Wednesday, May 31, 2017

Esquivel-Quintana as a Cousin of Bond and Yates (Not Everything Is About Chevron)

There's a puzzle about Esquivel-Quintana that my last post didn't explore.  The Court could have avoided reaching the question of deference by finding the statute unambiguous either way; indeed, both the government and Esquivel-Quintana argued the statute was unambiguous.  And as a textual matter, if we're to pretend that "sexual abuse of a minor" has a definite meaning, the government would seem to have the better of it.  Statutory rape is easily seen as sexual abuse, particularly where, as the agency required, there's a meaningful age differential between the victim and the rapist, and "of a minor" clearly means any minor, not just minors under 16.  Esquivel-Quintana couldn't argue that "of a minor" meant "of some minors," and therefore had to argue that because of a (nonexistent) consensus that the age of consent is 16, statutory rape of minors 16 and over in the 16 states with ages of consent at 17 or 18 isn't "sexual abuse."  This seems, to me anyway, a much harder textual pull.  Yet given two arguments for unambiguity, the Court unanimously chose Esquivel-Quintana's.  How come?

Esquivel-Quintana strikes me as of a piece with Bond and Yates, two infamous recent criminal cases in which the Court reached textually impossible results out of discontent with the government's prosecutorial choices.  In Bond, the defendant spread a mildly toxic chemical on her victim's doorknob in hopes of causing her a rash, and was prosecuted in federal court for use of a chemical weapon.  The chemical unambiguously was a chemical weapon, under the Chemical Weapons Convention Implementation Act's definition of chemical weapons, which, as one would expect, parrots the Chemical Weapons Convention's definition.  But the Court, bothered by the government's exercise of prosecutorial discretion, simply read the definition out of the statute because of its dissonance with the Court's "ordinary" understanding of chemical weapon and replaced it with its surpassingly vague ordinary understanding, incidentally taking the country out of compliance with the Convention, at least at the federal level.

In Yates, a scofflaw went fishing for dozens of undersized grouper, in violation of federal conservation regulations promulgated pursuant to federal marine fishery conservation law, and then, after receiving an order from a deputy of the National Marine and Fisheries Service to return to the harbor with the undersized fish, attempted to conceal his offense by dumping them into the ocean and replacing them with legally caught fish.  Yates was then prosecuted for concealing a "tangible object," i.e., physical evidence, with the intent to obstruct a federal investigation.  The Court, inexplicably feeling that this didn't rise to the level of what ought to be prosecuted as federal obstruction, held that fish weren't tangible objects and that the phrase "tangible object" in the obstruction statute only covered information-recording/storing objects—a meaning which no speaker of the English language had ever given to the phrase before Yates' able federal public defender came up with it, and which was completely redundant given that the statute already criminalized the destruction of records and documents.  The result of the Court's misguided solicitude has been, among other things, that the Second Circuit recently held that a man who covered up his son's triple-fatality arson by repainting the getaway car was innocent of destruction and concealment of evidence, and only liable, at most, as an accessory after the fact.  I would charitably say that bad facts make bad law, except that the Court was aware of the father's conviction at the district-court level and all but acknowledged that its holding would wipe it out.

Bond and Yates's smarter (and more qualified) defenders have claimed that the statutes in Bond and Yates were at least barely ambiguous, or at least that the Court non-frivolously thought they were; this is simply wrong, inasmuch as any claim about the meaning of language can be wrong, but even they acknowledge that the statutes in Bond and Yates were at best only mildly ambiguous, and that if not for its concerns about prosecutorial discretion the Court would likely have deemed them clear.  It strikes me that a variation on this move is going on in Esquivel-Quintana.  If not for the Court's normative objections to deporting college-aged men from the country on account of their relationships with their high-school senior girlfriends, the Court would likely have acknowledged "sexual abuse of a minor" was unclear or even read it the other way.  Having, however, a problem with the government's choice to deport Esquivel-Quintana, and having no way to police that exercise of discretion directly, the Court read that exercise of discretion out of the statute.  

I find this sort of stealth legislation stupendously inappropriate, in addition to usually being extraordinarily maladroit.  At least in this case the Court has announced a clear rule that probably shouldn't generate too many unexpected consequences, though one can of course imagine many extremely abusive relationships between adults and late adolescents, quite unlike Esquivel-Quintana's, that ought to be deportable offenses and won't be under the Court's rule, absent a state statutory-rape statute with special elements for abuse of trust or age differential or other blameworthy aggravating factors.  But this is a cost of the categorical approach (which is also a creature of the Court's creation) as much as it is a cost of the Court's rule, and one which the Court indirectly contemplates. In any event, though the Court has no business replacing Congress's criminal and immigration law with kinder, gentler criminal and immigration law of its choosing, at least decisions like Bond, Yates, and Esquivel-Quintana aren't fooling many people.

