Procedurally Taxing has been at the forefront covering the evolution of the “supervisory approval” requirements of IRC § 6751 (see posts here, here and here). The most recent post from Patrick Murray, a University of Minnesota Law student, argued that in its present, post-Walquist form, IRC § 6751 is of little help to low-income taxpayers. In a nutshell, this is because the IRS no longer needs supervisory approval for “automated” IRC § 6662(b)(2) penalties and the vast majority of low-income taxpayers experience fully automated examinations.
Mr. Murray’s post advocated for a Congressional fix, mandating that all IRC § 6662 accuracy penalties be subject to supervisory approval requirements as a way to level the playing field. I’ll reserve judgment on that, but instead ask whether a Congressional fix is actually needed for a level playing field. A scintillating footnote from a recent Tax Court opinion may reveal battle lines already being drawn on that issue.
read more…First off, my bet is a fair number of readers didn’t roll their eyes or do a double-take at the preceding phrase “scintillating footnote from a recent Tax Court opinion.” Please take a moment to reflect on that, as I have. Ok. Now let’s get into the fun stuff: a possible changing of the guard in IRC § 6751 supervisory approval requirements.
How do we get there? Through the idyllic (I’m assuming) town of Plentywood, Montana, its lone pharmacy and the case of Plentywood Drug, Inc. v. C.I.R., T.C. Memo. 2021-45. It is here we find that scintillating footnote I promised. Specifically, footnote 10 which provides:
“The Commissioner argues that the penalty for substantial understatement of tax at section 6662(b)(2) falls within the automatic-computation exception to section 6751. Because Plentywood Drug is the only party alleged to have under-stated its income tax and failed to carry its burden of production for the section 6751 issue, the Commissioner’s argument is moot and we will not address it.”
So we have a footnote, in a non-precedential memo opinion, where the Tax Court basically says “we aren’t touching that argument because it is moot.” You may fairly ask, “is that really ‘scintillating’?”
Yes, I’d say it is.
It is scintillating because it (potentially) represents the IRS taking a litigating position that would vastly expand the reach of Walquist. If the Plentywood examination was not fully automated this litigating position is a big deal because it takes the “automation” factor out of the equation. And there is at least some reason to believe the Plentywood exam was not automated. For one, it is a corporate taxpayer being audited on a Fair Market Valuation question -not the usual automation fodder of “our records show you have one more W2 than listed on your tax return, please explain.” For two, based on the information I was able to glean from the consolidated Tax Court dockets (Dkt. #s 17753 through 17755-16), the penalties as applied to the individual taxpayers did require (and have) proof of supervisory approval, and therefore were not automated. To me, it would be odd if the penalties against the corporation were automated but the penalties against the individuals were not.
So let’s assume this was not an automated exam. What would that mean? In essence, it would mean that the litigating position of the IRS appears to be that every “substantial understatement” penalty is “automatically computed” (and thus meets the exception of IRC § 6751(b)(2)) regardless of how the examination is initiated. In other words, Walquist would stand for “no supervisory approval needed for purely mathematical penalties, of which IRC § 6662(b)(2) is one example -regardless of whether it is a computer or a human that punches the numbers into the calculator.”
That would indeed be a step towards leveling the playing field between rich taxpayers and low-income taxpayers… by making things worse for everyone. All things considered, I’m basically fine with that.
My biggest complaint heretofore has been that under Walquist two identical taxpayers making identical mistakes, subject to identical IRC § 6662(b)(2) penalties are given different protections depending on how the IRS initiates the examination. You can see this played out, and this point made, in Bryan Camp’s post here. I think it is important that this issue be brought to the court’s attention, and especially if the taxpayer is represented.
As has been noted by both Judge Holmes in the Graev opinion itself and Professor Bryan Camp (here among other places), the statutory language of IRC § 6751 is problematic. I don’t envy the Tax Court’s role in parsing out the “real” meaning and requirements of an almost nonsensical statute -where ambiguity lies, a good lawyer will pick apart every word and nuance to advocate for their client.
And so the Tax Court has had to weigh in on timing issues (here), evidentiary issues (here), the definition of what a “penalty” is (here), and many other creative or at least plausible arguments put forth by zealous advocates.
