Defending R&D Credits

By Gerald H. Parshall Jr., J.D.



  • The IRS has taken a hard-line stance on the substantiation of R&D credits and routinely disallows any credits that are based on estimates of qualified research expenses (QREs). In its guidance to its auditors, the IRS has been extremely critical of the use of R&D credit studies prepared by outside consultants to support R&D credit claims.
  • Decisions in recent court cases have been steadily eroding the support for the IRS’s position on the substantiation required for R&D credits. Notably, the IRS suffered a full reversal on appeal in McFerrin, a case that it has cited extensively when disallowing claimed R&D credits due to insufficient substantiation.
  • In Union Carbide, the Tax Court held that a taxpayer could use estimates in determining R&D credits and that the oral testimony of employees could be used as evidence to prove the amount of qualified research expenses. The Tax Court also held that a court was obligated to apply the Cohan rule to estimate the correct R&D credit amount where a taxpayer had proved the existence of some qualified research expenses but not their exact amount.

This article highlights and analyzes some recent decisions concerning the research and development (R&D) tax credit and IRS administrative practices when auditing R&D credit claims, most notably the Union Carbide decision in the Tax Court. These recent developments may herald greater reasonableness about R&D credit claims—at least by the courts, and possibly by the IRS as well. For several years, the IRS has been relying on anti-taxpayer R&D credit opinions with little precedential value. This article points out how that strategy conflicts with Treasury’s public statements about the persuasiveness of court opinions and points practitioners to better authorities with which to counter the IRS’s methods in conducting R&D credit audits.

In 2000, the IRS and Treasury introduced and then dropped a requirement in the regulations that R&D credit claims be supported by “contemporaneous documentation.” 1 This requirement demanded that researchers prepare and retain, before or in the early stages of experimentation, a description of the technical problems they faced and what was planned that “exceeds, expands or refines the common knowledge of skilled professionals in the relevant field of science or engineering.” 2 The IRS and Treasury also withdrew the “exceeds common knowledge” requirement, often referred to as the “discovery test.” Both requirements in the proposed regulations had received substantial public criticism.

However, at least until recently, the news has been more bad than good for taxpayers. The relaxed attitude shown by Treasury in several iterations of R&D regulations has not filtered out to the field. The IRS has been less reasonable in the guidance it is providing its auditors; predictably, this has been reflected in those auditors’ examination stances. This guidance relies on questionable case law, primarily Eustace 3 and McFerrin. 4 The IRS has frequently cited these cases for the proposition that the Cohan rule does not apply to R&D credits and that R&D credits cannot be based on estimates of creditable expenses developed through after-the-fact R&D credit studies.

Revenue agents and their managers have been taking this centralized advice to heart, often completely disallowing the credit as their first negotiating point. They are supported in this by the IRS’s Large and Mid-Size Business (LMSB) Division, which has designated the R&D credit as a Tier I issue, subject to centralized audit management and now almost mandatory assertion of penalties.

The advice that revenue agents are getting from the LMSB characterizes most R&D credit studies as “prepackaged submissions,” often based on oral testimony, estimates, and ad hoc methodologies, which are all very difficult to audit. These criticisms appeared most recently in an audit guide (the RCCATG) 5 and an industry director directive (IDD No. 2). 6

The newest twist appearing in the IDD is the requirement that in all examinations in which a research credit refund is partially or wholly disallowed, the revenue agent must address the penalty and obtain concurrence from one of a handful of R&D technical advisers for not asserting it. This will make the assertion of the Sec. 6676 penalty for erroneous refund claims much more prevalent, almost mandatory, for R&D credit claims.

Unfortunately for many taxpayers whose businesses are just becoming profitable or who are just becoming aware of the R&D credit, many of the criticisms leveled in those two publications are unavoidable because the traditional method of calculating the R&D credit compares the current credit years with a 1984–1988 base period, and many taxpayers’ physical records may no longer exist for such years.

The tax adviser is thus faced with a dilemma: Forfeit the credit (or the greatest part of it) for a given taxpayer, even when it is clear that significant resources have been dedicated to innovation over a long period of time, or find witnesses who can recall what R&D was being conducted in the base period and obtain testimony and R&D documentation for more recent years.