Tuesday, May 30, 2017

Esquivel-Quintana and Chevron Avoidance

Take a trip with me back to 1996.  The Macarena is playing on your pre-digital car radio; I'm at my third day camp in three years arguing with my swimming instructor over whether Bob Dole will pick John Engler as his vice-presidential nominee; Bill Clinton is burnishing his well-earned reputation for law and order by signing the Anti-Terrorism and Effective Death Penalty Act and the Illegal Immigrant and Immigrant Responsibility Act (IIRAIRA) into law; and the five most populous states in the country (CA, NY, TX, IL, FL), along with eleven other states which all together account for about 51% of the United States' population,* place the age of sexual consent at 17 or higher.  What odds would you have given that 21 years later, the Supreme Court would hold that IIRAIRA's undefined "sexual abuse of a minor" unambiguously only embraced, in 1996, statutory rapes of minors aged 15 or younger, tracking statutory-rape laws that govern only about 49% of the country?  30 to 1?  20 to 1?  Maybe a little better than that, but presumably you wouldn't have given much of a chance to the Court's holding that the generic definition of an offense was unambiguous on a point on which the country was split 50/50.  Unless, at least, you were a psychic and knew that by 2017 the Court would be falling over itself in order to avoid deciding questions of Chevron's domain.

*Relying on congressional districting allocation in 1996, which probably slightly undercounts the population of these 16 states.

Of course, that's just what the Court unanimously did today in Esquivel-Quintana v. Sessions, in order to continue to avoid deciding whether the Board of Immigration Appeals receives deference on the definition of aggravated felonies (of which "sexual abuse of a minor" is a part), a term in immigration law that carries a mix of civil and criminal ramifications and which people have argued is outside the Board's delegated authority to interpret on a variety of grounds: (1) that the Board cannot receive deference on the interpretation of criminal statutes and that "aggravated felony" must be given a unitary meaning in every statute in which it appears, civil or criminal, given its unitary definition (but see Duke Energy, Utility Air); (2) that the immigration rule of lenity automatically trumps Chevron (no and no); (3) that the Board lacks expertise in criminal law (irrelevant; misunderstands Chevron; assumes that agencies are in the business of interpreting the law and deciding what terms like sexual abuse of a minor "really" mean rather than making policy choices as between permissible interpretations, which is in fact Chevron's theory).  

The claim that "sexual abuse of a minor" unambiguously only encompasses statutory-rape offenses with a maximum victim age of 16 is not, with respect to all the serious people who purport to believe it, a remotely serious position.  That the Court alighted on it, and unanimously no less, just in order to avoid a fractious and perhaps equally divided opinion on whether Chevron applied is obvious.  However, there are a few mildly interesting things to be said about this and other maneuvers in Chevron avoidance.

The first is just that Chevron avoidance is an increasingly frequent and important phenomenon.  As Professor Mila Sohoni insightfully argued at a recent Chevron symposium, questions of Chevron's domain and application are difficult, transsubstantive, and akin to the constitutional questions courts use constitutional avoidance to avoid deciding, inasmuch as Chevron is constitutive of the relationship between courts, agencies, and Congress.  Moreover, Chevron's domain is itself increasingly a constitutional question for the Justices.  If all these transsubtantive, administrative-constitutional (or literally constitutional) questions can be avoided by a decision that merely interprets the statute at hand, it's little wonder that the Court will increasingly be attracted to various strategies of Chevron avoidance, which include good-for-this-case-only exceptions to Chevron that punt tougher questions about how Chevron works, disingenuous unambiguity determinations, or simply ignoring Chevron altogether.

Second, Chevron avoidance, at least as practiced in Esquivel-Quintana, is extremely problematic.  Constitutional avoidance's famous vice is that it creates a doubtful penumbra of stealth enforcement of constitutional norms around the norms themselves by avoiding interpretations of statutes that might be unconstitutional but may very well be perfectly lawful.  (The Chief Justice's choice to decide constitutionality before reaching avoidance in NFIB should, for this reason, be praised, not grumped about for being somewhat outside the modern norm.)  Chevron avoidance has a similar problem in that it will tend to deny deference whenever, as a doctrinal matter, deference is merely in doubt, thereby leading to less deference than would obtain if the Court simply decided whether various asserted exceptions to Chevron really exist or not.  That should be true of all Chevron avoidance.

What's particularly offensive about avoiding Chevron by finding unambiguity is that it requires the Court to prevaricate.  It's one thing to say a statute can bear a weak constitutional reading, and that alternative constitutionally doubtful readings should be avoided; it's really something else for the Court to say a hopelessly vague term unambiguously must be read to mean something impossibly precise in order to dodge questions of deference that would arise were the term deemed ambiguous.  The unstated logic of Esquivel-Quintana is that if it's barely possible to call the statute unambiguous, the Court should say so to avoid a lot of trouble But if a statute's merely barely arguably unambiguous, it's not unambiguous at all.  Note too that constitutional avoidance is explicit, while Chevron avoidance of this flavor must be covert, by its logic; to admit that one's avoiding a question of deference by calling the statute unambiguous is to admit that the statute's not ambiguous. Moreover, the cost of finding unambiguity to avoid delicate questions about Chevron's domain is that it subverts Congress's actual delegation of the question at hand to the agency, replaces that delegation with a fantasized meaning of the Court's creation, and prevents the agency from exercising its wise policy judgment about the gap Congress left the agency to fill.  And while Congress can correct a mistaken act of constitutional avoidance if it wants by making its constitutionally doubtful meaning clearer, it's not clear to me how Congress would go about clarifying that "sexual abuse of a minor" is in fact ambiguous.  We're likely stuck with the meaning that the Court imposed in order to save itself from having to make some difficult decisions.

