court framed the question presented in terms of whether the agency view was permissible, full stop. What’s more, the Court’s recitations of the Chevron framework, like Justice Scalia’s past writings, often sounded like the one-step view. For instance, the Court said that “[t]he question for a reviewing court is whether … the agency has acted reasonably and thus has ‘stayed within the bounds of its statutory authority.’” Yet just after the Court wrote these words, it proceeded to ask whether one of the EPA’s readings flowed from “the Act’s unambiguous language” and whether the Act “compelled” the agency’s reading. Only after answering those questions in the negative did EPA go on to ask the question it had set for itself at the outset — namely, whether the EPA had adopted a “reasonable construction of the statute.” This two-step analysis is more succinctly put in the Court’s syllabus, which notes that “[t]he Act neither compels nor permits” the agency’s reading. Justice Alito’s separate opinion similarly noted that “the EPA is neither required nor permitted” to adopt its interpretation.
The Court’s first Chevron holding leaves something for everyone. One possibility is that UARG actually followed the traditional view of Chevron articulated by the D.C. Circuit, where Step 1 inquires into the unambiguous intent of Congress before Step 2 considers permissibility. Put another way, UARG separately asked both about mandatoriness (Step 1) and about reasonableness (Step 2). Another possibility is that, consistent with the Stephenson/Vermeule approach, the Court really meant it when it said it was concerned only about permissibility, which, after all, is the ultimately dispositive issue in any agency deference case; it just so happened that one very important argument for permissibility would necessarily also mean that the agency’s view was compelled. Finally, UARG might even be compatible with the Bamberger/Strauss view. This reading is more of a stretch, since the Court didn’t mention arbitrariness review and certainly seemed to be interpreting the statute, rather than merely passing on the agency’s stated rationale. Still, the Court’s Chevron analysis could conceivably be viewed as an extended Step 1 inquiry into statutory meaning that obviated the need to reach arbitrariness at Step 2.
The picture changes a bit when it comes to the Court’s second Chevron ruling, which went in favor of the agency. Again, the Court framed Chevron as a one-step inquiry. This time, however, the agency’s reasoning didn’t run afoul of the statute’s unambiguous meaning. More than that, there was a plausible argument that the agency’s view was unambiguously correct. As the Court put it, the statutory text at issue in the second Chevron ruling was “far less open-ended,” and “the more specific phrasing of the BACT provision suggests that the necessary judgment has already been made by Congress.” The Court further noted that certain arguments would apply “[e]ven if the text were not clear,” thereby suggesting that the text was clear. Despite all this, the Court found only that the agency had acted permissibly, stating that its “narrow holding” was “that nothing in the statute categorically prohibits EPA from interpreting the BACT provision” as it had. The Court thus chose to find the agency’s reading reasonable, without passing on whether it was also mandatory — contrary to the traditional view of Chevron, which seems to demand that courts first determine whether the statute is unambiguous. Because of this restraint, the EPA could in the future reject the reading that the Court sustained and instead adopt, for example, the different reading offered in Justice Alito’s separate writing.
UARG illustrates that debates over the structure of Chevron tend to have relatively low stakes when the agency loses, since that finding necessitates that the agency’s view was neither required nor reasonable. Yet agencies sometimes win. And, when they do, it is important to know whether their readings are mandatory. For instance, the D.C. Circuit didn’t have to write its opinion in terms of agency compulsion; it could simply have said that the agency’s view was permissible. Should the D.C. Circuit have been: required to consider mandatoriness, forbidden from doing so, or simply allowed to do it? The Court has never squarely confronted this question, but its rulings over time suggest that mandatoriness findings are neither prohibited nor required, but optional. That is, the Court sometimes finds agency readings to be unambiguously correct. In other cases, like UARG, the Court settles for finding reasonableness, while casting mandatoriness findings as a distinct and viable possibility.
Nor have Justice Scalia’s past opinions taken a definite position on this issue. In a passage that appeared at the outset of Stephenson and Vemeule’s paper, Scalia wrote that “any agency interpretation contradicting what Congress has said would be unreasonable.” That statement is silent as to what happens when the agency prevails. More recently, in United States v. Home Concrete & Supply LLC, Justice Scalia wrote in a concurrence that “the so-called ‘Step 1’ determination of ambiguity vel non” is a “customary” but “hardly mandatory” component of Chevron. In other words, the mandatoriness inquiry normally ensconced in Step 1 is a distinct but optional analytic step. In a footnote, Scalia tacked back toward the Stephenson/Vemeule view by asserting that “[w]hether a particular statute is ambiguous makes no difference if the interpretation adopted by the agency is clearly reasonable.” But, again, that assertion is wrong. While mandatoriness findings don’t make a difference as to whether the agency wins, they most certainly do make a difference as to the agency’s ability to revise its interpretation. Since Scalia is well aware of that fact, his footnote may have meant to say that there is no legitimate reason for a court to issue mandatoriness findings, since doing so would amount to a gratuitous holding. UARG supplied a chance to put that normative view into practice.
My own view (explained in more detail here) is that Chevron originally required consideration of mandatoriness (at Step 1) before consideration of reasonableness (at Step 2), much as qualified immunity cases once required consideration of the merits in addition to reasonableness. In both contexts, structuring judicial decisionmaking according to a two-step process forced courts to clarify the law, to the benefit of regulators and litigants alike. But, also in both contexts, forcing courts to issue unnecessary rulings created problems, including by increasing the risk of error and raising legitimacy concerns. In response to those problems, qualified immunity jurisprudence has now embraced a more flexible case-specific approach to deciding whether to reach the merits. Chevron jurisprudence should explicitly adopt a similar approach: while courts should always resolve the question of reasonableness, they should also have the option to rule on the distinct mandatoriness question. In other words, Chevron (like qualified immunity) should be understood to have two steps, where the second step is optional.