Paul L. Caron

Monday, June 4, 2018

Lesson From The Tax Court: A Haunting

Four cases from the last couple of weeks illustrate the continued fallout from the Tax Court’s recent about-face in its reading of §6751(b)(1). Graev v. Commissioner, 149 T.C. No. 23 (Dec. 20, 2017)(commonly called Graev III because it was the Tax Court’s third published opinion regarding Mr. Graev’s case). The good folks at Procedurally Taxing have been following Graev III’s impact here, here and here (to name a few). These four cases add a new wrinkle.

In all four cases, the Service had failed to produce evidence in the initial trial that it had complied with §6751(b)(1). And for good reason. All four cases had gone to trial before the Court issued its opinion in Graev III. At the time of trial the Tax Court’s fully reviewed position on §6751(b)(1) was that consideration of penalty approval was premature when contesting an NOD.

In all four cases the Service asked the Tax Court to re-open the record to allow it to introduce the theretofore-unrequired-but-now-required evidence. The cases were heard by three different Tax Court judges. In two cases, the Court allowed the record to be reopened and in two cases the Court refused. Taken together, the cases illustrate how the fallout from the Tax Court’s Graev decision continues to elevate procedure over substance. As a result, similarly situated taxpayers receive very different outcomes based both on which IRS attorneys work the cases, what information the attorneys have, perhaps most importantly, which Tax Court judge decides. Four cases, three judges, two opposing outcomes, all in one discussion, waiting for you below the fold.

The Law

Section 6751(b)(1) says, in relevant part: “No penalty under this title shall be assessed unless the initial determination of such assessment is personally approved (in writing) by the immediate supervisor of the individual making such determination or such higher level official as the Secretary may designate.”

Now, if the phrase “initial determination of such assessment” bothers you, then congratulations! You know something, both about the English language and about tax procedure, that the drafters of that language did not: IRS employees do not “determine assessments.”

As a matter of English language that phrase is problematic. The word “assessment” is a word akin to “judgement.” To assess something is to judge it, as in “I assessed my chances of winning at zero and so folded my hand.” Used that way, it means the same thing as “determine,” as in “I determined my chances of winning were zero and so folded my hand.” As applied to the language in §6751, that means the drafters prohibited the IRS from determining a penalty until the “initial determination of such determination” was personally approved.

As a matter of tax administration the phrase fares no better. In tax administration, the term “assessment” is just the act of recording a liability in the Service’s account system. §6203(a). So defined, the term takes on a meaning different than the term “liability.” That is, IRS employees exercise judgement about tax liabilities (they “determine” liabilities) and then they assess (i.e. record) the liabilities so determined on the books of the IRS. That’s the act of assessment. Judge Holmes gives a nice exposition on the difference between a tax “liability and the “assessment” of a tax liability in his Graev III concurring opinion. He gets it.

Up until December 20, 2017, the Tax Court had focused on the tax administration meaning of the word “assessment” in §6751(b)(1) to hold that taxpayers could not raise compliance with §6751 during a deficiency proceeding because there was no assessment yet. The Tax Code in fact prohibits the IRS from making an assessment of the liability proposed in a Notice of Deficiency (NOD) until after the Tax Court proceeding and any ensuing appeals are over. §6213. Without an “assessment of the penalty” on the books, the Court was not in a position to see whether the IRS had complied with the requirement that the “initial determination” of “such assessment” had been approved.

However, on December 20, 2017, the Tax Court changed its collective mind, meekly following the 2nd Circuit Court of Appeals decision in Chai v. Commissioner, 851 F.3d 190 (2nd Cir. 2017). There, faced with the task of making sense of nonsense, the 2nd Circuit went with what it thought the purpose of the statute was and decided that "the written approval requirement is appropriately viewed as an element of a penalty claim and therefore requires written approval of the initial penalty determination no later than the date the IRS issues the notice of deficiency (or files an answer or amended answer) asserting such penalty.” In Graev III, the Tax Court basically adopted this holding as its own. Here’s how Judge Lauber’s concurrence defends it:

“The Court adopts a more sensible approach. It treats the "initial determination of such assessment" as referring to the action of the IRS official who first proposes that a penalty be asserted. This is a reasonable construction of the statute, giving primacy to the word "initial" (a term that appears in the statute) rather than to the penalty-asserter's scope of authority (a term that does not appear). And by requiring supervisory approval the first time an IRS official introduces the penalty into the conversation, the Court's interpretation is faithful to Congress' purpose by affording maximum protection to taxpayers against the improper wielding of penalties as bargaining chips.”