Third, I don't quite see what the Court thinks it gains by avoiding hard Chevron questions.  Perhaps the Court would have fractured over deference, but what signal does the Court send lower courts when it transparently avoids questions about deference?  Are lower courts permitted to act on those signals?  On the one hand, the best reading of extant Supreme Court precedent is that there is no exception to Chevron for civil/criminal statutes in general or the aggravated-felony definition in particular; on the other, some unknown number of Justices obviously think there should be one and have persuaded their colleagues to do anything to avoid deferring to agencies interpreting those statutes.  As it's unclear whether lower courts can even act on these sorts of signals—some people would say they're merely predictive and that predicting the Court's impermissible, others would say predicting the Court is great, others would justify use of signals on something more like a command theory of precedent—it's extremely difficult to say what lower courts should do with what's now been almost a decade of Chevron avoidance on this question.  A fractured opinion, the mysteries of the Marks doctrine notwithstanding (on which see this exciting cert petition), would provide much more guidance than this.

Fourth, when courts avoid Chevron and interpret irreducibly indeterminate statutes, what you end up with is the interpretive equivalent of junk science.  "Legislative voids," as Justice Gorsuch aptly described them in Gutierrez-Brizuela, don't talk.  What's left of interpretation is a judicial game of spin the bottle.  So we learn, for example, in this opinion that in 1996 a "reliable dictionar[y]" said that the age of consent was "usually" 16.  So it "usually" was, in a sense; to be more exact, it was the age of consent in 31 of 50 states comprising 45% of the population.  The reliable dictionary's generalization adds less than nothing.  

Next, there's a federal definition of sexual abuse of a minor which, the Court admits, "would categorically exclude the statutory rape laws of most States."  No matter; the Court takes the part of the definition it likes, the victim-age minimum, and says that that's unambiguously part of the generic definition because, in part, of the "evidence" provided by the good half of the federal definition.  That's not how evidence works.  Then there's the 51%/49% multi-state survey, discussed above, from which the Court mystifyingly derives a "general consensus," presumably because 34 states are a lot more than 16 and usage in Vermont and Wyoming counts more than usage in California.  Finally, and insultingly, the Court claims that there's "something special" about California's having an age of consent at 18, though (1) the Court holds that the generic age of consent is 16, not 17, making California's specialness irrelevant; (2) half the country at the relevant time had an age of consent over 16; and, (3) in any event ten states (including CA, FL, VA, and WI) accounting for about 28% of the population had ages of consent at 18 at the relevant time, making California not so special at all.*  Were the Board left to make a policy choice about which statutory-rape offenses are serious enough to be deportable, which it actually attempted to do on the basis of studies showing that statutory rapes with meaningful age differentials are especially harmful to their victims, law in this area would be reasoned, rather than a game of dictionary-definition random chance and counting error.

* I hope Justice Thomas enjoys seeing his opinion cited back to him the next time he dissents from an opinion dubiously announcing a national consensus against some modality of punishment.

Sunday, May 7, 2017

Another Addition to the Chevron Anticanon: Judge Kavanaugh on the "Major Rules" Doctrine

(Apologies at the outset for the unnecessary length of this post, though I think it's worth your time; I'm writing an article and haven't had time to revise and shorten.)

I hate to be the guy who's always picking on people's attempts to undermine his favorite doctrine, but what else can I do when Chevron's under attack?  Some people compare Erie, not without reason, to George Washington declining to seek a third term; along analogous lines, I compare Chevron to Gorbachev's peaceful resignation in favor of Yeltsin.  You may not like what came afterwards, but would you like to go back to what came before?  To a certain cast of mind, the most heroic thing the Article III judiciary can do, at least in a post-Brown world, is to make law abnegating its lawmaking power.  So I have to comment on last Monday's dissent from Judge Kavanaugh in the D.C. Circuit's net-neutrality case on his discovery of the "major rules" exception to Chevron

One* of the many moves in Team Death to Chevron by a Thousand Cuts' arsenal is what's heretofore been known as the major-questions exception:  the idea, formulated most clearly in King v. Burwell, that agencies shouldn't receive Chevron deference on really important questions, just the not-so-important ones. (Judge Kavanaugh thinks the doctrine stands for an entirely different idea; I'll get to that shortly.)  The theory of the major-questions exception is not, as some defensive Chevronistas charge, that Congress wants major open questions to be decided by courts.  Rather, it's that Congress probably made a decision, even if obscurely expressed, on any given "major" question, and doesn't usually intend to punt such questions to agencies.  I find this a perfectly defensible premise; the trouble is that the major-questions exception doesn't follow from it.  What follows is that major questions likely present Step One problems where Congress "spoke to the precise question at issue," as Chevron puts it, and decided it one way or another; in the rare event a court should find a major question was genuinely undecided, that would seem a case for administrative gap-filling, not judicial.

