Last April, I wrote (see here and here and here) about the Federal Circuit’s decision in General Mills, Inc. v. United States, 957 F.3d 1275 (Fed. Cir. 2020), aff’g 123 Fed. Cl. 576 (2015). The briefs on appeal are available here: the taxpayer’s opening brief, the government’s answering brief, and the taxpayer’s reply brief. This was a complex case, involving the intersection of TEFRA partnership audit procedures and the “large corporate underpayment” (LCU interest, or “hot interest”) provisions of the Code.
There was one argument made by the plaintiff on the TEFRA issue that I addressed rather briefly, because the Federal Circuit simply swatted it away. That argument really warrants more discussion because it’s a serious problem that goes beyond just this specific context, one that we need to continue challenging: the adequacy of notices to taxpayers. So let’s talk about that case a bit more.
The Background
This is a very abbreviated version of the facts, because I want to focus on that one argument General Mills raised and the court rather perfunctorily swatted down. Interested in more? Check the opinions or prior blog posts.
The General Mills consolidated group (“GMI”) filed corporate tax returns for the 2002-2003 and 2004-2006 tax years, and General Mills Cereals, LLC (“Cereals”) filed partnership tax returns for the same years. Various members of the GMI consolidated group were partners in Cereals, so the tax returns—and any audit adjustments—for Cereals flowed through to GMI. The IRS audited both sets of tax returns for all these years. Although there are some slight differences between the audits for those two periods, for simplicity I will focus on the 2002-2003 tax returns.
Deficiencies were identified for both the corporate and the partnership tax returns. As this was back in the days before the Bipartisan Budget Act of 2015, the TEFRA audit procedures were still in effect. For the remainder of this post, references to Code sections in the 62xx range will be the old, pre-BBA versions, which have been replaced in the current Code by the new BBA procedures.
TEFRA “Computational Adjustments”
TEFRA required “partnership items” to be resolved at a partnership-level proceeding. After that proceeding ended, the effects of the adjustments to partnership items were translated into partner-level tax liabilities by means of “computational adjustments.” Notices of deficiency at the partner level are required only if a partner-level “determination” is required; otherwise the IRS can assess immediately. (Penalties can be assessed immediately even though a partner might raise partner-level defenses in a refund claim/suit.) Immediately assessable computational adjustments also include “any interest due with respect to any underpayment or overpayment of tax attributable to adjustments to reflect properly the treatment of partnership items.” Treas. Reg. § 301.6231(a)(6)-1(b).
TEFRA also provided a mechanism for taxpayers to challenge computational adjustments that did not require a deficiency proceeding, in section 6230(c). The taxpayer must file a refund claim “within 6 months after the day on which the Secretary mails the notice of computational adjustment to the partner.” Thereafter, a refund suit can be brought within the period specified in section 6532(b) for refund suits. Section 6511(g) provides that section 6230(c) applies, rather than the two-year period specified in section 6511(a), with respect to tax attributable to partnership items.
You guessed it. The partners (individual members of the GMI consolidated group) paid the assessed liabilities and then filed refund claims. Those refund claims were timely under the two-year period in section 6511, but not under the six-month period in section 6230(c). So GMI had to argue that the longer period applied.
There were a lot of different arguments by GMI that I discussed at length last April. Here, I want to focus on GMI’s argument that the six-month period never commenced, because none of the communications from the IRS qualified as a “notice of computational adjustment.”
A Brief Pause to Consider Other Types of Notices
A notice of deficiency clearly identifies the tax year, the amount of the deficiency, and the reasons for the deficiency. The IRS sends it by certified mail. It clearly identifies what the taxpayer must do to challenge it – file a petition with the Tax Court – and the period within which the taxpayer must do that. It even calculates the deadline and states it explicitly on the first page, and the taxpayer can rely on that even if the IRS erroneously calculates the date. Stat notices are not without their flaws, but this is about as good as you could hope for – the “gold standard.”