And then you get Walquist. That case had zealots, but I’m not sure I would call them zealous advocates. The petitioners were pro se and advanced a raft of nonsense tax-protestor arguments that the Court dismissed. Maybe the outcome would have been the same with a lawyer, but I believe at the very least with competent counsel the contours of the decision would have been different. Perhaps if the IRS continues to raise the arguments it did in Plentywood we just may see that happen.
One issue I would like to see raised by counsel for taxpayers is the bizarre policy effects of Walquist. If an exam is fully automated fewer protections are offered to the taxpayer. Fully automated exams are generally limited to relatively low-dollar issues. And these issues, in turn, tend to involve middle-and-lower income taxpayers. In a nutshell, as I tell my students, high-income taxpayers get agents and low-income taxpayers get automation. And since automated penalties don’t require supervisory approval, low-income taxpayers therefore don’t get supervisory approval prior to getting hit with penalties.
To me, this has the perverse effect of protecting most those who need it least.
Pause to think about this, for a moment, while I step down from my soapbox. Accuracy penalties are largely an automatic default for the least sophisticated taxpayers, while the most sophisticated get layers of review and contemplation before determining (maybe, not conclusively) a penalty should apply. If IRC § 6662(b)(2) penalties are really just a matter of math, why should that be the case? Why should there be any discretion at all, automated exam or not, as to whether IRC § 6662(b)(2) applies when the math says it does? There are some hiccups in the statutory language of IRC § 6751(b)(2) that would cause some issues: that the penalties must be calculated “through electronic means” seems to imply that a computer is running the whole show. But as with so many other aspects of that troublesome code section, there is room for debate.
Someday I may write about (or litigate) why I don’t think IRC § 6662(b)(2) is a “pure math” question amenable to automation. But for now, I am happy that Plentywood brings that issue to the forefront, and hopeful that it arises again.
About Caleb Smith
If a human inputs numbers into a computer’s calculator program and tells the program what operations to perform on the numbers she inputted, the computation is manual.
If a human inputs numbers into a computer’s database system and tells the database system what operations to perform on the numbers stored in the database, the computation is automated.
Disclaimer: As my profession is programmer not lawyer, the above explanation is NOT my professional opinion and is NOT my personal opinion.
It’s an interesting issue. There could be a lot of differentiation, e.g.
Negligence penalty should be pretty easy, because it depends on finding *by a human* is necessary to know whether it applies. Same with a substantial understatement penalty for which the taxpayer responded claiming a defense against the penalty. Human decision required = supervisoruy approval required.
For substantial understatement penalty, the initial determination is purely mechanical: Is it above $X? But there can be a defense that would eliminate the penalty. That’s not purely mechanical and would require a finding *by a human* to decide that it doesn’ty apply. So:
– Is supervisory approval required if the taxpayer doesn’t respond? Maybe, maybe not.
– What if the taxpayer responds but doesn’t raise the penalty defense? Maybe, maybe not.
In either of these cases, could the approval be in advance *for all such cases with no response that otherwise meet the requirement* – approval for all cases meeting specific guidelines, versus approval on a case by case basis? That feels inconsistent with the underlying idea of the approval, to make sure the penalty is not being asserted for an improper purpose. But in those cases where the computer initiated the penalty assertion, was there an improper purpose? A person didn’t make the decision for that specific case, but a person did decide how the system would be programmed. Can you have an “improper purpose” at the programming level? Maybe, maybe not.
Fun times!
But, yes, absolutely: taxpayer rights are abstract, but the application is concrete, and they do not apply to all equally. When they don’t, it’s the (unrepresented) low income taxpayer who gets the weaker version and the rich taxpayer who gets the stronger version.
.
‘ Can you have an “improper purpose” at the programming level? Maybe, maybe not.’
Yes you can, and there’s no maybe about it.
Anyway, the calculation depends on numbers that were input into the system. You can have an “improper purpose” at the input level. At least one IRS employee was jailed for it. And according to a LinkedIn page, the employee was rehired by the IRS.
Note that the Greenbook proposes to eliminate the requirement for written approval of penalties under sections 6662, 6662A and 6663. See https://home.treasury.gov/system/files/131/General-Explanations-FY2022.pdf, page 100.