* The other 999 cuts include a civil/criminal dual-use statute exception—by the by, if you're someone who's been thinking about or working on Esquivel-Quintana lately, that one's a very bad idea for reasons laid out in the linked postuse of alternative ambiguity-resolving canons, like the not-so-existent doctrine of immigration lenity, at Step One, clause-by-clause inquiries into intended delegation, Breyerian Step Zero balancing, and denying deference to agencies that change their minds based on "shifting political winds [i.e., democracy]," like Chevron says they're supposed to.  

While the major-questions doctrine misguidedly narrows Chevron's domain, it has nothing at all to say about which way major questions should be decided.  It only says that when it comes to major questions, courts must decide de novo (perhaps with Skidmore deference to persuasive agency interpretation) what the statute means, on the theory that when it comes to major questions statutes must mean something.  The agency may win on de novo review, as it did in Burwell, or it may not; the exception itself places no thumb on the scale in any particular direction.  

Judge Kavanaugh, however, disagrees with this account of the major-questions exception.  On his account of the "major rules doctrine," a neologism of his invention, not only is judicial review of major rules non-deferential, it's dispositively tilted against them.  Whereas so-called "ordinary rules" are upheld under Chevron unless they are clearly impermissible or arbitrary and capricious, "major rules" are invalid under the major rules doctrine unless clearly authorized.  This inversion of Chevron for major rules cannot plausibly be described as existing law.  More interestingly, if Judge Kavanaugh is to be taken at his word that his doctrine is neutral as between regulation and de-regulation, it also is logically impossible.

A.  The Major-Questions "Exception" Before Burwell

To lay some descriptive groundwork, there are many partially divergent accounts of the major-questions cases, but the best view of the major-questions exception is that it didn't truly exist until King v. Burwell was decided two years, or if you're fancy, two terms ago.  This may sound startling; we had all, after all, heard of a major-questions exception to Chevron long before Burwell was decided.  Talk of a major-questions exception to Chevron before Burwell, however, was really just a case of loose speech.  Major-questions cases before Burwell had, with maybe half an exception, fallen in the genre of what we might colloquially call "Step One (or Two) cases that make you say 'huh.'"  They were cases where the Court, far from applying an exception to Chevron, applied Chevron itself, albeit in ways that felt less deferential than traditional Chevron review.  These cases are generally thought to have followed two patterns: using the importance of a question to modestly torque Step One, and using the importance of a question to modestly torque Step Two (though I will argue below that this second pattern never really existed).  

1.  The Elephant-in-Mouseholes Canon at Step One

In the more representative and important pattern, one that describes Brown & Williamson, Whitman, and MCI, the Court would encounter a statute that was linguistically ambiguous on whether it allowed an agency to embark on a regulatory course of profound economic and political significance.  Normally, that linguistic ambiguity would suffice for the agency to survive at Step One.  But the Court, perceiving an incongruity between the profoundly consequential regulatory power the agency was claiming on the one hand, and "the vague terms," "ancillary provisions," "modest words," or "subtle device[s]" the agency claimed authorized that power on the other, would conclude at Step One that "Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion."  As Scalia would put it, Congress does not hide elephants in mouseholes.  

As that famous line suggests, this small handful of cases didn't solely turn on gauging the size of the elephant; they also came down, just as much, to the size of the mousehole.  After all, the annals of Chevron cases in the Court are hardly free of instances where the Court deferred to (or in Burwell, upheld) rules of politically or economically elephantine proportions.  The theory of the elephant/mousehole cases was never that Congress must clearly authorize an elephantine rule, but merely that Congress is presumed not to authorize elephantine rules in a manner clearly inconsistent with having authorized an elephantine rule.  

In MCI, for example, the FCC argued that its statutory power to "modify" statutory rate-filing requirements allowed it to read them out of the Communications Act altogether, gutting what the Court called "the heart" of the Act's regulation of common carriers.  In a dictionary sense, "modify" could mean "eliminate"; in context, it couldn't.  In Brown & Williamson, the FDA argued that a statute enacted in 1938 gave it the previously undiscovered and disavowed power to regulate tobacco as a drug—a reading of the statute that was not only highly anachronistic but an extremely awkward fit with the way it regulated drugs generally, as it required the FDA to choose between simply banning tobacco, which Congress had all but forbidden the FDA to do, and absurdly deeming tobacco safe for its intended use.  Linguistically, the statute's definition of "drug" could embrace tobacco; contextually, it couldn't.  And in Whitman, the EPA (CORRECTION: industry, which takes Whitman out of the major-questions canon proper altogether and illustrates that elephant/mousehole is a generalizable rule of interpretation, not limited to the Chevron/pseudo-major-questions-exception context; thank you, commenter gv516) argued that its mandate to set national air quality standards that were "requisite to protect the public health" with an "adequate margin of public safety" allowed it to consider costs in setting down markers for the nation's air quality, relying on attenuated readings of "requisite" and "adequate margin."  The lesson of these cases isn't profound; it's simply that in order for a court to interpret a statute to delegate an agency profound regulatory or deregulatory power, the statute must do so in terms that are at least plausibly proportionate to, and not contextually inconsistent with, such a delegation.  Far from reviving the non-delegation doctrine by interpretive means, or really embodying any doctrine at all, these cases merely represent ordinary context-sensitive statutory interpretation at work.