Notice how this re-writing of the statute subs in the word “penalty” for “assessment” to rewrite the statute as follows: “No penalty under this title shall be assessed unless the initial determination of such penalty is personally approved (in writing) by the immediate supervisor of the individual making such penalty determination or such higher level official as the Secretary may designate.” (emphasis supplied).

So here’s the new rule, as it works with §7491(c): the Service must now produce evidence in every case that any penalty proposed in the NOD was approved in writing by the immediate supervisor of whomever made the “initial determination” to impose the proposed penalty.  If the Service fails to introduce such evidence, the taxpayer escapes the penalty.

In his long concurrence in Graev III Judge Holmes expressed a concern that this interpretation of the statute meant that “a procedural restriction on assessment has been transformed into a protection in deficiency cases against liability for penalty imposed by the Code.”  Judge Holmes then goes on to list a parade of horribles that might arise from this judicial excursion into tax administration policy-making. He calls them “ghouls.” That’s an apt word, as these four cases show. They illustrate ghouls that Judge Holmes did not fully anticipate: what to do about reopening the record and what to do about jeopardy assessments.  They illustrate his concern that formalism is trumping substance.  Ironically enough, it's an opinion of Judge Holmes that best illustrates the very concerns he raised in his concurrence.

The Cases

First, in Bradford J. Sarvak v. Commissioner, T.C. Memo 2018-68 (May 21, 2018), the NOD proposed deficiencies of $360k for 2011 and $452k for 2012. The NOD also proposed §6662 penalties of $72k and $94k. Judge Cohen held trial in April 2017. After Graev III, the Service moved to re-open the record to admit evidence showing the proper penalty approval.

The Tax Court has discretion to allow a party to reopen the record to submit additional evidence. That discretion is cabined by an indeterminate multi-factor test created by the reviewed opinion in Butler v. Commissioner, 114 T.C. 276 (2000). As explained by Judge Cohen, the Court will grant a motion to reopen “only if the evidence relied on is not merely cumulative or impeaching, is material to the issues involved, and probably would change some aspect of the outcome of the case.”

You would think that is not a difficult standard to meet in these “Chai Ghoul” cases. In Sarvak, for example, the evidence the IRS wanted to get in the record was a declaration by the examining agent and a penalty approval form signed by the agent’s supervisor and dated before the NOD was issued. Judge Cohen had no difficulty finding the IRS met the Butler factors and she then sustained the application of §6662 penalties. Mr. Sarvak argued that allowing the Service to reopen the record would be unfair because the Court was, in other cases, not allowing the Service to reopen the record and so other taxpayers, equally inaccurate and negligent, were escaping penalties. Hey, when some $160k in penalties are on the line, that sounds like a good argument. Judge Cohen rejected it out of hand, with no word of explanation. None is really needed, at least to those of us who have raised children.

Second, in Slawomir J. Fiedziuszko and Alicia M. Fiedziuszko v. Commissioner, T.C. Memo 2018-75 (May 31, 2018), the NOD proposed a tax deficiency of $31k and a §6662(a) penalty of $6k. The trial record closed before the Tax Court issued its Graev III opinion. As in Sarvak, the Service wanted to introduce a declaration by the examining agent and a penalty approval form.  The Service asked Judge Pugh to reopen the record.

You would think Judge Pugh would use the same indeterminate multi-factor test that Judge Cohen used. You would be wrong. She used a different indeterminate multi-factor test, pulled from a Fifth Circuit case. She writes: “In reviewing motions to reopen the record, courts have considered when the moving party knew that a fact was disputed, whether the evidentiary issue was foreseeable, and whether the moving party had a reason for the failure to produce the evidence earlier.” She does not even mention Butler. Applying th3 5th Circuit test , she allowed the motion, “because the record was closed in this case before we issued Graev III and because petitioners never raised section 6751(b) as an issue before the record closed. Therefore, respondent was justified in concluding that introduction of the Civil Penalty Approval Form was not necessary.” Judge Pugh then sustained application of §6662 penalties.

Third, in Yasmin Azam and Muhammad Ayub Azam v. Commissioner, T.C. Memo 2018-72 (May 29, 2018), the NOD proposed deficiencies of $85k and $44k for 2008 and 2009 and §6662(a) penalties of $17k and $9k for the same years. As with the other two cases, the trial record had closed before Graev III. Unlike the other two cases, the Service did not seek to introduce a nice clean Penalty Approval Form, authenticated by an accompanying declaration. Nope. Instead, the Service wanted to put in the record “a heavily redacted 12-page” memo from field counsel in San Francisco to Appeals.  It seems the reason the Service wanted to introduce the memo and not the standard penalty form was because the examining agent had proposed a fraud penalty but when the case had gone to Appeals, field counsel had recommended the §6662(a) penalty instead of the §6663 fraud penalty.  In addition to memo the Service asked to submit a declaration from both the attorney who wrote the memo and his supervisor to explain the memo and explain whose initials appeared on it and why.