The Good News

Good arguments can be made that the IRS’s reliance on Eustace and McFerrin is clearly inconsistent with the regulations. This was so even before the Fifth Circuit reversed the district court’s holding in McFerrin on June 9, 2009. While audit guides and IDDs are binding on neither the IRS nor taxpayers, regulations are.

The recent reversal of McFerrin and the opinion by the Tax Court in the Union Carbide case should come as good news. 7 Both of these cases, which are discussed in more detail below, are blows to the hard-line stance that the IRS had taken in R&D credit cases.

In addition, in June 2009, in the FedEx case, the IRS was forced to abandon its “whipsaw” argument—that taxpayers wanting to use the IRS’s internal use software rules appearing in older proposed regulations were required to submit to those older proposed regulations’ adoption of the “exceeds, expands or refines the common knowledge of skilled professionals” discovery test, long since abandoned by the IRS and Treasury and recently criticized in the McFerrin reversal. 8

Finally, in November 2009 and again in January 2010, the IRS’s position on prototypes was substantially challenged by adverse decisions, first in Tax Court and then in a district court opinion. In the first decision, T.G. Missouri, 9 a manufacturer of injection-molded products sought to claim as qualified research expenses (QREs) contract work performed by third parties on molds it sold to customers. The IRS tried to resist this on the basis that the molds were “property of a character subject to the allowance for depreciation” and so were disqualified from the credit. The Tax Court held that the language used in the statute tied creditability to the proper financial accounting treatment of the molds in the hands of the taxpayer, not the taxpayer’s customers. Because in the taxpayer’s hands the molds were not properly depreciable, they were also not disqualified from the credit.

A district court decision 10 held that a shipbuilder’s prototypes (“first in class” vessels) qualified as creditable business components and could further qualify as R&D where the evidence of novelty was particularly strong. Where the evidence was less strong, the taxpayer lost because it could not produce evidence allowing the district court to “shrink back” four of six first-in-class ship design projects to their experimental aspects.

What Is a Good Tax Precedent?

Treasury laid out its guidelines for evaluating tax precedents in the regulations for the penalty for substantial understatement of tax. Under Regs. Sec. 1.6662-4(a), such penalties can be reduced by the amount of an understatement attributable to an item for which there is substantial authority. Regs. Sec. 1.6662-4(d)(3)(ii) evaluates authorities as follows:

The weight accorded an authority depends on its relevance and persuasiveness, and the type of document providing the authority. For example, a case or revenue ruling having some facts in common with the tax treatment at issue is not particularly relevant if the authority is materially distinguishable on its facts, or is otherwise inapplicable to the tax treatment at issue. An authority that merely states a conclusion ordinarily is less persuasive than one that reaches its conclusion by cogently relating the applicable law to pertinent facts.

Therefore, a party generally cannot rely on a case with unique facts unless the subsequent situation’s facts closely match it. Similarly, a reversed case has no persuasiveness. Eustace and McFerrin, which the IRS has been relying on, cannot pass muster under Regs. Sec. 1.6662-4.


In Eustace the taxpayer sought the R&D credit for software it had developed in house, but the taxpayer’s main witness stated that they were not doing any research:

Petitioners’ computer expert testified, relating to the expansion of the rating module: “It’s hard for me to think about writing each of those programs as constituting research.” Applied Systems’ vice president in charge of First Rate [an insurance risk-rating software package] testified: “Really from 1990 to ’92 there were minor enhancements, modifications to make things easier. There weren’t a lot of major changes.” 11

It is little wonder that later in the opinion, once the taxpayer had “presented insufficient evidence relating to evaluations/R&D and failed to address this issue on brief,” 12 the court declined to apply Cohan, 13 stating: “The rule of Cohan v. Commissioner . . . does not require the Court, in this case, to make . . . an allocation [of salaries to functionality].” 14 The Eustace court therefore acknowledged that Cohan was still good law but simply declined to apply it where the taxpayer’s primary witness testified that the company was not conducting R&D. Eustace was decided on unique facts (the taxpayer’s lack of evidence); applying the regulation’s standards for tax authorities leaves Eustace sorely wanting. 15

Practice tip: Review testimony with witnesses before they have any contact with IRS personnel. Impress on these witnesses that what does not seem groundbreaking to them may still pass the tests in the regulations, and ask them to stress technical uncertainties and the processes they use to evaluate different approaches to these uncertainties.