2.   Majorness at Step Two?

Utility Air Regulatory Group v. EPA, a 2014 opinion by Scalia, is read more often than not as a case where the importance of a question inflected how the Court applied Step Two, or even as a holding that it's likely unreasonable at Step Two for an agency to interpret an ambiguous statute to authorize vastly important regulation.  I don't think that's right.  The standard account, though, goes like this.

In Massachusetts v. EPA, the Court held that the Clean Air Act's definitional section's definition of air pollutant encompassed greenhouse gases.  The EPA, therefore, naturally reached the conclusion that it was required to regulate greenhouse gases as air pollutants under all the Clean Air Act programs that regulate "air pollutants" - including its permitting of pollutant-emitting stationary sources.  Because, however, EPA has to issue a permit for any source that emits 100–250 tons or more of an air pollutant, and because a great many sources emit greenhouse gases at that level, reading greenhouse gases into the permitting provisions' references to air pollutants would have required EPA to issue permits to 6 million sources, up from 15,000 and spend $21 billion a year on permitting, up from a cost of mere tens of millions. The EPA, therefore, sought to "tailor" the tonnage triggers to the enormity of greenhouse-gas emissions by increasing them by a factor of a thousand.

The Court, rejecting this rule, first held at Step One that the EPA was not compelled to read the permitting program's references to air pollutants to include greenhouse gases; the statute's definition of air pollutant could "yield to context" where context so demanded, and hence the references to air pollutant in the permitting program were ambiguous.  Then, at Step Two, the Court held that the EPA's reading of those references to include greenhouse gases was an unreasonable choice as between the alternatives the statute permitted—unreasonable because of the "enormous and transformative expansion in the EPA's regulatory authority" its reading would work.  The Court, it concluded, "expect[s] Congress to speak clearly if it wishes to assign to an agency decisions of vast economic and political significance"; at the least, when an agency claimed to discover previously undiscovered vast regulatory power in an old statute, "a measure of skepticism" was warranted.

Were this picture of Utility Air correct—had it really held that an agency interpretation of an ambiguous statute to give the agency vast regulatory power is presumptively, though not necessarily, unreasonable at Step Two—Utility Air would be a big deal indeed, and go a fair way towards supporting Judge Kavanaugh's claims about a major-rules doctrine.  But it's not.  Rather, Utility Air is just another major-questions Step One case, and an object example of how the knotty relationship between Steps One and Two can trip up very smart people into mistaking the one for the other.  

Utility Air first held that the EPA's reading of air pollutant was not compelled before moving to address an alternative argument that it was at least reasonable; from this structure, people seem to infer that Utility Air held that the statute was ambiguous on whether the permitting programs applied to greenhouse gases.  But that never happened; nothing in the opinion actually says so.  A holding that an agency's reading isn't compelled isn't the same thing as a holding that a statute's ambiguous; it's only a holding that the agency isn't unambiguously right.  What remains possible is that the agency is unambiguously wrong, and that's just what Utility Air said about the EPA's reading of air pollutant.  

To wit, the Court "conclude[d] that EPA's interpretation is not permissible"—not just arbitrary or unreasonable—because EPA itself "admitt[ed] the statute [wa]s not designed to grant" it permitting power over greenhouse-gas stationary sources, given the impossibility of actually regulating them at the statute's tonnage triggers, and the illegality of the EPA's alterations to those unambiguous triggers.  As in Brown & Williamson, the Court found that a semantic ambiguity in a statute's definition of what it regulated was contextually unambiguous in light of the substantive provisions of the statute that acted upon it. The relevant section of the opinion is replete with claims of unambiguity; for example, "applying the  . . . permitting requirements to greenhouse gases would be inconsistent with—in fact, would overthrow—the Act's structure and design," or, "the authority claimed would render the statute unrecognizable to the Congress that designed it."  Likewise, every statement of principle on which the Court relied en route to its supposed Step Two holding concerned unambiguity, not reasonableness; the Court cited cases holding that "a statutory provision that may seem ambiguous in isolation is often clarified by the remainder of the statut[e] . . . because only one of the permissible meanings . . . is compatible with the rest of the law," and that agency interpretations that are "inconsistent with the design and structure of the statute as a whole" are, of course, wrong.

