More on Fact Finding Tools and Statutory Interpretation through Chevron Deference (1/21/19; 1/25/19)

The Chevron Framework that is so ubiquitous in administrative law now (see e.g., Law Finding (The Chevron Framework) and Fact Finding in Trials (Federal Tax Procedure Blog 1/19/19), here) offers another interesting relationship to fact finding.

A standard formulation of the preponderance of the evidence fact finding standard is that, if the fact finder is in "equipoise" as to the existence of the fact, the party bearing the burden of persuasion loses on that fact.  Equipoise is that point (stated in percentages at 50% belief in the existence or nonexistence of the fact) where the trier cannot decide with the comfort level of more likely than not (in percentages greater than 50% (existence) less than 50% (nonexistence)).

That same phenomenon conceptually occurs in determining a proper interpretation of a statute.  If there is more than one reasonable interpretation of the statute, then presumably the most reasonable interpretation applies.  The most reasonable interpretation could be the one that is more likely than not the correct one.  In the percentages, it is the interpretation that the court is persuaded to a level greater than 50%.  But the most reasonable interpretation can mean something less than 50% if there is more than two reasonable interpretations of the statute.  Then, presumably, the most reasonable interpretation could be the one at a 40% level if the other two are at 30%.  But, even though not the more likely than not interpretation, a court has to pick one interpretation and would, presumably, pick the 40% level.

But what does a court do when the interpretations are 50-50 with no one of them more likely than not?  I don't have an answer to that question, but I do want to pursue the question in the context of an agency interpretation in a regulation.  The court would then apply the Chevron Framework (based on Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984) and its progeny).

Chevron deference would mean, theoretically, if the agency in the above three interpretation choice example, chose a 30% interpretation, the court should defer to it (assuming 30% is within the range of reason even though it believed the 40% interpretation was the most reasonable (actually I would say was the more reasonable)).

Now, going back to the two interpretation choice example, Chevron would say that, if the Choice 1 were 60% and Choice 2 were 40% and the agency interpretation was Choice 2, then the agency interpretation controls (provided the court agrees that Choice 2 is at least a reasonable interpretation).  Perforce, that would mean that if the interpretations are equally persuasive (50% for each, the state of equipoise under the fact finding analogy), then the agency interpretation governs.

I recently picked up this on Chevron and equipoise (Paul A. Larkin, Jr., Reawakening the Congressional Review Act, 41 Harv. J.L. & Pub. Pol'y 187, 209 (2018), here, in discussing judicial review of agency interpretations:

[T]he controversy is biased in the agency's favor. Chevron and other Supreme Court decisions place a thumb on the government's side of the scale when it comes to the meaning of federal law, with the agency winning when it has the better of the argument and when courts find themselves in equipoise.

Notice that the author says Chevron applies (i) when the agency has the better (in percentages more than 50%) or is in equipoise (50%-50%).  As I suggest above, the way I read Chevron is that Chevron deference to the agency interpretation may apply in a third category -- when an agency interpretation is not the better and there is no equipoise, so long as the agency interpretation is reasonable.

Indeed, to work this further, if the agency interpretation is better (in the court's mind), then deference means nothing because that is the interpretation the court would have chosen anyway.  So, deference is meaningful only in the equipoise situation (a rare occurrence in fact-finding and, I think, equally rare in law finding) and where the agency interpretation is not the better interpretation but is reasonable.

Addendum 1/25/19: 

I developed the foregoing in terms of the Chevron Step Two inquiry which asks whether the agency interpretation is a reasonable interpretation entitled to deference.  In revising my article, I found the following which uses the equipoise concept at the Chevron Step One inquiry which asks whether the statutory text is ambiguous (at least ambiguous within the scope addressed in the agency interpretation).  As I state in the article, the question of whether the statute is ambiguous is a bit like Justice Potter Stewart's definition of pornography -- "\I know it when I see it."  Jacobellis v. Ohio, 378 U.S. 184, 197 n. 2  (1964) (concurring opinion).

On the issue of ambiguity, I then state:

While I don’t expect a court to articulate a trial fact finding-type analog with percentages of confidence, a noted commentator, Cass Sunstein,  recently presented the following as possible ways of conceptualizing the test: 

1.  A statute counts as ambiguous if and only if courts are in genuine equipoise after they have used the appropriate tools of statutory construction. Judges simply do not know the right answer; it is a 50-50 call. If that is the proper understanding of equipoise, there is a fair question: How often does that happen? Some judges think that it is fairly common; some judges think that it is pretty rare. 