McFerrin involved a chemicals manufacturer that had received a refund from an R&D credit claim. The United States sued the taxpayer to recover the refund; because the government was suing, it bore the burden of proof. The Department of Justice’s trial attorney misbriefed the court on the applicable legal authorities, and as a result the decision contained three surprising rulings. First, the court applied the “discovery test,” 16 which the IRS and Treasury had eliminated from the regulations in 2004. 17 Second, the court stated 18 that research had to aim only at capability uncertainty—whether something could be done at all—to satisfy the statute, even though Regs. Sec. 1.41-4(a)(3)(i) states that uncertainty as to capability, method, or design will qualify. 19 Third, the opinion tried to apply to chemical engineering—the science involved in that case—the “high threshold of innovation” standard. 20 Since the Tax Reform Act of 1986 there has been such a standard in Sec. 41, but only for “internal use software” development. 21 This high threshold of innovation standard has no applicability outside the software context.

The taxpayers did object to all the statements of law by the Department of Justice. They also debunked the discovery test and pointed out to the court that more than simple capability uncertainty (to include method and design uncertainty according to Regs. Sec. 1.41-4(a)(5)(i)) qualified. The taxpayers never addressed the high threshold of innovation standard applied by the court, perhaps recognizing that no such test exists anywhere but in the legislative history of the internal use software exclusion.

The Fifth Circuit, in reversing the district court’s McFerrin holding, noted that the authorities the lower court relied upon had been superseded by Treasury regulations and represented “an erroneous interpretation of law.” 22 Noting the government’s burden of proof (McFerrin was a rare “erroneous refund” case), the Fifth Circuit mandated that QREs be estimated by the district court under the Cohan rule.

This was only defensive good news, pointing out that the government’s favorite R&D credit cases are inherently flawed. What taxpayers needed was a case allowing the type of R&D evidence to which the IRS traditionally objects. In Union Carbide they got such a case, which clearly allowed (and even relied upon) numerous types of evidence regularly used to support R&D credit claims.

Union Carbide

In Union Carbide, Union Carbide Corporation (UCC) was seeking an additional $8.4 million research credit for tax years 1994 and 1995. UCC based the additional credit on $201.6 million in QREs that it claimed it had subsequently identified. The court finally determined that UCC was entitled to only $1,045 in additional QREs—a complete denial of the additional credit sought and thus a government victory—but this was the taxpayer’s second attempt after previously receiving $18 million in R&D credits for those two years.

Nevertheless, the case merits substantial attention by taxpayers and their advisers seeking to compute research credits. It is interesting because unlike most R&D credit claims, which are wage driven, the bulk of the additional QREs in Union Carbide consisted of supplies. (The Code allows four types of expenses as QREs: internal wages (Sec. 41(b)(2)(A)(i)), internal supplies (Sec. 41(b)(2)(A)(ii)), computer services bought from a third party (Sec. 41(b)(2)(A)(iii)), and external contract research (Sec. 41(b)(1)(B)).)

Further, although many R&D credit claims involve product research, Union Carbide centered on process research. Regs. Sec. 1.41-4(b)(2) makes it clear that more than just products qualify as business components as the object of qualified research:

The requirements of section 41(d) and paragraph (a) of this section are to be applied first at the level of the discrete business component, that is, the product, process, computer software, technique, formula, or invention to be held for sale, lease, or license, or used by the taxpayer in a trade or business of the taxpayer.

In the end, UCC lost its second case for two reasons. First, the court applied a rule of law under Sec. 174 that costs to build out a concept are not deductible. Second, UCC did not give the judge anything on which to rely to differentiate its production supplies from its experimental supplies (production expenses do not count as QREs, while experimental ones can). But the Union Carbide opinion allows many elements of proof and calculation that the IRS criticizes in its RCCATG and IDD No. 2. Just as important, in Union Carbide the Tax Court applied Cohan to the taxpayer’s advantage.

Oral Evidence

The IRS has argued against the use of oral evidence to support R&D credit claims for several years now, both in its guidance issued to examiners and in audits. This was not always the case. The 2005 RCCATG encouraged agents to review testimony and take witness interviews in tandem with another IRS employee. However, when the IRS issued the 2008 RCCATG, it shifted the focus to documentary evidence, and information document requests were suggested to determine the adequacy of the taxpayer’s claim. All references to testimony in the 2008 RCCATG were disapproving.

