Tuesday, August 1, 2017

Supreme Court 2016 Statutory Term in Review: Textualist Pathologies in Advocate Health Care Network

I'm going to do a review of the Court's most methodologically interesting OT 2016 statutory cases, less the ones I've already written on.  I was inspired to write this series of posts by a good column in The Hill by Professor Jonathan Nash, who describes Justice Kagan's opinion in Advocate Health Care Network v. Stapleton as a "tour de force in textualist interpretive technique."  I'm not even sure if Advocate Health Care is a good textualist opinion; it seems to turn almost entirely on a basic mistake.  But it is, I think, a very representative opinion of how textualism is practiced today.  Advocate Health Care exhibits three textualist pathologies that we textualists should cure ourselves of:  conflating literal meaning with legal meaning, overconfidence about what a statute literally means, and overaggressive application of the anti-surplusage canon.

1.  The Statute

Advocate Health Care involved the church-plan exemption from ERISA.  A "church plan" is defined as "a plan established and maintained . . . for its employees . . . by a church" or association of churches.  A subsequent amendment to ERISA clarified that a "plan established and maintained for its employees . . . by a church includes a plan maintained by an organization the principal purpose or function of which is the administration or funding of a plan or program for the provision of retirement benefits or welfare benefits . .  for the employees of a church . . . if such organization is controlled by or associated with a church[.]"  Call these organizations principal-purpose organizations, or PPOs.  

The question then arose whether a plan maintained and established by a principal-purpose organization, i.e., not by a church, counts as a church plan, or whether a plan must at least be established by a church, as the church-plan definition says, to be a church plan.  Several circuits held the latter, reasoning that while the statute broadened the definitional requirement of church maintenance to include principal-purpose organization maintenance, it never said that a plan both maintained and established by a principal-purpose organization counted as a plan established and maintained by a church.  The entities that offered such plans, on the other hand, argued that that's just what the statute says; the statute literally equates merely being maintained by a principal-purpose organization with being "established and maintained . . . by a church" when it says that plans established and maintained by churches include PPO-maintained plans.

2.  The Opinion

What does Justice Kagan do with this problem?  Essentially, she question-begs her way to a conclusion.  The heart of the opinion is this syllogism:
Premise 1: A plan established and maintained by a church is an exempt church plan.
Premise 2: A plan established and maintained by a church includes a plan maintained by a principal-purpose organization.
Deduction: A plan maintained by a principal-purpose organization is an exempt church plan.
The question Justice Kagan begs is whether "includes a plan maintained by a PPO" means "includes all plans maintained by PPOs," rather than simply "includes (or put otherwise, doesn't exclude, or includes some) plans maintained by PPOs"—implicitly, the ones established by churches.  It's not obvious what "includes" means here; that's just the point at issue.  Consider, for example, as Justice Kagan gets around to doing some pages later, a hypothetical statute that says "a disabled veteran of the active Armed Forces includes a person who served in the National Guard."  Does that mean that every person who served in the National Guard is deemed a disabled veteran?  No—just that disabled veterans include National Guardsmen, namely the ones who are disabled.

What does Justice Kagan do with this hypothetical?  She says it's a special case.  Generalizing from the features of this hypothetical, she claims that statutes that say a two-part definition/defined term—like "disabled veteran," or "established and maintained by a church""includes" some type of thing and mean that the defined term includes some subset of that type of thing that otherwise meet the definition will have two features.  First, the two parts of the definition or defined term will be relatively "distinct" and "independent," which increases the chance of their "standalone relevance" such that a gloss on what can be read as just one part of the definition is less likely to modify both parts.  

Second, and much more importantly, in the case of the disabled veteran, we know that defining disabled veterans to include all National Guardsmen "could not possibly have been what Congress wanted"; "the mind rebels against reading the statute literally, in line with the logical . . . principles [the syllogism] described above."  In other words, though she doesn't quite say this, we can depart from the supposed literal reading of "includes" to mean "includes all" only when reading "includes" that way is absurd, or very near to it; at one point she suggests her literal reading of "includes" in ERISA would have to be "utterly untenable" to be rejected.  (Here I agree with Professor Nash, who also detects an absurdity subtext at this point in the opinion.)  In the case of the church plans, she claims, the mind doesn't rebel against deeming plans neither established nor maintained by churches, but rather established and maintained by church-affiliated PPOs, to be both established and maintained by churches.  (Your mind's mileage may vary.)  

