Statistical Sampling for Large Dataset Issues in Tax Litigation (3/23/24)

In Kapur v. Commissioner, T.C. Memo. 2024-28, GS here,  the Court (Judge Pugh) set out the issue as follows (Slip Op. 2, two footnotes omitted):

    Before the Court is petitioners' Motion for Protective Order. The parties dispute whether discovery and trial should be limited to a sample of projects at this stage of litigation. We decline to order sampling for the reasons summarized below. n4
   n4 This appears to be a recurring issue. See, e.g., Phx. Design Grp., Inc. v. Commissioner, No. 4759-22 (T.C. Aug. 29, 2023) (order); Feller v. Commissioner, No. 11581-20 (T.C. Aug. 10, 2023) (order). Respondent referred us to these orders but of course they are not precedential.

The Court reviews (Slip Op. 5) sampling by agreement of the parties (whether encouraged by the Court or not). The Court then says (Slip. Op. 5-6, bold face supplied by JAT):

    Respondent also claims that we do not have discretion to order sampling at the request of petitioners if respondent objects. We disagree: We do have authority to limit discovery (including by ordering sampling) over the objection of a party. See Rule 70(c)(1). Nonetheless, we agree that exercising our discretion to limit the scope of discovery and trial in 6*6 accordance with petitioners' Motion for Protective Order is improper at this stage. The only issue in this case is whether petitioners are entitled to the research credits claimed for the years in issue. Evaluating compliance with section 41 necessarily involves consideration of the underlying business components. And petitioners agree that they have the burden of showing entitlement to the claimed research credits. See Feigh, 152 T.C. at 270. As we have said previously, "[a]bsent an agreement between the parties, project sampling improperly relieves the taxpayer of its burden of proving entitlement to the research credit claimed." Betz v. Commissioner, T.C. Memo. 2023-84, at *77 n.30 (citing Bayer, 850 F. Supp. 2d at 538, 545-46).

As I understand the Court said it can order sampling for discovery purposes but cannot order sampling over IRS objection for resolution of the merits on issues as to which the taxpayer bears the burden of persuasion.

It is not clear to me that the final conclusion is consistent with the Court’s earlier rejection of the IRS claim that Court does not have authority to order sampling when the IRS does not agree. Another way of reading the paragraph is that the Court will not order sampling based on the stage of pretrial development right now in Kapur.

This raises some issues for me, but let me start with my understanding of good sampling that permits reasonable inferences about the universe of data that is sampled. Those reasonable inferences can be stated in possibilities or margins of error for the inferences, which generally can be slimmed down by increasing the sample size.

Where resolution of issues related to large data sets (say a large corporation’s trade or business expenses (or some subcategory thereof)), good sampling choices can be made that will give reasonable comfort as to the universe.

At least that is my understanding of how the sampling process works. I realize that all large data sets may not be appropriate for sampling. But many are, and that is why the parties confronted with having to try such cases may opt for sampling rather than actually proving the entire data sets; the problem is if one party refuses to agree to sampling. Given that the taxpayer usually bears the burden of persuasion, it may be usual that the taxpayer is the one who potentially suffers if the IRS obstinately refuses to agree.

Of course, I think that a court would be aggrieved if it is clear that the IRS unreasonably refused sampling to resolve the issue for the universe. A court could in the pre-trial sparring certainly encourage, if not direct, the IRS to agree.

 Some other questions:

 1. Could the taxpayer introduce into evidence an unagreed sampling with expert testimony that it is representative of the whole set and has a margin of error that is, say, 2%. Or, perhaps, 1% or .50%? Could a Court be persuaded then as to the universe?

2. In such a case, could the Court use the range process based on the interplay with the burden of persuasion that I developed in John A., Burden of Proof in Tax Cases: Valuation and Ranges — An Update (2020). 73 Tax Lawyer 389, 2020. Available at SSRN: https://ssrn.com/abstract=3599481. Since the taxpayer bears the burden of persuasion, the Court might pick the IRS favorable end of the range spectrum with an acceptable margin of error (i.e., one that persuades).

3. Could the IRS use unagreed sampling in defense of the taxpayer’s claims? I think that it comes back to the issue of the persuasiveness of the evidence on the issue the Court must resolve.