However, the Union Carbide decision cited numerous fact witnesses whom UCC’s professionals had interviewed (157 current and former employees) to determine what to include in the R&D claim. The decision does not criticize UCC’s need for such testimony to support its claims. In at least one notable instance, the IRS insisted that UCC had failed to produce sufficient documentary evidence to support the qualification of certain activities (surrounding a chemical process abbreviated as UCAT-J). The judge disagreed, ruling that through fact witnesses, corroborated by some documentation, UCC had met its burden of proving that UCAT-J was qualified research.

The court stated that it was considering the “record in its entirety.” 23 A similar rejection of the IRS’s complaint that UCC had not sufficiently substantiated certain wages claimed as QREs was later answered by the judge’s ruling that the testimony of three fact witnesses sufficiently substantiated those wages. 24

Practice tip: Papers and taxpayer documents always look more reliable than notes of interviews. If a practitioner can reduce testimony to transcripts or even sworn statements, this is better than relying on the IRS’s recollection of what a witness said.

R&D Credit Claim Overall Methodology

The IRS has complained that many R&D credit claim studies are “prepackaged.” According to the IRS, these studies, while utilizing a fairly consistent methodology in preparing research credit claims, frequently fail to substantiate that the taxpayer paid or incurred qualified research expenses, and the methods used (particularly interviews of employees) to prepare the claims cannot be tested. This criticism appears in both the 2005 and 2008 RCCATGs. However, in Union Carbide the court largely disregarded the IRS’s concerns about prepackaged claims. The 2008 RCCATG’s description of what makes an R&D claim prepackaged is discussed below.

Typical Prepackaged Research Credit Claim Study

Research credit claim studies typically include the following three phases of preparation:

  • Initial evaluation. An assessment of issue feasibility is conducted to determine whether a taxpayer is entitled to the research credit or, if already reported on the return, is entitled to additional credit. This information is used to develop a plan for proceeding (or not) to phase 2 of the study. An engagement letter may be prepared before the feasibility assessment and then updated or revised based upon the findings of the study.
  • Execution of the study plan. Procedures typically consist of requesting, reviewing, and compiling accounting records; securing and reviewing the corroborative evidence that is still available; physically inspecting some operations; and conducting employee interviews at various levels within the taxpayer’s organization. Questionnaires or surveys may also be used in this process.
  • Compilation of findings. The documentation is compiled and organized for submission to the IRS. The claim or amended return is prepared and presented to the taxpayer. Computational workpapers are prepared and finalized, summarizing the findings of the taxpayer’s analysis. 25

The actual conduct of the study involves reviewing records, visiting R&D operations, and interviewing witnesses. This process is what the court explicitly approved in Union Carbide over the IRS’s objection that it was not testable or able to be duplicated: Not only did the method used (as above) meet with approval from the court, but so did the fact that UCC used outside CPAs. At numerous points in the opinion, the Tax Court quotes CPAs as experts with approval, and nowhere does it question the need for their thousands of hours of effort.

Practice tip: The practitioner’s “deliverable” is often a nicely bound analysis of the taxpayer’s facts and a compilation of its records, with an analysis of the law, often bearing the practitioner’s firm name and/or logo. Since the legal analysis never changes (or does so only slowly), seeing the same legal discussion in every study may contribute to the IRS’s perception of such studies being prepackaged. Practitioners should consider leaving it out entirely. Also consider helping the taxpayer develop its own R&D study, producing something that looks much more like an internal tax department report than an outside vendor’s efforts.

Other Methodology Concerns: Estimates, Interpolations, Ad Hoc Approaches

Even though taxpayers are regularly required by the simple structure of the R&D credit to look so far back into their history that records become scarce—and thus estimation using the best evidence available (often testimony) is the only alternative—the IRS recently greatly tightened its restrictions on estimates and continued to argue against estimates in Union Carbide.