If the EPA were so clearly wrong, why, one might ask, did the Court bother to say it expected Congress to "speak clearly" if it wished to assign agencies vast regulatory power?  Doesn't this suggest it thought the statute unclear?  No.  Rather, the importance of the power the EPA asserted appears as a makeweight in the Court's opinion (if not indeed a gratuitous dictum), punctuating a discussion of the absolute irreconcilability of the EPA's reading of air pollutant with the regulatory scheme in which that term was embedded, that the Court seems to have merely adduced as further evidence that the statute was indeed clear.  As in the elephants/mouseholes cases, the Act's extreme obscurity, at best, on authorizing a regulation of vast consequence was an indication that the Act clearly did not authorize a regulation of vast consequence.  Given the somersaults the EPA admitted it had to turn to fit greenhouse gases into the permitting program, Utility Air can hardly be said to stand for the proposition that whenever a statute is merely unclear on whether it authorizes a regulation of vast consequence, it doesn't.

Finally, assuming I'm wrong and Utility Air is a Step Two case, would it then stand for the proposition that an agency reading of an ambiguous statute to vest it with profound regulatory authority is per se unreasonable?  No.  As even those who think Utility Air's a major Step Two precedent acknowledge, it at most says the Court is skeptical of profoundly consequential regulations at Step Two, not that it's always unreasonable to read an ambiguous statute to authorize one.  The Court wrote that agency claims to have discovered vast, previously undiscovered regulatory power in an old statute are owed "a measure of skepticism"—not that such claims are off the wall.  And while the Court did say in the very next sentence of its opinion that it "expect[s] Congress to speak clearly" when giving agencies vast regulatory powers, that can only be read as an expectation or presumption, not a clear-statement rule.  Otherwise, the Court would greet interpretations of ambiguous statute to yield vast regulatory powers with more than skepticism, and otherwise, all the extensive discussion of how EPA's interpretation of air pollutant didn't cohere with the permitting program would have been unnecessary.  Moreover, had Utility Air announced a clear-statement rule for major rules, Chevron's presence in the opinion would make no sense; a clear-statement rule would displace Chevron altogether.

B.  Burwell—Sometimes an Exception is Just an Exception

Before Burwell, the Court had never held that the majorness of a question took it out of Chevron altogether.In Burwell, the Court did—but then upheld the agency rule it declined to defer to.  Burwell, of course, involved whether Obamacare's health-insurance tax credits were available to participants in federally run insurance exchanges, or only state-run insurance exchanges.  Critically, most states failed to establish their own exchanges; a decision either way would have enormous consequences on the efficacy of Obamacare.  

* Gonzales v. Oregon, nine years earlier, mentioned majorness as one of a multitude of reasons to hold at Step Zero that the Attorney General lacked the authority to even interpret a statute with the force of law on a particular question, quite apart from whether the interpretation he rendered was correct; interestingly, even at Step Zero, the Court used elephants/mousehole reasoning on the interpretive question of whether the A.G. had rulemaking power on the underlying question in the case, arguing that it would be incongruous for Congress to hide rulemaking power on the subject of assisted suicide in a provision giving the A.G. the power to make rules concerning the registration of physicians to dispense certain drugs.

The IRS, which had rulemaking authority with respect to the tax credits, had issued a regulation granting the tax credits to participants on federally run exchanges, notwithstanding language in the statute at least suggesting they were ineligible.  But the Court refused to defer to the IRS's rule.  Inappositely citing two elephant/mousehole cases, the Court reasoned that Congress wouldn't have wanted to delegate such an important question to an agency and must have decided it itself.  Though the Court refused to apply the ordinary Chevron standard of review, it nevertheless took pains over whether or not the statute was ambiguous, which it seemed to view as a threshold question, and found at some length that it was.  It then held that the correct way to resolve that ambiguity was in favor of the IRS's conclusion that it was required to make tax credits available on federal exchanges—thereby holding that the statute ambiguously authorized the IRS to issue a rule of "deep economic and political significance" that was "central to th[e] statutory scheme."  In the closest the Court has ever come to making a doctrine out of the major-questions exception, the Court clearly held that ambiguous statutes can authorize major rules.

C.  Judge Kavanaugh's "Major Rules Doctrine"

1.  Its Lack of Legal Pedigree

Judge Kavanaugh, reading exactly the same cases I've just discussed, believes they stand for the following proposition: "For an agency to issue a major rule, Congress must clearly authorize the agency to do so.  If a statute only ambiguously supplies authority for the major rule, the rule is unlawful."  This is puzzling.  It is also, given his definition of "major rules" to include major regulatory or de-regulatory rules, completely incoherent.

The heart of Judge Kavanaugh's argument that we have a major-rule doctrine is a series of bullet-pointed overreadings of the major-questions cases.  Each bullet-pointed overreading concludes with a few sentences that go like this: "It would have been a major step for the [agency] to [do what it wanted to do].  Yet there was no clear statutory authority for the [agency] to do so.  The Court therefore rejected the rule/thus invalidated the rule/as a result, vacated the rule."  Unfortunately for Judge Kavanaugh, this just isn't what the major-questions cases say.  The Court refused in MCI and Whitman to find elephants in mouseholes, or in the case of Brown & Williamson and Utility Air, to find them in incongruous tundras—details of the Court's reasoning which he conveniently skips over.  Nothing in those cases remotely stands for the proposition that the Court will never find an elephant in a statute that's merely ambiguous about whether it contains an elephant.  Indeed, in Burwell the Court found authorization for a rule that "involve[d] billions in spending each year and affect[ed] the price of health insurance for millions of people" in a statute that the Court struggled to avoid holding unambiguously foreclosed that authority.