Finally, while with DOJ Tax Appellate, I was assigned to work with others to consider the use of statistical methodologies (including sampling) in tax litigation. I can’t recall (i) if it was a committee or other named group or (ii) the names of other members of the group; I don’t recall that the group ever did anything productive. I am not sure why I was assigned to work on it; the only statistics I had was psychological statistics in undergraduate school. But still, I then could bandy about some of the buzzwords about statistics. Those buzzwords had tangentially appeared in a case I handled which may have been why I was assigned. Northern Natural Gas Company v. United States, 470 F.2d 1107 (8th Cir. 1973) here, cert. denied 412 U.S. 939 (1973).

The issue in Northern Natural Gas was whether the taxpayer the allocation of lump-sum going business purchase costs between the depreciable tangible assets and nondepreciable intangible assets (good will, going concern, etc.). Purchasers in lump sum purchases often load up the costs to the depreciable tangible assets; where, as in Northern Natural Gas, multiple smaller businesses were purchased, individual asset valuations were rarely done. In assigning costs to the many depreciable tangible assets that were too numerous to individually appraise, the Government’s brief, here, said (p. 37 n. 15 boldface supplied by JAT):

15/ The elements of value included by the District Court in its replacement cost formula were the container costs, utilization equipment costs and labor installation costs. (R. 36.) The court thus included every possible element of value relating to the tangible asset itself * * * * The container costs, utilization equipment costs and labor installation cost were determined from studies and data from taxpayer's files and were confirmed within a reasonable range by an independent appraiser. (R. 377, 378, 671-678, 722; Exs. 19, 245, Ex. Vol. 35-38, 223-224.) With respect to the useful life element, the District Court gave taxpayer the benefit of any doubt by assigning a longer total useful life (25 years on tanks and 20 years on cylinders (R. 36)) for the equipment than taxpayer's own witness testified to as being his best judgment on the useful life (16 years) (R. 536-537)). Finally, the remaining useful life was determined by subtracting the average age of the customer equipment from the total useful life. The Commissioner's average age  determinations were sustained in each acquisition except one because  there was no evidence of record to establish any shorter average age. (R. 32-33, 36.) In this respect, the Commissioner’s average age determinations were made under accepted appraisal techniques. (R. 631-635, 639, 668-669.)

 The boldface statement is cryptic as to the methodology used, but suggests (and as I recall) some statistical methodology to determine a comfort level within ranges. (This footnote discussion was included virtually verbatim in the brief in opposition, here, to the taxpayer's petition for certiorari at p, 4 fn. 3.)

 As an aside, I deployed in the Northern Natural Gas brief one of my favorite concepts on burden of proof. Prior to writing the brief, I had been assigned to do a supposedly definitive Tax Division memorandum on burden of proof in tax cases to guide Tax Division attorneys in what they said about burden of proof. My anecdotal understanding is that most attorneys did not pay much attention to my memorandum. Nevertheless, the concept that I used in the Northern Natural Gas brief (pp. 39-40, n. 16):

The District Court recognized and properly applied these elementary rules in saying that taxpayer's claim must fail either because it failed to show error in the Commissioner's determinations or because, even if it had, it had failed to meet its burden of showing the correct values. The court did not rest its decision on the presumption of correctness.16/ Rather, it rested its decision on the unpersuasiveness of taxpayer's evidence of a higher valuation.

   16/ It has been noted that raising a presumption in favor of the Commissioner's determination is akin to placing a handkerchief over something already covered by a blanket Chicago Stock Yards Co. v. Commissioner, 219 F. 2d 937, 948 (C.A. 1,192), rev'd on other grounds, 318 U.S. 693 (1943). A presumption can only have significance when it is raised in favor of the party bearing the burden of persuasion, since it serves merely to shift the burden of production to the opposing party. Thus, a presumption has no significance when it is raised against the party bearing the burden of persuasion, for that party already bears a production burden by virture (sic) of the fact that he must introduce persuasive evidence anyway. See 9 Wigmore, Evidence (1940 ed.), Sec. 2498(a), p. 349; Louisell & Hazard, Cases on Pleading & Procedure (1968), p. 1007.
   Not only is discussion of the presumption of correctness irrelevant, it can lead to confusion as to the allocation of the burden of persuasion and, for this reason, it has been recognized that any discussion of the presumption must not obscure the fact that the taxpayer must sustain his burden of persuasion even though he has knocked the presumption out of the case by meeting his production burden. United  Aniline Co. v. Commissioner, 316 F. 2d [701, 704] ( 1^st Cir. 1963).]*

That footnote echoes what I recommended in the memorandum—that DOJ Tax attorneys should avoid relying on the presumption of correctness and focus on the burden of persuasion. That recommendation was respected in the breach.

* The bracketed content was omitted from the brief through inadequate brief checking.