Judge Pugh was not having any of this Chai ghoul.  She now recited the Butler indeterminate multi-factor test, but then ignored it in favor of deciding that the proffered evidence was not admissible in the first place.  Why?  Because the declarations failed to recite magic words that would take the 12-page memo out of the hearsay rule. At the end of the opinion, Judge Pugh wrote some dicta that even if she ruled the memo admissible, “the declarations will not change the outcome of the case because the memorandum alone will not satisfy respondent’s burden of proving that the penalty was approved by the immediate supervisor.” So she ties it back to Butler.  And yet.  And yet.  Judge Pugh then goes on to chastise the taxpayers that “their failure to keep records was not what a reasonable person would do under the circumstances.” In other words: “yeah, I would have approved the penalty had the IRS not goofed up the procedure.”

Judge Pugh was careful to note that the Service had not requested “further trial on the matter.” I read that as suggesting that Judge Pugh might have allowed such a motion, to give the taxpayer the opportunity to question the attorney and the supervisor and to give the Service the ability to give direct evidence of compliance with §6751(b)(1) in the absence of a penalty form.  But notice how the formalism here gets in the way of substance.  On the substance, these taxpayers should have paid an accuracy related penalty.  

Fourth is the case of Joseph C. Becker & Mercy Grace Castro, et al., v Commissioner, T.C. Memo 2018-69 (May 21, 2018).  Keith Fogg has a good blog about it here. There, Judge Holmes unearthed a “new species” of Chai ghouls: “the jeopardy-assessment Chai ghoul.” In 2012 the Service issued an NOD for tax years 2007 and 2008 which proposed §6662(a) penalties. The taxpayers petitioned Tax Court. The Service later that year made jeopardy assessments for 2007, 2008, and added 2009 for good measure, including a fraud penalty assessment for each year. When the Service later filed an answer to the Tax Court petition, it sought to add the §6663 fraud penalty which had been in the jeopardy assessment but not the original NOD. Later, the 2010 year gets added to the case and for that year the Service sought penalties only against Mr. Becker.

As with the other cases, the trial record closed before Graev III and the Service moved to reopen the record to introduce various forms along with a declaration from the supervisor to authenticate the forms. As in the Azam case, the Service did not ask for additional trial on the evidence it proffered. In a separate order (here), Judge Holmes rejected the Service’s motion to reopen the record but then devotes the last eight pages of his 52 page opinion to explain why the Service had proved by clear and convincing evidence that Mr. Becker committed fraud.  Of course, that is all dicta because, as Judge Holmes writes: “even though Mr. Becker's fraud is evident, we hold that he and Ms. Castro are not liable for civil fraud penalties or accuracy-related penalties in these cases” because “the Commissioner introduced no...evidence that he complied with section 6751.” Well, hell, the Commissioner sure tried to! As in Azam, these were taxpayers who should have paid penalties.  Only the procedural problems prevented that.  Let’s look at what went wrong.

Two aspects of Judge Holmes’ separate order are noteworthy. First, as in Azam, the Service was seeking to introduce non-standard information and yet did not ask for further trial. Here, for example, the proffered penalty forms were supposedly signed by the same supervisor but apparently had different-looking signatures.  Having the ability to put on a witness who could be subject to cross would have solved any prejudice to the taxpayer from introducing the evidence. 

The big problem for Judge Holmes, however, was with the forms offered to support the penalties contained in the jeopardy assessments.  Keep in mind that the Service had used the jeopardy procedures to actually make the assessment before the taxpayer got to Tax Court.  “Even before Chai and Graev III, the plain language of §6751(b)(1) required that written supervisory approval for penalties be secured before assessment.”  The dates on the forms to support the fraud were after the date of the jeopardy assessment. Whoops.

To me, Judge Holmes is being unduly formalistic here.  First, he overstates the "plain language" of the statute.  Folks, it ain't plain.  More importantly, if the thrust of the Tax Court’s new interpretation of §6751(b)(1) is to ensure that penalties are “genuinely” proposed and not just thrown in to create bargaining chips, then the Service ought to be able to show compliance with §6751 through non-standard means. Particularly in this case, it would seem to me that the approval process for Jeopardy assessments would satisfy the §6751 requirement. See IRM (07-15-2011) which says “A jeopardy assessment must be approved by the Area Director and Area Counsel AND be made within 24 hours of the Area Director’s approval. See IRM, Securing Approval of the Package.” Further, “Written Approval by Area Counsel is required under the provisions of the IRS Restructuring and Reform Act of 1998. Failure to do so will result in the abatement of any assessment made under IRC 6201.”

