In denying taxpayers’ claims for various deductions and credits offered by the tax code, courts are fond of saying that these are matters of “legislative grace.” Congress, we are told, allows a taxpayer to reduce his tax liability through sheer beneficence, and a taxpayer cannot infer a credit or deduction. Rather, his entitlement to them must be “clearly demonstrated.”

But scholarly discussions of the premium assistance tax credit in King v. Burwell seem to ignore this allegedly fundamental legislative grace canon. Why is this so?

I suspect that the unusual nature of the case provides an explanation. Scholars who lean left on policy issues, and who are more likely to support the ACA, are also the ones who support a robust legislative grace canon in other contexts. But in this context, restricting the scope of the tax credit does not comport with their policy preferences (the expansion of the premium assistance tax credit). And scholars who lean right and who may be hostile to the ACA usually do not favor the legislative grace canon, so they might hesitate to invoke it. They would likely prefer the canon that preceded it — until around the 1930s, courts tended to strictly construe the revenue laws against the government.

One might use the disappearance of the legislative grace canon to accuse both sides of hypocrisy. But unlike some of the over-the-top editorials we’ve seen in the past week, I’m not inclined to presume bad faith regarding any judge or scholar who offers an intellectual defense of her position. I think a reasonable person can believe that the King v. Burwell issue should be decided at Step One of Chevron in favor of the challengers, and I also think a reasonable person can believe that the Court should proceed to Step Two and ultimately hold for the Obama administration.

However, I think the disappearance of the canon highlights one of the most troubling aspects of strong purposivist approaches to statutory interpretation. That is, under the purposivist approach, canons that are usually followed can be easily ignored or distinguished when they do not comport with the policy that the interpreter perceives in the statute. Thus, a usually favored canon, like the legislative grace canon, will be swept aside.

I suspect that the government’s defenders would respond that there is nothing inconsistent about their approach. Even though other tax credits should be construed narrowly, this one should not. The so called three-legged stool supports expansive interpretation of Section 36B, after all.

The problem is that in describing any tax credit or deduction, one can always invoke a three-legged stool or an equally vapid metaphor. The foreign tax credit at issue in PPL v. Commissioner , decided a couple years ago, could be described as “supporting free trade, preventing double taxation, and respecting source-country autonomy.” But various prominent tax academics submitted an amicus brief urging the Court to rule against the taxpayer, because the tax credits it sought reflected an “extraordinarily generous measure” and must be “strictly construed” in favor of the IRS. The Court ruled in favor of the taxpayer 9-0.

King v. Burwell highlights another potential arbitrary use, or really non-use, of a dice-loading canon. Judge-made canons that tip the scales in favor of the government or in favor of taxpayers are necessarily troublesome. I would prefer that the courts interpret the tax laws neutrally, not purposively in favor of the preferred litigant.

I would reject the use of three-legged stools to support the tax credit and I would reject the use of the legislative grace canon to defeat it. That the 4th Circuit ignored the canon in upholding the IRS’s interpretation, but the Oklahoma district court cited it in rejecting that interpretation, only highlights the manipulability of dice-loading canons.

If nothing good comes out of the King v. Burwell litigation, as half of the case’s followers will eventually conclude, then at least we should take the opportunity to examine and completely discard rules of interpretation that are boisterously invoked when they support one’s position and freely ignored when they do not. Whatever way the Court decides King, I hope it does so with close regard to statutory text and without regard to judicially-perceived underlying purposes. I don’t yet take a position on whose side the text best supports, not having reviewed all the relevant statutory provisions, but I look forward to the merits briefs.

Andy Grewal