The Impact of Chevron Deference is Exaggerated (8/19/21; 8/21/21)

I recently wrote on Judge Lauber’s Opinion in Lissack v. United States, 157 T.C. ___, No. 7 (2021), here.  See Tax Court Holds that Collected Proceeds for Whistleblower Awards under § 7623(b) Do Not Include Unrelated Collections (8/18/21; 8/19/21), here. I discussed the Court’s discussion regarding Chevron deference.  Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984).  Chevron established what is now referred to as Chevron deference (although deference had been in the law for a long time, with Chevron provided only a regular structure to apply deference).  I believe that Chevron is much cited and much misunderstood.  I pointed out in the blog that Judge Lauber stumbled into Chevron but finally got the analysis correct by testing the interpretive regulation for reasonableness of the interpretation within the scope of the ambiguous statutory text.  I also said in the blog (comment 5) that I would be writing another separate blog about a key aspect of the Chevron jurisprudence that is also misunderstood – when does a court really defer under Chevron?  As I said, by not clearly stating what they are doing, courts overstate the outcome-determinative effect of Chevron deference.  I write now to further explain.  (I attach here a pdf of the current state of the working draft for my article to be published on SSRN later this month (hopefully).)

First, here is my crisp statement of outcome-determinative deference:

Deference is a court applying a reasonable agency interpretation of ambiguous statutory text despite the court’s belief that there is a more reasonable interpretation of the ambiguous statutory text.  That’s it.

Second, inherent in that definition of deference is a very limited domain for outcome-determinative Chevron deference, far more limited than the commotion, particularly among conservatives and libertarians, support.  Consider these categories of judicial interpretation related to agency interpretations (note that I develop this in the context of agency interpretive regulations doing no more than interpreting statutory text):

            1.        If the statute is unambiguous, the statute controls without either the court or the agency interpreting it further.  No deference there.  

All further Categories assume that the statute is ambiguous:

            2.         If the agency has not interpreted the statute, then the court interprets it.  No deference there. 

            3.        If the agency has interpreted the statute, and the court believes the agency interpretation is the best interpretation, then that best interpretation applies. Since the court reached that same interpretation, the court did not defer to the agency interpretation.  No deference there.

            4.        If the agency has interpreted the statute. the court believes a better interpretation applies, and the court believes the agency interpretation is unreasonable, then the court applies its interpretation rather than the agency interpretation.  No deference there.

            5.        If, in the same circumstances as Category 4 except that the agency interpretation is reasonable but not the most reasonable interpretation (in the court’s mind), and the court applies the agency interpretation, then the court has deferred to the less reasonable agency interpretation.  Deference there.  (This Fifth Category is what is now called Chevron deference but similar deference was accorded before Chevron.)

            6.        If the agency interpretation and at least one other interpretation are reasonable within the scope of the ambiguity, and the court applies the agency interpretation without determining which interpretation among the reasonable interpretations is the best, then the court has not necessarily deferred in the sense that it applied an agency interpretation instead of its own best interpretation.  Category 6 is ambiguous as to whether there is deference.  The court’s holding does not tell whether the court deferred to a less reasonable agency interpretation.  Only in Category 5 does the court defer to a less reasonable agency interpretation.  In this Category 6, the agency interpretation could be the best interpretation (a Category 3 case) in which there is no deference. Or it could be a Category 5 case if the court had actually determined the best interpretation and applied the agency less reasonable interpretation instead.  Category 6 may look like a Chevron deference case; it is not because Chevron is outcome determinative only if the court applies (defers to) an agency interpretation instead of its own better and more reasonable interpretation (Category 5).  I caution readers to be aware of this Category 6-type case because treating this category as Chevron deference cases exaggerates Chevron’s true impact, which is only in Category 5.  (I add my belief that true Category 5 cases are far fewer than the ideologically and politically heated commotion about Chevron warrants.)