A notice of determination denying innocent spouse relief can be pretty good too. The most recent we received, by Letter 3288, was sent by certified mail and clearly identified the tax year(s) at issue and a (too) brief explanation of why relief was denied. (A request for the administrative file is definitely in order!) It clearly specified, at the top of page two, the taxpayer’s right to challenge the determination by filing a petition in Tax Court and the period within which the taxpayer must do that.
A notice of intent to levy is, frankly, horrible. They’re too long and poorly written, hard to understand for a layperson; even for some students at an academic LITC. They muddle the message by combining requests to pay with notifying the taxpayer of her rights. The next-to-last notice (CP504) is misleading, as discussed below. The final notice (LT11, CP90, etc.) explains the right to a CDP hearing reasonably well, including the period within which the taxpayer has to request a CDP hearing, but the message can be drowned out by the scare tactics elsewhere within the notice. On the positive side, both these notices at least specify their statutory basis – sections 6331(d) and 6330(a) respectively, and both must be delivered personally or by certified mail. There are other problems with collection notices, but those will do for a start.
Keith has a great blog post here about the distinction between the 6331(d) notice (CP504) and the 6330(a) “final notice” providing the right to a CDP hearing (LT11, CP90, etc.) and the history. It also points out how misleading, if not knowingly false, the 6331(d) notice is. Go read Keith’s post now if you haven’t already/lately. I like to think of the difference as that the 6331(d) notice by itself only allows the limited categories of levy described in section 6330(f), most notably state tax refunds or jeopardy collections. For those categories, a CDP hearing is available only post-levy. For all other categories, a CDP hearing is available pre-levy but the IRS must first send the 6330(a) “final notice” to allow the taxpayer to request a hearing. It’s a bit of a mess procedurally.
Math error notices are . . . well, look at this post by Keith (including the comments and the NTA blog post he links to) for some of the many defects of math error notices.
Multiple Communications
GMI, on the other hand, was dealing with a notice of computational adjustment. The Court of Federal Claims and the Federal Circuit considered several of the communications from the IRS to GMI as possible notices of computational adjustment, either separately or collectively, that would have started the statute of limitations running.
First, on August 27, 2010, the IRS sent a letter with Form 5278, “Statement—Income Tax Changes” enclosed. That form included a line for “Balance due or (Overpayment) excluding interest and penalties” with a corresponding dollar amount, the additional tax owed by the partners from the settlement of the partnership audit. No amount was shown for interest, but the cover letter stated that the IRS “will adjust your account and figure the interest.”
The Court of Federal Claims quoted a statement in the earlier Form 870-LT (AD) settlement agreement for the Cereals audit that the taxpayer consented to the immediate assessment of deficiencies in tax and penalties “plus any interest provided by law.” I emphasized that last phrase because the court did. Three times, once for each place that phrase occurred on Form 870-LT (AD). Pause for a moment and consider the reaction if a settlement agreement or notice of deficiency had not specified the exact amount of tax and penalties and merely said “the amount of additional tax and penalties provided by law.” And here, there was enough ambiguity about what the law provided regarding “hot interest” that “provided by law” was not sufficient anyway.
Second, the IRS assessed the adjustments to GMI’s returns, flowing through from the partnership audit and including interest, on September 3, 2010. The assessment, of course, would have just shown up as a lump sum. I defy anyone to reverse engineer the calculations underlying an interest assessment on a large corporate return, including flow-through adjustments from a partnership. It can be done, but anyone in their right mind would give up long before they reached a solution. GMI proceeded to pay the assessed amounts on April 11, 2011.
Finally, on April 18, 2011, the IRS sent GMI detailed interest computation schedules. As described, those were probably sufficient to identify the IRS application, which GMI disputed, of the “hot interest” provisions. (This discussion focuses on the 2002-2003 tax years. For the 2004-2006 tax year, the IRS sent two different sets of interest computation schedules, the first one of which implied that “hot interest” wouldn’t apply at all.)