Thus, in the first RCCATG in 2005, the LMSB told its agents that estimates in R&D credit calculations were permissible, but only when documentation was unavailable and, where needed, estimates required factual support. 26 But by 2008, the Audit Techniques Guide stated that the IRS was not required to accept estimates or extrapolations, and it proposed extensive and burdensome information document requests reinforcing this. In Union Carbide, the IRS similarly complained that UCC “made unreliable assumptions and estimations to calculate its claimed credits” 27 and argued that “estimates are legally impermissible.” 28

Rejecting this position by the IRS, the Tax Court criticized the taxpayer’s use of estimates at only one point in the opinion, otherwise allowing estimates at numerous junctures. In fact, Union Carbide can be read to endorse a view expressed in the 2005 RCCATG: “Audit adjustments based solely upon critiques of the taxpayer’s methodology and prepackaged submissions, in many cases, stand little chance of being sustained in Appeals or in court.” 29 That was the IRS’s litigating position in Union Carbide—pointing out many discrete problems with UCC’s methodology but proposing nothing as an alternative that would be better. And, as the 2005 RCCATG predicted, that approach was ultimately unsuccessful.

How Important Is Base Period Consistency?

Taxpayers using the “traditional” method of calculating their R&D credit are faced with computing a fixed base percentage, which compares their gross receipts with their QREs for four years between 1984 and 1988. 30 Obviously records from 1984 to 1988 are often difficult to locate, particularly for QREs. 31 As a consequence, base period records are often less reliable than records available for the credit or claim years.

Recognizing the difficulties involved in gathering the evidence for a base period calculation, the IRS often focuses its attention on such calculations on examination, and it did so in Union Carbide as well.

Compounding taxpayers’ difficulties in using the traditional method of calculating the R&D credit is the consistency requirement, which requires taxpayers to retroactively “correct” their base period QREs and gross receipts to be consistent with changes reflected in their credit year(s). 32 As stated in the regulations, this means altering the base period figures to reflect changes in the law in effect for the credit year, as well as changes in the taxpayer’s interpretation of the law in the credit year. 33

In Union Carbide, UCC used two different teams and two different methodologies to calculate QREs in the base period and the credit years. The court determined that this was not fatal to its case:

Neither section 41(c)(4) nor section 1.41-4(d), Income Tax Regs., imposes any requirement that a taxpayer use the same types of documents to identify qualified research in the base period as it used to identify qualified research in the claim year if the taxpayer can otherwise show that it has satisfied the consistency requirement. 34

UCC showed that it had “otherwise satisfied” the consistency requirement with oral evidence.

Cohan Reaffirmed and Applied

Cohan stands for the general proposition that if a taxpayer has an activity that requires certain expenses, but lacks documentation for those expenses, a court cannot completely disallow those expenses. In the R&D credit context, if the taxpayer has clearly done research (it has new products, processes, or significant improvements in either), difficulties with documentation cannot doom the entire credit.

Cohan’s continued validity in the R&D context is highlighted by its citation in Union Carbide to “accept petitioner’s close approximation of all of the qualified research activities.” 35 The question is what is more appropriate: Try to work with the taxpayer’s best efforts at calculations or reject those efforts completely?

[G]iven petitioner’s expansive reading of section 41(b) we find it highly unlikely that petitioner’s calculation . . . understates the amount of QREs that were actually incurred. Accordingly, under the principles provided in Cohan v. Commissioner, 39 F.2d at 543-544, we find it more appropriate to accept petitioner’s calculation . . . than to reject petitioner’s efforts as a whole. 36

Caveats Highlighted by Union Carbide

While the Union Carbide opinion provides good news to taxpayers, UCC ultimately lost the case, and the opinion offers several caveats.

Regs. Sec. 1.41-4(a)(5)(i) states that a taxpayer need identify only one alternative to test when doing research—after all, why keep testing if you make something that works? If the process the taxpayer is using could have evaluated other alternatives, the taxpayer is doing a “process of experimentation,” one of the statutory requirements for the R&D credit.

The Tax Court held, with respect to a particular chemical used in UCC’s production processes, that if a single alternative is tested, the testing must go beyond simply determining that it works. There must be a systematic analysis of data so that, if the single alternative fails, comparisons with other alternatives could have been made. Because for this one compound UCC determined only that the compound worked as expected, the Tax Court held that there was insufficient process of experimentation to qualify for the credit. 37

In addition, throughout the Union Carbide opinion the Tax Court cites, although without commenting negatively, numerous instances in which UCC overincluded supplies costs in its credit claim. When discussing a particular process, the opinion regularly points out where UCC claimed more for experimental supplies than it actually produced. 38 Later in the opinion, the judge cited the fact that “petitioner has not allocated its claimed QREs between the experimental process business components and the nonexperimental product business components” 39 as a major reason for his denial of additional supplies QREs.