Readers with me so far, particularly trial lawyers, will spot a possible problem.  Chevron deference depends upon a court being able to find its own most reasonable interpretation that is different from the agency’s less reasonable interpretation.  Trial lawyers will be familiar with burden of proof concepts where the burden of persuasion requires the bearer of the burden (often a plaintiff) to lose in a state of equipoise – where the trier of fact is unable to find that a key fact is or is not more likely than not.  Taking that concept of equipoise into the Chevron persuasiveness of interpretation analysis, consider what happens where the judge is unable to determine whether any interpretation is the most reasonable, meaning that there are competing reasonable interpretations with the same degree of persuasiveness thus preventing the judge from determining that either the agency interpretation or any other interpretation is the most reasonable.  One easy answer to the concern is that, like with equipoise as to facts, equipoise as to legal interpretations would be rare, indeed more rare.  But just imagine that it could happen.  Then, in the fact-finding context, the burden of persuasion provides a default tie-breaker result. 

What does the judge do when he is in equipoise as to the legal interpretation?  Does he flip a coin, consult the auguries, or invoke other forms of arbitrary tie-breaker rules?  Is that the zone in which Chevron is intended to provide the tie-breaker for the agency interpretation over an equally persuasive other interpretation?  If so, Chevron has very limited effect because the equipoise or practical equipoise phenomenon for legal interpretations is very rare.

Now, I suppose the Chevron deference issue is set up where the judge determines that the best interpretation is an interpretation other than the agency interpretation, but it is only marginally better than the agency interpretation.  Let’s say in terms sometimes used for burden of persuasion analysis, the judge believes to a level of 51% persuasion that its own interpretation is best but that the agency interpretation is just slightly less persuasive, in this case 49%.  Of course, this thin an edge is unlikely to convince a judge that he is not in a state of practical equipoise.  Indeed, I think if percentage points were added or subtracted as appropriate, say to get to 55%-45%, the judge would likely be in a state of practical equipoise.

 Perhaps, though, we could posit a case where the judge is persuaded 67% for his own interpretation and is persuaded that the agency interpretation is 33%.  Now that is a difference that a judge can perceive.  Would any judge apply the agency interpretation in this case?  I suspect not.  The judge could determine at Chevron Step One that the judge’s own interpretation is the plain meaning of the statute because the competing agency interpretation is just not enough within the reasonable scope of the statute’s ambiguity (33% is not enough to be reasonable).  Or the judge could determine at Step Two that the agency interpretation is unreasonable.  (This phenomenon has been why some conflate the two Steps.)  Either way, the judge’s interpretation trumps the agency interpretation, without any deference being accorded to the agency interpretation.

So the analysis must focus in-between 67-33 and 55-45.  Posit a judge who is persuaded that his own interpretation is 60% most persuasive and the agency’s interpretation is only 40% persuasive.  Will a judge really defer to an agency interpretation in that situation?  I suspect not.  Chevron has the tools to prevent deference.  I suspect the judge will invoke them.

Based on the foregoing analysis, I think one improvement (perhaps an interim one) in Chevron jurisprudence would be to require courts proceeding to Chevron Step Two to articulate first which legal interpretation they think is the best interpretation (courts can stop there is the agency interpretation is also that best interpretation), to articulate whether the less reasonable agency interpretation is reasonable (if unreasonable, they can stop without deference), and then stating with analysis the relative persuasiveness of each interpretation.  I suspect that a court going through this drill will find his own interpretation sufficiently stronger than the agency interpretation so that the agency interpretation is not sufficiently reasonable to require deference.

Added 8/20/21 11:00 am:

I thought it might be useful to readers not deep in the weeds on deference to test my general speculation above that, when the hypothetical judge does the work to determine the best interpretation of the statute, it would not be common that the judge will then actually defer to a less persuasive agency interpretation.  Let’s focus on Judge Lauber’s opinion in the case that started this discussion,  Lissack v. United States, 157 T.C. ___, No. 7 (2021), here.