Besides claiming that "a plan established and maintained by a church includes a plan maintained by a PPO" literally means "includes all plans maintained by PPOs," Justice Kagan also reasons that the alternative reading would make surplusage of the two words "established and."  Had Congress only wanted to modulate the meaning of "maintained by a church" to include PPO maintenance, it could have said that a plan maintained by a church included a plan maintained by a PPO; the phrase "established and" must add something lest we suggest that ERISA contains two surplus words, and that something is to make clear that the PPO maintenance proviso overrode the entire church-plan definition, not just the maintenance half.   

3.  Textualist Pathology #1:  Overconfidence About Literal Meaning  

A key premise of Justice Kagan's opinion is that "a plan established and maintained by a church includes a plan maintained by a PPO" at least literally means that every plan maintained by a PPO is defined as a plan established and maintained by a churchand that sentences like "a disabled veteran of the active Armed Forces includes a person who served in the National Guard" literally mean that every National Guardsman is a disabled veteran.  Is it true, though, that "a disabled veteran of the active Armed Forces includes a person who served in the National Guard" even literally means that every former National Guardsman is a disabled veteran?  I don't think that's at all obvious.  

Consider, to take just one of countless examples of an alternative meaning of "includes," the following exchange.  Someone tells a man he has to buy a pair of "nice black shoes."  He asks for clarification about what counts as a nice black shoe.  Do, he asks, loafers count as nice black shoes?  Yes, his interlocutor replies, "nice black shoes include loafers," or, if you want to get the grammar on all fours with ERISA's, "yes, a 'nice black shoe' includes a loafer."  What has she literally said?  That every loafer of any color and quality is a "nice black shoe"?  Must we invoke anti-absurdity to interpret this sentence?  No; what she literally has said is simply that the set of things called "nice black shoes" includes (or doesn't exclude) some loafers of an indefinite number, a number which, as far as purely literal meaning goes, may or may not be equal to the number of all the world's loafers What she implicitly means is that "nice black shoes" include loafers, so long as those loafers are (1) black and (2) nice.  As we know that not all loafers are black or nice, we know that she is not saying that every loafer is a nice black shoe.

How did Justice Kagan get the Court to unanimously agree that "includes" literally means "includes all" and that the mind must rebel against this super-inclusive reading of "includes" in order to read "includes" otherwise?  I have no idea.  When my mind sees "the term 'disabled veteran' includes a National Guardsman," it doesn't rebel against what I perceive to be the sentence's literal meaning; I just think the sentence is literally saying that some disabled veterans are National Guardsmen and clarifying that National Guardsmen can be disabled veterans.  

Now, I don't doubt that "includes" sometimes means what Justice Kagan says.  A statute that helpfully clarified that "a fruit includes a tomato" would just mean that all tomatoes are fruits, not that only the really fruity ones (whatever that means, if anything—oddly enough, I've never eaten a tomato) are.  But a statute that said "the term 'seedless fruit' includes a tomato"?  That only means that seedless fruits under that statute include seedless tomatoes, and it doesn't literally mean all tomatoes either.  What it literally means is simply that "seedless fruit" is not exclusive of tomatoes; the requirement that tomatoes be seedless to count as "seedless fruit" is technically, I think, implied (rather than literally said) given our assumption that the statute isn't self-contradictory and doesn't mean that seedless fruits include seeded fruits.  

What this shows, I think
—though it's hard to say for sure, as the linguistics of "includes" are surprisingly complicated—is that even in definitions, "includes" is a word of multiple meanings.  "Includes" can just mean "includes all of," but it can also mean "includes at least some" or "does not exclude."  Neither is the word's singular literal meaning; both are literal meanings of includes, and it will require context to figure out which is meant—to dismambiguate, as linguists say, the word.  But it is characteristic of textualists these days to hastily conclude that one of several possible meanings of a word is its literal meaning and presumptive legal meaning, subject to some heavy burden of persuasion otherwise.  More broadly, overconfidence about normal meaning of all sorts, whether of a single word or the number of antecedents a word or phrase normally modifies or any other number of usage patterns, is one of contemporary textualists' most characteristic traits.