2. A statute counts as ambiguous whenever there are objectively reasonable arguments both ways (that is, arguments that reasonable people hold in good faith). Under this view, a court would defer to an agency unless it is nearly certain that the agency is wrong – certain, say, with at least 90 percent confidence. If that is the proper understanding of Chevron, we can understand Judge Kavanaugh’s use of the word “amazing.” If all judges on a three-judge panel are 89 percent sure that an agency’s interpretation is wrong, and if the agency wins anyway, then Chevron is indeed a significant shift in interpretive authority. 

3. A statute counts as ambiguous unless courts think, with a fair degree of confidence, that one interpretation is best – say, 65 percent confidence. This approach would grant a large degree of discretion to agencies, but it would not exactly count as amazing. After all, judges are deferring when they think that agencies are 35 percent likely to be right, and 35 percent is not so low.  [footnote]

[Footnote]
Cass R. Sunstein, Chevron as Law 10 (Preliminary Draft of 8/3/18 (marked “provisional, incomplete and subject to revision) (footnotes omitted) (available at SSRN: https://ssrn.com/abstract=3225880).  I do note that the starting point in paragraph 1–the state of equipoise–is frequently used to describe fact finding under the preponderance of the evidence standard.  A fact-finder, whether jury or judge, should find the fact or not find the fact unless he is at the 50-50 point; if the fact-finder believes that the likelihood of the the fact is more than 50% (say even just 50.01%), then the fact is found to exist and the bearer of the burden prevails, but if the fact-finder believes that the likelihood of the fact just 49.99% (or less), then the fact finder finds that the fact did not exist and the bearer of the burden loses on that fact.  Sunstein seems to offer a similar conceptual analysis.  If I am right on that, then, I am intrigued by his last statement under par. 1 that some judges think the state of equipoise is “fairly common,” but some other judges think it is “pretty rare.”  In the accompanying footnote for that statement, Sunstein cites only Justice Scalia’s comments indicating that Justice Scalia and other like-minded judge would rarely find ambiguity as thus formulated; he cites no authority for the statement that some judges find the state of equipoise “fairly common.”  Going back to the trial evidence analog, my understanding is that fact-finders rarely find a state of equipoise rarely.  I reached this understanding both from my many years of trial practice and my research which I have incorporated into my annual Federal Tax Procedure Book, John A. Townsend, Federal Tax Procedure (2018 Practitioner Ed.) 619 n. 2424 (July 23, 2018) (available at SSRN: https://ssrn.com/abstract=3218532): Schaffer v. Weast, 546 U.S. 49 (2009) (“In truth, however, very few cases will be in evidentiary equipoise.”); see also Justice Ginsburg’s dissent in Schaffer, p. 68 (“And judges rarely hesitate to weigh evidence, even highly technical evidence, and to decide a matter on the merits, even when the case is a close one. Thus, cases in which an administrative law judge (ALJ) finds the evidence in precise equipoise should be few and far between.”); Cigaran v. Heston, 159 F.3d 355, 357 (8th Cir. 1998) (“The shifting of an evidentiary burden of preponderance is of practical consequence only in the rare event of an evidentiary tie . . . .”); see also Polack v. Commissioner, 366 F.3d 608, 613 (8th Cir. 2003) (citing the Cigaran case).  See also Neil Buchanan, The Burden of Proof and Tax Law: Deja Vu Silliness (Dorf on Law Blog 6/14/13), where Professor Buchanan notes that, although it is conceptually conceivable that there might be a 50-50 case where outcome is determined by the assignment of the burden of persuasion:
In the real world, however, it is never that close (in tax cases, or in any other civil case, as my CivPro-teaching colleagues can attest).  In fact, a study in 2008 (ten years after RRA98) showed that shifting the burden of proof under the 50%-plus-a-tiny-amount standard simply makes no difference in tax cases.  The outcome is the same, no matter who formally bears the burden of proof.
Also, finally, note the concept in both Sunstein’s analysis as the standard trial fact-finding analysis is that it implies some risk that the process will reach the wrong conclusion.  Let’s say that the judge is 49% sure that his finding of no ambiguity is right, that implies some considerable risk (say 51%) that he is wrong.  Even where the threshold burdens are higher (as in paragraphs 2 and 3), there is still some significant risk that the judge may be wrong and that the statute is really ambiguous.  Of course the higher percentage thresholds would mitigate the risk, but it would not eliminate the risk.