Finally, at several places in the Union Carbide opinion the judge pointed out that there was no separate budgeting for the supplies UCC sought to claim as additional QREs. Further, UCC had lost its R&D budget documents for all but one base period year and both credit years. The IRS complained about the lack of R&D budget documents.

Practice tips: Find out what the taxpayer’s R&D department does to justify its existence and how it accounts for its payroll and other costs. Preserve any such documentation or ensure that the client creates and preserves such records going forward. Also, be wary of “single alternative” situations, and bolster all evidence of testing and evaluation to pass the process of experimentation requirement. Further, if supplies are a major factor in a taxpayer’s QREs, the practitioner would be well advised to have some rational means of segregating production supplies from experimental supplies and to be wary of overestimating them, as happened to UCC. Finally, if a taxpayer has an R&D budget with accountability in place for amounts (wages, supplies, contract research) spent on R&D, that fact should be highlighted and documentation preserved for eventual IRS examination. If a taxpayer is not doing such budgeting, advisers should strongly encourage it.

Concept vs. Production

In Union Carbide, the Tax Court apparently had no problem with efforts to “scale up” a pilot process qualifying as R&D. 40 It also was apparently not upset by UCC’s customer motivations (quality meeting customer demands, quantities in otherwise experimental production runs meeting certain quotas), ruling that the primary goal in all the production runs was to eliminate uncertainties in process design. 41 However, UCC lost its case because the Tax Court, citing its 1964 decision in Mayrath, 42 held that while the costs to design the concept of an experimental business component were deductible under Sec. 174, the costs to actually build it out were not. However, an argument can be made that Mayrath has no continuing applicability to the R&D credit after the 1981 enactment of Sec. 41 and accompanying legislative history. The credit was enacted explicitly endorsing prototype material costs—build-out costs—as creditable. 43


Union Carbide will likely be appealed. However, it is unlikely that the court’s reasonable consideration of the “evidence as a whole” and its unwillingness to let the credit be completely disallowed for technical flaws in an overall reasonable calculation methodology will be reversed. This does not mean that the IRS is going to adopt the court’s evidentiary standards wholesale as its own in audits and appeals.

Practice tip: The IRS continues to quote Eustace in examination and appeals and has even cited McFerrin’s reversal as a taxpayer loss. 44 When this occurs, the practitioner should ask the agent, the Appeals officer, or the examination manager for a written determination from the National Office explaining why Eustace (specifically the trial court opinion in Eustace) has any validity as an authority, as that term is defined by Regs. Sec. 1.6662-4(d)(3)(ii). The practitioner should also ask how the government’s complete reversal on the law (as occurred in McFerrin) can have any negative consequence for taxpayers.

The reaffirmance by Union Carbide (and in the Fifth Circuit’s reversal of McFerrin) of Cohan as a viable precedent is reason for optimism, but it is unlikely that the IRS will immediately give up on this issue.

Union Carbide is not a license to prepare sloppy R&D studies, nor is it permission for practitioners to relent on demanding from their clients as much documentation of their R&D activities in their base periods and credit years as is practical. However, it does endorse the notion that a sensible, reasonable approach to capturing (or even estimating) QREs should not doom an entire R&D credit claim from the beginning. Since the testimony of outside consultants and CPAs was so integral to its decision, Union Carbide should dispel the IRS’s position that R&D credit studies performed by outside consultants somehow lack validity. Perhaps this will cause the IRS to be more reasonable with Sec. 41 in its auditing and appellate functions.

It also may be that the McFerrin reversal and the FedEx decision will cause the IRS to better educate the attorneys representing it in court as to the true state of the law and regulations. Finally, with several courts’ recent analyses of taxpayers’ prototypes as qualifying for the credit, it is long past time that the IRS conformed its position on this issue to the clear statements in the statute’s initial legislative history.


1 The “contemporaneous documentation” requirement appeared in December 2000, when Treasury issued final Sec. 41 regulations in T.D. 8930. These regulations were “final” only briefly; the IRS announced less than a month later in Notice 2001-19 that it was reviewing the regulations, and it suspended the effective date of the regulations pending that review.