4.  Textualist Pathology #2:  Conflating Literal Meaning with Meaning Simpliciter and Relegating Non-Literal Meaning to the Absurdity Doctrine.

Suppose Kagan were right that "a disabled veteran includes a former National Guardsman" literally means that every former National Guardsman is a disabled veteran.  Perhaps I am right about what "includes" can mean in sentences like "nice black shoes include loafers," but not about what "includes a ____" means in sentences like "a seedless fruit includes a tomato."  "Includes a ____," we might think, always literally means "includes every such thing" while "includes ___" can mean "includes some ___."  If we think that she's right about literal meaning in the disabled-veteran case, is it necessary that we find that reading "utterly untenable" and conclude it "could not possibly have been what Congress wanted" before we reject the statute's literal meaning?  I don't think so, because literal meaning doesn't exhaust meaning.

Suppose the disabled-veteran hypothetical read a little differently.  In light of a dispute in the courts about whether a veteran with PTSD can count as a disabled veteran, Congress amends its disabled-veteran statute to say that "a disabled veteran includes a veteran suffering from PTSD."  Assuming that the statute literally means that disabled veterans include all veterans with PTSD, must we read the statute that way?  I don't think so.  It is possible—the mind doesn't rebel against it—that Congress wanted to protect every veteran with PTSD, however slight.  But maybe what it meant was only that veterans with PTSD can be disabled veterans; they still have to suffer from PTSD badly enough to be "disabled," as the statute elsewhere defines disability, and an extremely mild case of PTSD won't suffice.  This is possibly the more plausible guess about what Congress meant.  But if we follow Justice Kagan, we will have to say that because it's only more plausible than the statute's literal meaning and not absolutely compelled by absurdity doctrine, we're stuck with the statute's literal meaning.

This, I think, is mistaken, because literal meaning isn't all that meaning is.  A statute's meaning can consist of what the statute implies, but doesn't literally say.  For example, if it weren't explicitly addressed, the disabled-veteran statute would literally embrace veterans of any country's armed forces.  But no one would think the statute means that disabled veterans of the Syrian Army receive disabled-veteran benefits.  Limitation to service in the United States' armed forces is implied.  This is a cousin of what linguists call quantifier-domain restriction:  the idea that if you say "everybody will be at the party," the quantifier "everybody" is impliedly restricted to only extend over some "domain" or subset of persons that "everybody" literally signifies, like your circle of friends in the place where you live.

Similarly, when the Court claimed in Yates that "any tangible object" only meant record-keeping objects because the phrase was preceded in a list by words like "record" and "document," the Court was using noscitur a sociis to restrict the domain over which "any" extended, though "any tangible object" literally means any tangible object and definitely does not literally mean any record-keeping object.  I happen to think the Court restricted "any tangible object" in a way that no one would ever use the phrase, literally or otherwise, but I have no objection to restricting the phrase's domain in principle.  We would at least all agree, for example, that if a child's told to pick up "every object" off his bedroom floor, he isn't being told to move the furniture or search for dust mites, though "every object" most definitely includes furniture and dust mites.  To claim that the child really has been told to pick up his furniture and that the only reason he shouldn't is because his parent obviously misspoke and meant to say something like "pick up every object on your floor, except the furniture and dust mites," is quite confused.  The parent expressed herself perfectly well and explicitly exempting furniture was unnecessary; the implicit limitation on "every object," given the communicative context, was not only comprehensible, but all but unambiguous.

If these sorts of domain restriction are part of statutory meaning, not just instances where we determine that what Congress wrote was an absurd mistake, then it seems permissible to limit "a disabled veteran includes a veteran who suffers from PTSD" to "a disabled veteran includes a veteran who suffers from PTSD and is disabled by virtue of it."  We shouldn't have to prove that Congress couldn't possibly have meant what it literally said (assuming that what it literally said was that all veterans with PTSD are disabled); we should only have to show that the better reading of what the statute implies and thus means is that veterans with PTSD must be disabled, just as the Yates Court only had to show that its limited reading of "any tangible object" was the phrase's most likely reading given its context.  Likewise, even if ERISA literally says that any PPO-maintained plan is established and maintained by a church, that doesn't mean it's out of bounds to say that the statute implies, and thus means, that only church-established PPO-maintained plans are established and maintained by a church.  And again, if we can make this move, we shouldn't have to prove the statute definitely doesn't bear its literal meaning before we make it; literal meaning is just one level of meaning, and not necessarily the most important or controlling.  