What, then, does Judge Kavanaugh do with BurwellBurwell, he says, is different.  The major rules doctrine, he explains, only applies to rules that "regulate or de-regulate (as opposed to tax or subsidize) some major private activity."  When an agency merely "interprets a major government benefits or appropriation provision of a statute," all that follows is that Chevron may not apply; the agency's interpretations can still stand, absent deference, even if only ambiguously authorized.  

I suppose this is the arbitrary distinction one might have to draw if one read a major-rules doctrine out of the elephants/mouseholes cases and needed to distinguish Burwell.* That said, I can't understand Judge Kavanaugh's distinction.  Why would Congress have to clearly authorize agencies to regulate or de-regulate major private activities, but not have to clearly authorize agencies to tax, subsidize, or de-subsidize major private activities?  Do the non-delegation doctrine or "the presumption that Congress intends to make major policy decisions itself," which Judge Kavanaugh claims motivate the major-rules doctrine, have more strength as applied to private activity?  If so, why?  One would assume it's just as unconstitutional for Congress to delegate how to tax or spend than how to regulate (maybe more so given how central taxing and spending is to Article I), and that Congress is just as determined to make major taxing and spending decisions itself as it is to make major regulatory decisions (again, maybe more so).  Besides, is the distinction between "regulating or de-regulating" major private activities and "taxing or subsidizing" them even a meaningful one?  Aren't taxes and subsidies regulatory?

* Though I can probably offer Judge Kavanaugh a better one.  How about the distinction that in Burwell, the Affordable Care Act necessarily authorized a major rule in some direction on the availability of tax credits on federal exchanges—the IRS had to say something on the subject—whereas in the other major-questions cases, mere inaction was an alternative? 

Finally, Judge Kavanaugh quotes a series of academic writings by then-Judge Breyer, William Eskridge, and Abbe Gluck and Lisa Bressman.  The strange thing about all his quotations is that they rather plainly summarize or anticipate the Court's real major-questions exception, not Judge Kavanaugh's major-rules doctrine.  Eskridge, for example, writes that "judges presume that Congress does not delegate its authority to settle . . . major social and economic policy decisions"; Gluck and Bressman describe a "presumption of nondelegation" as to major policy questions, and quote a congressional drafter who says that "drafters don't intend to leave [major questions] unresolved."  All any of that says is that  Congress decides major questions itself—not that it necessarily announces its decisions in unambiguous terms.  The fact that Congress is ambiguous on the authority for a major rule doesn't mean, as Burwell illustrates, that it didn't decide to authorize it.

2.  Its Incoherence

As Lisa Heinzerling has persuasively argued about the much milder anti-major-rule skepticism on offer in Utility Air, presumptions against major rules are inherently biased against regulation. As a formal matter, however, Judge Kavanaugh is careful to insist that his rule applies to rules that "regulate or de-regulate . . . some major private activity," and reads MCI as a case where the major-rules doctrine precluded major deregulation.  This begs a tough question.  If on Judge Kavanaugh's rule, a statute that is ambiguous on a major question neither authorizes major regulatory nor de-regulatory activity, won't anything the agency does on the question be unlawful?  

For example, suppose the major-rules doctrine really were the law and the Court applied it in Burwell.  How would that work?  According to the Court, the statute was ambiguous on whether federal exchanges included tax credits, and according to the Court, the question of whether they did was one of major economic and political significance.  It would seem to follow that the IRS's rule giving credits on federal exchanges couldn't stand; such a rule would undeniably be a major rule.  So too, however, would a rule denying credits on federal exchanges; that rule also would massively affect federal spending and insurance prices.  So it would seem to follow that that rule couldn't stand either.  Under the major-rules doctrine, whatever the IRS did about credits on federal exchanges would be invalid.

Now it's true that Judge Kavanaugh gerrymanders Burwell out of his rule, though only because it was decided in a way that refutes his theory, not because he has an exception for cases where his rule paradoxically renders any disposition of a major question illegal.  That said, consider, to be fair, how the major-rules doctrine would apply in the cases where Judge Kavanaugh says it applies:  cases that involve the regulation or deregulation (not the taxing or subsidizing) of some major private activity.  Here, Judge Kavanaugh may say that the doctrine only cuts one way:  barring, in cases of ambiguity, major regulation or de-regulation that differs from the status quo.  Merely standing pat isn't itself a "major rule."  And so, in Brown & Williamson, simply not regulating tobacco as a drug would not have been a major rule; in MCI, not eliminating rate-filing would not have been a major rule; and in the net-neutrality case actually before him, simply not reclassifying Internet service providers as common carriers would not have been a major rule.