GMI’s Argument
What was missing from all these communications? The August 27, 2010, letter didn’t state the amount of the computational adjustment. The September 3, 2010, assessment gave the total amount of interest assessed, which did not break out the portion attributable to the increased tax and penalties included in the computational adjustment and was insufficient to identify how the IRS calculated that amount. The April 18, 2011, interest computation schedules did provide sufficient information to identify what GMI considered in error. None of them were explicitly identified as a “notice of computational adjustment.” None of them specified the applicable statutory provision. None of them specified the deadline for filing a refund claim, a deadline that was different from the normal statute of limitations for filing refund claims.
When I compare this notice of computational adjustment (at least with respect to the interest amount) to other notices described above, it seems: (a) significantly worse than the notice of deficiency and an innocent spouse notice of determination; (b) arguably worse than collection notices; and (c) at least as bad as some of the worst examples of math error notices. Being “no worse than math error notices” is not a good standard for the IRS to strive for.
GMI argued that
a notice of computation adjustment must (1) contain the amount of adjustments, (2) contain a statement of the increased rate of interest that will apply, and (3) provide the partner with some indication that the document is intended to be a notice of computational adjustment that triggers a 6-month period of limitations under § 6230(c)(2)(A).
It cited McGann v. United States, 76 Fed. Cl. 745 (2007) as establishing that standard.
The Federal Circuit Was Not Convinced
The court noted GMI’s arguments but rejected them because those asserted requirements went beyond the statutory text.
GMI contends that the notices were defective for various reasons. First, GMI says that the IRS was required to give notice that “a jurisdictional period was being triggered,” and the schedules failed to mention § 6230(c) or the six-month limitations period. GMI also argues that the schedules were tainted by the failure to mention the Partnership proceedings and the failure to separate the accrued interest on underpayments resulting from the corporate proceedings from that of the Partnership proceedings. These contentions lack merit. The Court of Federal Claims stated that the notice of computational adjustment need not be in any particular form, and we agree. Indeed, the Internal Revenue Code does not define what a notice of computational adjustment should contain.
(citations omitted) At most, the court would have required “the information [GMI] needed to assess whether the IRS may have erroneously computed the computational adjustment,” but it concluded the various communications provided that.
The court’s apparent conclusion that a notice, unless it is misleading, need only comport with statutory requirements is disturbing. McGann stands for the proposition that there is a minimum requirement of due process that may exceed statutory requirements. The Federal Circuit distinguished McGann as involving a misleading notice of computational adjustment and concluded the notices sent to GMI were neither misleading nor contradictory. It didn’t mention the statement from McGann that “the notice of balance due bears no indication that it is to be taken as a notice of computational adjustment, nor does it disclose that Mr. and Mrs. McGann would have had to contest any amounts said to be due within a six-months’ period thereafter.” The McGann court also pointed out that “neither the Form 4549A nor the accompanying Form 886-A previously sent to Mr. and Mrs. McGann contained such an advisory.” The McGann court also looked to cases involving notices of deficiency as “instructive” in determining whether a notice of computational adjustment was adequate.
Conclusion
I can understand why Congress and the IRS might not specify, and courts might be reluctant to impose, rigorous requirements for notices of computational adjustment. With respect to additional tax and penalties, the amount of the aggregate adjustment was already determined in the TEFRA proceeding. The partner-level adjustments seem merely mechanical and unlikely to be in error. But it doesn’t make sense with respect to complex interest computations, which were not addressed in the TEFRA proceeding. That rationale is probably also the unstated reason for why the IRS treats math error notices so cavalierly. But we know those math error notices also increasingly include adjustments that go far beyond the simple mechanical adjustments that have an extremely high likelihood of being correct.
Not only math error notices but also notices of computational adjustments require improvement, beyond the statutory requirements, to protect taxpayer rights. We can and should work with the IRS and Congress to achieve better notices. But pushing for the courts to recognize and enforce higher standards is also a worthy fight, even if it may feel like tilting at windmills.