Of course, perhaps ERISA in no way implies that a plan established and maintained by a church must be established by a church when it says that a plan established and maintained by a church includes a plan maintained by a PPO; requiring church establishment but allowing PPO maintenance may not be a plausible reading of the statute at all.  To say that the statute might imply a church-establishment limitation on the PPO gloss is not to say that it does.  But if we accept the claim that a definitional clause like "a 'disabled veteran' includes a former National Guardsman" literally means that all former National Guardsmen are disabled veterans, it does seem that we have to acknowledge that definitions of this sort can imply a limiting condition borrowed from the terms they define, e.g., that the disabled-veteran clause implicitly means "a disabled veteran includes a former National Guardsman who is disabled."  To claim instead that we're only sure in that case of what Congress meant to write but mistakenly didn't fails to explain why we so naturally read the clause as if it in fact ended "who is disabled."  By contrast, when Congress famously and absurdly wrote that CAFA appeals had to be filed "not less than seven days" after the appealed order's entry, one didn't naturally read the statute as if it read "not more than seven days"; it was obvious that Congress had meant to write "not more," but no one would have ever claimed that Congress implied "not more" by writing "not less."

While textualists are at times quite cognizant of the fact that literal meaning can be implicitly delimited or modulated by context, at other times—as in this opinion—they seem to believe that a statute means whatever it literally means unless what it literally means is absurd.  That need not be a tenet of textualism.  It would be more correct, and perfectly textualist, to say that a statute means what its language means unless what its language means is absurd and transparently not what Congress meant, and to say that what a statute's language means is not always coextensive with what its language literally means, given that meaning can encompass implication.  No textualist should feel compelled to invoke the absurdity doctrine to explain why "a seedless fruit includes a tomato" doesn't mean that a seedless fruit includes a tomato with seeds.  Rather, he should feel free to say that while the statute might literally mean that a seedless fruit includes any tomato (though I'm not at all sure of that), the statute implies that a tomato must be seedless to be a seedless fruit.

5.  Textualist Pathology #3: the Anti-Surplusage Obsession

One of Justice Kagan's non-question-begging arguments is her argument that if Congress had only meant to clarify that PPO-maintained plans counted as church-maintained plans, but hadn't meant to eliminate the requirement that a church plan be church-established, it would have only written that a plan maintained by a church includes a plan maintained by a PPO, not that a plan established and maintained by a church includes a plan maintained by a PPO.  This strikes me as a singularly weak invocation of the anti-surplusage canon.

Courts typically justify anti-surplusage on the theory that Congress presumably doesn't write, as Justice Kagan puts it, "stray marks on a page," and that every word they write must convey some additional non-redundant meaning.  Empirical research on congressional drafting suggests that this is dead wrong.  Of course, a survey of congressional drafters doesn't tell us much about whether in any given case a drafter avoided redundancy; it is also at least theoretically vulnerable to the objection that, even if the point of anti-surplusage is to get legislative intent right, it's possible that Congress, in reviewing its drafters' handiwork, would assume that the bill the drafters drafted isn't redundant and that "established and" does some additional work that "maintained" doesn't.  

Be that as it may, I think Justice Kagan's invocation of anti-surplusage is uniquely weak as a matter of estimating congressional intent, for the reason that it seems quite natural for Congress to repeat the whole of the definition of church plans when clarifying that a defined term doesn't exclude, and includes some, PPO-maintained plans.  Take the case of the nice black shoes.  When you ask for clarification on whether nice black shoes include loafers, what you're really asking for clarification on is whether "nice" shoes include loafers, not whether black loafers exist.  But when told to buy a "nice black shoe," you're at least as likely to ask if your friend's concept of nice black shoes includes loafers as to more precisely ask whether a loafer meets their notions of a "nice" shoe; it's simply more natural to repeat the whole term you're seeking definitional clarification on, not the particular part of the definition that needs clarifying.  