This works as far as it goes, but it doesn't go very far.  For at the back of agency stasis, there's always an antecedent agency decision that the stasis maintains; on any given ambiguous major question, there will at some time be a time for agency choosing, as there was shortly after Obamacare's enactment in Burwell.  Take the reclassification of information service providers.  It's true that what Judge Kavanaugh says the major-rules doctrine required of the FCC—simply continuing to sit on their hands while Internet service providers discriminated between different sources of content—isn't a major rule, indeed isn't a rule at all.  But before the FCC reclassified broadband ISPs as common carriers, it classified them some time after the technology came along.  And at that moment, whatever the FCC would have to say on the ambiguous subject of ISP classification would necessarily be a "major" rule, one which would profoundly affect the economy if not indeed society at large.  So it would seem that the FCC's initial classification of broadband ISPs in 2002, on Judge Kavanaugh's view, was necessarily invalid, though that's obviously not what the Court thought in Brand X when it deferred to that classification.

Judge Kavanaugh hasn't completely overlooked this point, at least not as to ISP classification specifically.  (He seems unaware that his rule generally results in paradox.)  His puzzling response is that the FCC's original decision that broadband Internet was an information service, not a telecommunications service subject to common-carrier regulation, wasn't a major rule, but a mere "ordinary" one.  How so?  Because that classification "did not entail common-carrier regulation and was not some major new regulatory step of vast economic and political significance."  Absent a definition of major rules that turns on how regulatory they are, rather than their significance, which Judge Kavanaugh suggests elsewhere in the opinion is not the definition he's using (recall that a major rule can de-regulate), I can't understand this distinction.  Surely the FCC's decision to regulate ISPs as non-common-carriers was a decision of vast economic significance for precisely the same reasons that their later decision to regulate ISPs as common carriers was one.

Judge Kavanaugh, then, is trapped in a dilemma.  On the one hand, he can claim, as he initially does, that his rule is neutral as between regulation and deregulation, or rules on profound questions that impose more regulation and rules on profound questions that impose less.  This seems to follow from the premises he says animate his rule:  that Congress is presumed not to delegate major questions to agencies and that such delegations are constitutionally questionable, neither of which seems to suggest courts should look more kindly on delegations to deregulate than delegations to regulate.  However, if he sticks to this formulation of his rule, what follows is that an agency can't act in any direction on an ambiguous major question, even though the agency must act in one way or another.  (The FCC couldn't, in Brand X, have simply not classified ISPs, anymore than the IRS in Burwell could have declined to either make tax credits on federal exchanges available or unavailable.)

On the other hand, Judge Kavanaugh can acknowledge that his rule is anti-regulatory, and has to be formally anti-regulatory for the rule to work.  He can define major rules as rules on a major question that choose to impose more regulation rather than less, and in this way save his rule from the paradox of forbidding agencies to do anything.  But this comes at a severe cost:  neither he nor anyone else has any argument for why ambiguous statutes cannot be read to authorize profoundly important onerous regulations, but can be read to authorize profoundly important choices to regulate lightly or not at all.  

It's true that such a rule neatly avoids delegation problems; on Judge Kavanaugh's rule, the less regulatory of two choices on an ambiguous major question isn't just permissible, but mandatory (though he doesn't appear to have thought this through).  For example, in Brand X, if the FCC's choices were the "major" rule of treating ISPs as common carriers and the "ordinary" rule of treating them as mere information-service providers, the major rule was illegal and the ordinary rule FCC's only permissible option.*  So this version of major-rules doctrine would address Judge Kavanaugh's delegation concerns by foreclosing all delegations on ambiguous major questions.  The trouble is that the direction in which it forecloses them is completely unmotivated.  Judge Kavanaugh can explain why we'd want a rule that presumes delegations on major questions away; he just can't explain why we'd want a rule that presumes them away in favor of less regulation.  A rule that Congress always prefers more regulation in cases of major ambiguity follows equally from his premises as the rule he's proposed.

* This is yet another illustration, incidentally, of why Judge Kavanaugh is wrong about the current state of the law, as Brand X was decided on the theory that the FCC had a choice.

The reason Judge Kavanaugh finds himself in this corner is that the major-rules doctrine isn't really responsive to his anti-delegation premises at all.  What's logically responsive to a presumption that Congress doesn't delegate major questions is de novo review on the question of what Congress decided—that is to say, Burwell.  Whether the court concludes Congress opted for more regulation or less, whatever it decides will, in theory at least, be Congress's decision; either way there won't have been any delegation.  To require, on top of withholding Chevron deference on major questions, that courts invalidate any "major rules" on those questions is completely unnecessary to addressing whatever non-delegation concerns or presumptions against delegation one has.  It also results in logical impossibility; absent invalidating any statutory ambiguities on major questions and asking Congress to decide them more clearly, something which Judge Kavanaugh doesn't call for, what follows is that agencies must be silent on major questions altogether, though they practically and legally can't.  In order to save his rule from incoherence and arrive at a place that appears to respond to his non-delegation concerns, Judge Kavanaugh has to presume an anti-regulatory answer to major questions.  That answer, however, has nothing at all to do with the concerns that supposedly motivated the project in the first place.