The second noteworthy feature of Judge Holmes’ order is his total contradiction to Judge Pugh. Remember that one of the myriad and indeterminate factors that courts can pick and choose from in deciding whether to reopen the record is whether the party seeking to reopen the record “should have known” to use the evidence for which permission is now sought. Judge Pugh said, in effect: “No duh, the record was closed before Graev III so the IRS had no way to know we were going to change our collective mind on how to interpret §6751(b)(1).” Judge Holmes said, in actuality: “Our decision in Graev III didn’t create new law; it interpreted a section of the Code that was in effect at the time of the trial in these cases, and which we then applied to the parties before us in that case.”

Judge Holmes’ idea that the Tax Court’s flip in Graev III “didn’t create new law” is a beautiful example of either ingenuousness or disingenuousness, I cannot decide. Either way, it is again unduly formalistic. Only those who were either born before 1800 or have consumed more than three beers would agree that Graev III “didn’t create new law.” The idea is especially strange coming from the very Judge who so ably explained exactly why (in his concurring opinion) the “section of the Code” at issue was incomprehensible as written. Congress wrote gibberish into §6751(b)(1). The gibberish could have several different meanings. It is just wrong to say that the gibberish has “always” had whatever meaning the Tax Court latest opinion gives it.


I draw two lessons from these four cases.

First, the Tax Court should have made Graev III prospective only. You might say that not doing so was a "Graev" mistake.  If it had done so, it would have avoided all four of these cases where the question of whether to reopen the record is affected by (1) the IRS counsel’s perceptiveness of (2) what the particular Tax Court judge will (3) accept as offered or want to have further trial about.

Second, putting aside the propriety of the courts re-writing a statute as the 2nd Circuit and Tax Court have done here, their justification for the interpolation is weak. Their idea---taken from the legislative history---is that Congress created this provision to prevent low level IRS employees from putting penalty crap into NODs for the sole purpose of forcing taxpayers into weaker bargaining positions. So the remedy is to require a supervisor to sign off on the crap. Then it won’t be crap. This is very similar to the idea you find in §7605(b), restricting multiple examinations.

Each step of the logic is problematic, however, and cautions against a strict or formalistic application of the statute. First, low level IRS employees generally put lots of non-penalty crap into NODs. They do that for a perceived good reason: the taxpayer is not cooperating. “Hey, you don’t want to give me information supporting some of your deductions? Fine, I’ll not only disallow the deductions for which you gave me nothing, I will also disallow all the other deductions and you can sort it out later.” So I would be shocked, shocked, to see an RAR that did not have a lot of “water” in it that the taxpayer needs to wring out in Appeals.

Second, the “purpose” of putting in the crap is not really about adding bargaining chips for the Office of Appeals. It’s more about punishing perceived non-cooperative taxpayers and rewarding cooperative taxpayers. That is why the standard operating procedure  when representing taxpayers in civil cases (very different from the SOP in criminal or even eggshell matters) is to cooperate: you will get the best resolution at the lowest level. If you interpret this statute formalistically and strictly, then that will, actually, create all kinds of opportunities for taxpayers to use it as a bargaining chip in Appeals. Especially given the myriad ways that a decision to impose penalties can arise. If the pathway to penalties is narrow and formal, taxpayers will use that (I sure would!) to bang on Appeals to eliminate a penalty on “hazards of litigation” grounds.

Finally, the signature requirement is pretty much an empty formalism. Supervisors regularly sign off on highly watered NODs without thinking. Compliance with the statute by signing a form does nothing to ensure than anyone actually thinks about the penalty.  

Those are my thoughts, FWIW.  I am sure others have views on how the Tax Court is interpreting the statutory gibberish in 6751. I would love to hear from folks with actual recent practice experience. Please feel free to use the comment section or just send me an email.    

Bryan Camp is the George H. Mahon Professor of Law at Texas Tech University School of Law.

Bryan Camp, New Cases, Tax, Tax Practice And Procedure | Permalink


Bryan, great write up. A quibble. You state: "The Tax Code in fact prohibits the IRS from making an assessment of the liability proposed in a Notice of Deficiency (NOD) until after the Tax Court proceeding and any ensuing appeals are over. §6213." Section 7485(a) permits assessment upon the initial appeal from the Tax Court unless the taxpayer files a bond with the Tax Court. So, the IRS can assess in most cases (where no bond) during the appeals process.

Posted by: Jack Townsend | Jun 5, 2018 7:36:37 AM