Similarly, if a statute defined and frequently used the term "disabled veteran," it wouldn't be at all surprising if Congress clarified the meaning of "veteran" by writing that a disabled veteran includes a National Guardsman rather than more precisely writing that a veteran includes a National Guardsman.  And it should equally come as no surprise that even if Congress only meant to clarify what it meant for a plan to be maintained by a church, it would copy/paste and gloss ERISA's entire definition of church plan rather than carefully reducing the definition to its most relevant constituent part and glossing that.  The idea that we would be consigning part of the statute to the status of "stray marks on a page" were we to conclude that Congress repeated a key statutory phrase when it was only trying to clarify the meaning of a part of that phrase (to be precise, the whole phrase less two words) takes anti-surplusage to rather pedantic and implausible extremes.

Now, I happen to think that anti-surplusage needn't get congressional intent right to be right.  Indeed, it seems a little incoherent to me to be a textualist on the one hand and say that one uses the canons because they accurately capture congressional intent on the other.  If the reason you use the interpretive techniques you use is because you think they get intent right, you probably shouldn't be a textualist and should resort to a more thoroughgoingly intentionalist approach altogether.  What I think justifies anti-surplusage is that people who don't know very much about congressional drafting, but who do read the statutes that regulate them, tend to assume that Congress doesn't lard the U.S. Code with redundancy.  That may be empirically false, but if it would thwart public expectations to throw anti-surplusage out, I don't think we should.  In addition to that justification for anti-surplusage, it also makes difficult and otherwise unpredictable cases relatively easy and predictable, and it is a rule against which Congress can draft if it wants to.

That said, if we're going to do anti-surplusage in order to honor public expectations, we should only take anti-surpulsage so far as the public imagines Congress takes it.  If the public wouldn't be surprised that Congress would repeat and gloss a whole definition when it only means to gloss the particular part of that definition to which the gloss is obviously germane, then there's no need to say anti-surplusage blocks interpretations to that effect.  For the reasons I just gave above, I don't think that churches would be shocked to learn that when Congress said that a plan established and maintained by a church includes a plan maintained by a PPO, it only meant to clarify that a PPO-maintained plan is a church-maintained plan and that plans established and maintained by churches include some PPO-maintained plans: the ones established by churchesThe people who respond to that interpretation by saying, "but if that's what Congress meant, why wouldn't it just say that a plan maintained by a church includes a PPO-maintained plan instead of writing the two extra words 'established and' before maintained" are, I suspect, very nearly limited to the lawyers and judges who come up with hyper-textualist moves of that ilk.

Finally, even if we justify anti-surplusage on the ground that it's some way of predictably disposing of hard cases, it seems to me a condition of that justification that anti-surplusage be a rule that Congress can practicably legislate around.  Even if courts aren't going to interpret statutes to mean what Congress intended them to mean, Congress should at least have some control over what courts interpret what they write to mean.  The more, however, we apply anti-surplusage to block extremely natural sorts of redundancy, like the repetition (or practically speaking, copying and pasting) of a whole defined term when Congress is only trying to clarify a particular ambiguous constituent part of it, the more difficult we make it for Congress to legislate against the background of anti-surplusage or anticipate how it will be applied.  It really might not occur to a drafter trying to clarify that plans established and maintained by a church include PPO-maintained plans, so long as they're church-established, that a court will think he was trying to say something much more categorical unless he omits the "established and" and only writes that a plan maintained by a church includes a PPO-maintained plan.

To tie this discussion back to contemporary textualist pathologies, anti-surplusage has always been with us.  But it strikes me as a relatively new (though quite unsurprising) development for the Supreme Court to denounce an argument that Congress conveniently repeated a definition when it really meant to gloss that definition less two words of it* as reducing part of the statute to "stray marks on a page—notations that Congress regrettably made but did not really intend."  This kind of hyper-anti-surplusage stance isn't, I hope I've shown, justifiable as a matter of textualist theory, or any theory.  But it is of a piece with several textualist tropes, or tendencies:  the denial of ambiguity and hard interpretive choice on the part of the courts and use of dubious presumptions about congressional usage to efface ambiguity and choice, and the assumption of superhuman degrees of linguistic precision on the part of Congress.  These tropes take us farther away from being good interpreters of statutory language, even as they create the appearance of rigorous fidelity to it.

* Though even this overstates the case.  Congress, on the view the Court rejected, was glossing the whole definition, and clarifying that plans maintained and established by churches do not exclude plans maintained by PPOs.  Congress, on this view, was only speaking redundantly inasmuch as it didn't mean to say that a plan is both maintained and established by a church if it's maintained by a PPO. 

No comments:

Post a Comment