I direct readers attention to the statutory interpretation discussion (pp. 13-23).  Judge Lauber concludes the Chevron discussion (p. 21):

            In sum, we conclude that the regulatory provisions at issue, taken together, are not “arbitrary, capricious, or manifestly contrary to the statute.” Chevron, 467 U.S. at 844. Treasury reasonably determined that a multiple-issue IRS examination could comprise more than one “administrative action.” And it reasonably concluded that the IRS does not “proceed based on” the whistleblower’s information unless that information substantially contributes to the “administrative action” that generates proceeds. We accordingly reject petitioner’s challenge to the regulation’s validity.

Judge Lauber does not say that the regulations interpretation is not as good as his own interpretation and is nevertheless deferring to the regulations interpretation.  Indeed, he does not say that he has a different, better interpretation than the regulations interpretation.  I think if you parse what he says otherwise (pp. 18-21), Judge Lauber was satisfied with the agency interpretation and not just because it was a less-best reasonable agency interpretation.  Judge Lauber expresses no concern about the agency interpretation.  In this case, he does chant Chevron Step Two.  But his actual discussion of the merits of the interpretation at least suggests that he thinks it is the best interpretation of the ambiguous statute.  If I am correct on that, this is a Category Three case in disguise where there is no deference to the agency interpretation.  Cf. Oakbrook Land Holdings, LLC v. Commissioner, T.C. Memo. 2020-54, * 25, here (where, in the regulations ambiguity Auer deference analog to statutory ambiguity, the Court said there is no need for deference where the “traditional tools of construction” on de novo review support the IRS interpretation without deference).

I would appreciate hearing from readers with views on this issue.  Please respond either by comment or by email.  Thank you.

Added 8/21/21 12:00pm:

In my addition to this blog entry yesterday, I speculated that Judge Lauber may have been persuaded that the Treasury / IRS regulations interpretation was the best interpretation.  If that were correct, then the case does not involve Chevron deference because Judge Lauber would be applying the best interpretation, albeit also the agency interpretation.  In other words, he would not be deferring to the agency interpretation.  Had there been no agency interpretation, Judge Lauber would have applied that interpretation on de novo review.  In reflecting on that overnight, I thought there might be a different model to consider this issue.  That is the equipoise and line-drawing model.

For many “interpretations” of ambiguous statutory text, the interpretation requires line-drawing.  United States v. Correll, 389 U.S. 299 (1967) is a good example.  The statute allowed trade or business deductions for “away from home” meals and lodging expenses.  In an interpretive regulation, Treasury interpreted the statutory text to require stopping for sleep or rest, an overnight rule.  Was that the only reasonable interpretation within the scope of the statutory text that the expenses be incurred “away from home?”  Surely not.  Some administrable line was needed.   I won’t speculate about the possible lines via interpretation that could be drawn within the scope of the “away from home” text.  Treasury chose one line among other possibilities. Among those possibilities, the one Treasury chose might be in a sense arbitrary – no line was more persuasive than the other, except that, for Treasury, this line was conducive to administration.  By conducive, I don’t mean that Treasury meant to take something away from taxpayers; Treasury meant to make the statutory text more administrable for taxpayers and the IRS.  Keep in mind that taxpayers have to plan and report their away from home expenses and needed a line that supported their compliance; so too the IRS needs a line to support their compliance initiatives.  Perhaps for Chevron purposes, that administrability consideration would mean that the interpretation was the most persuasive, particularly under a Skidmore analysis.  But what if at least one other interpretation had arguably similar administrability features so that there is equipoise between or among the interpretations and choosing between those interpretations would be arbitrary?

I ask but can’t answer (I will continue to reflect on it) whether that equipoise is the space in which the default position is the agency line-drawing “interpretations.”

I develop some thoughts on equipoise in a Chevron context in the excerpt linked at the beginning of this blog entry at p. 74 n. 266I need to think more about equipoise in interpretation in a Chevron context.

Any comments from readers will be appreciated.