This post’s about a puzzling opinion from the D.C. Circuit. The puzzle has to do with the Chevron two step and legislative history. This puzzle’s important, and not just for King v. Burwell.

In Council for Urological Interests v. Burwell (CUI), published last week and available here, the D.C. Circuit shuffled between one view and another of legislative history’s role in the Chevron analysis. On the first step, the court of appeals held that a Conference Report didn’t foreclose an agency’s rule. On the second step, however, the court held the same Conference Report required a remand of the rulemaking. (For those unfamiliar with the Chevron two step, check out this discussion or this much more entertaining video.) The flaw, the court held, was not in the agency’s reading of the statute’s text but rather in its “tortured reading” of the Conference Report. There’s something in the case for everyone who has debated whether Chevron has one step or two and what role, if any, legislative history should play in those steps.

In CUI, an association of physicians challenged regulations promulgated by the Secretary of Health and Human Services that “effectively prohibit[ed] physicians who lease medical equipment to hospitals from referring their Medicare patients to these same hospitals for outpatient care involving that equipment.” The court of appeals remanded one of the regulations to the Secretary with instructions to “grapple with [a] Conference Report” that might be read to preclude the Secretary’s view of the statute.

Interestingly, the panel members shuffled their votes between Chevron‘s two steps. At step one, one majority of the panel (Judges Griffith and Rogers) rejected the physicians’ argument that the Conference Report foreclosed the agency’s reading of the statute. At step two, however, a different majority (Judges Henderson and Rogers) remanded the rulemaking because of the Conference Report. Judge Henderson, who would have decided the case at step one, wrote the step two portion of the opinion. Of course, parts of her step two opinion read like step one analysis. For those who think Chevron has only one step, that’s not surprising. See Matthew C. Stephenson & Adrian Vermeule, Chevron Has Only One Step, 95 Va. L. Rev. 593 (2009).

The case also has something for those who think Chevron has two steps. See, e.g., Richard M. Re,Should Chevron Have Two Steps?, 89 Ind. L.J. 605 (2014); Kenneth A. Bamberger & Peter L. Strauss,Chevron‘s Two Steps, 95 Va. L. Rev. 611 (2009). Kenneth Bamberger and Peter Strauss have argued that step two entails “judicial oversight” of the “agency decisionmaking process,” with the court asking whether the agency “has appropriately taken account of other interpretive tools — like normative canons of construction or legislative history — when resolving ambiguity in the governing statute.” Bamberger & Strauss, supra, at 623. Parts of Judge Henderson’s step two analysis in CUI can be read in these terms. The agency had addressed the Conference Report in the rulemaking, but, Judge Henderson explained, its “jargon [was] plainly not a reasonable attempt to grapple with the Conference Report.” The court therefore remanded and instructed the Secretary to reconsider whether her interpretation of the statute “is consistent with the . . . Conference Report.” In other words, the agency failed reasonably to resolve statutory ambiguity, or so Judges Henderson and Rogers held.

By contrast, Judge Griffith thought the agency’s briefs saved its rulemaking. Citing Chenery, Judges Henderson and Rogers concluded they couldn’t rely on the agency’s litigating position. Judge Griffith replied that the physicians forfeited their Chenery argument by not raising it until the reply. So the case has something for those who think Chenery‘s founded in the nondelegation doctrine and “states a necessary condition for Chevron deference.” Kevin M. Stack, The Constitutional Foundations ofChenery, 116 Yale L.J. 952, 1006 (2007).

The case also has something for those who think courts should stop dancing the Chevron two step. See, e.g., Jack M. Beerman, End the Failed Chevron Experiment Now: How Chevron Has Failed and Why it Can and Should Be Overruled, 42 Conn. L. Rev. 779 (2010). There’s confusion in the competing opinions. Judge Henderson, for example, would have used legislative history to foreclose deference to the agency at step one. Her reason? “In the Chevron context, . . . a failure to consult legislative history would leave the text ambiguous and thereby transfer authority to an administrative agency, whose democratic accountability is nil.” Really, nil? Zip, zilch, nada? Not even a little bit? And if the concern’s accountability, then why use legislative history to transfer authority to an Article III court? Doesn’t that miss an important reason courts do the Chevron two step?

Judge Henderson’s reasoning thus tees up John Manning’s critique of the use of legislative history in the Chevron analysis. He argues that using legislative history to narrow an agency’s discretion “necessarily alters the scope of — and thereby contradicts — a delegation made by the relevant statute to the agency.” John F. Manning, Chevron and Legislative History, 82 Geo. Wash. L. Rev. 1517, 1521 (2014).

My own view is that CUI‘s about the agency’s changing interpretations of legislative history, not legislative history itself. As far as I can tell, the agency had one view of the Conference Report in the 2000s, another when it promulgated the rule, and a third in its briefing to the court. Unexplained changes in position and post hoc rationalizations can’t make a reviewing court all that confident about the agency’s decisionmaking. Chevron, however, seemed to frustrate rather than to facilitate the court’s analysis of that problem and its directions to the agency on remand. Apparently the agency needs to “grapple with the Conference Report.” How much of a struggle will be enough? Would it be enough, for instance, if the agency were to conclude that legislative history’s unreliable and thus irrelevant? Or does CUI suggest an agency has a special obligation to use legislative material when interpreting statutes? See Jerry L. Mashaw, Norms, Practices, and the Paradox of Deference: A Preliminary Inquiry into Agency Statutory Interpretation, 57 Admin. L. Rev. 501, 513 (2005).

It would be helpful, as Bradley Silverman has pointed out in connection with King v. Burwell, if federal courts specified how to dance the Chevron two step with legislative history in hand. But I wouldn’t be surprised if the doctrine just keeps shuffling from one foot to the other.