2 T.D. 8930.

3 Eustace, T.C. Memo. 2001-66, aff’d, 312 F.3d 905 (7th Cir. 2002).

4 McFerrin, No. H-05-3730 (S.D. Tex. 5/25/08), rev’d, No. 08-20377 (5th Cir. 6/9/09). It should be reasonable to expect the IRS to discontinue using the McFerrin case for its continued audits.

5 Research Credit Claims Audit Techniques Guide (RCCATG): Credit for Increasing Research Activities §41, LMSB-04-0508-030 (May 2008),,,id=183208,00.html.

6 “Industry Director Directive #2 on Research Credit Claims,” LMSB-4-0608-035 (January 15, 2009),,,id=202712,00.html.

7 Union Carbide Corp., T.C. Memo. 2009-50. All references to the opinion herein will be to the 298-page slip opinion.

8 FedEx Corp., No. 08-2423 (W.D. Tenn. 6/9/2009), Order Granting Plaintiff’s Motion for Partial Summary Judgment. This “hard” discovery test had its origins in United Stationers, Inc., 982 F. Supp. 1279 (N.D. Ill. 1997), aff’d, 163 F.3d 440 (7th Cir. 1998), cert. denied, 119 S. Ct. 2369 (1999). The author briefed and argued United Stationers for the IRS in the District Court in 1996 and 1997.

9 T.G. Missouri, 133 T.C. No. 13 (2009).

10 Trinity Indus., Inc., No. 3:06-CV-00726 (N.D. Tex. 1/29/10).

11 Eustace at 14.

12 Id. at 15.

13 Cohan, 39 F.2d 540 (2d Cir. 1930), in which the Second Circuit held that if it was clear a taxpayer had engaged in an activity that demanded paying certain expenses, it was impermissible to completely deny deduction of those expenses because the taxpayer retained no documentary evidence. The case further held that it was the trial court’s duty to make an estimate of the undocumented expenses, with any inexactitude weighing against the taxpayer.

14 Eustace at 16 (emphasis added).

15 The Seventh Circuit’s affirmance of Eustace did not comment on the lower court’s refusal to apply Cohan or on the state of the evidence at trial. The Seventh Circuit primarily stated that it was standing by its earlier interpretations of Sec. 41 in United Stationers and in Wicor, Inc., 263 F.3d 659 (7th Cir. 2001), interpretations that have been superseded by subsequent Treasury regulations.

16 McFerrin at 11.

17 2001 proposed regulations, 66 Fed. Reg. 66,353, finalized in 2004 by T.D. 9104.

18 McFerrin at 12.

19 This is an important liberalization of the rules since United Stationers. Very little business research begins with the conviction that something cannot be done at all. Usually it begins with a conviction that it can be accomplished but the researchers just do not know how yet (method uncertainty) or what design will be most efficient to accomplish their aims (design uncertainty).

20 McFerrin at 11.

21 In fact, “high threshold of innovation” appears nowhere in the statute or regulations. It currently exists as a standard only in the committee reports for the Tax Reform Act of 1986, P.L. 99-514, and only in the context of software, not chemical engineering. See H.R. Rep’t No. 426, 99th Cong., 1st Sess., 182 (1985), and S. Rep’t No. 313, 99th Cong., 2d Sess., 694–95 (1986).

22 McFerrin at 9.

23 Union Carbide at 243–44.

24 Id. at 284.

25 Research Credit Claims Audit Techniques Guide (2008), chapter 1.

26 Credit for Increasing Research Activities (i.e., Research Tax Credit) Audit Technique Guide (June 2005).

27 Union Carbide at 186.

28 Id. at 254.

29 2005 Audit Technique Guide, n. 26 above, §2.c.

30 Sec. 41(c)(3)(A).

31 Gross receipts should not be as much of a problem because many taxpayers maintain their financial records forever or at least keep tax returns for decades.

32 Sec. 41(c)(6); Regs. Sec. 1.41-3(d)(1).

33 The examples given in the regulations include a change (in the credit year) in the statutory definition of qualified research and a change (in the credit year) in what the taxpayer was claiming as qualifying research (Regs. Sec. 1.41-3(d)(1), Examples (1) and (2)).

34 Union Carbide at 252–53.

35 Id. at 271.

36 Id. at 295.

37 Id. at 226–27.

38 Id. at 98–106.

39 Id. at 282–83.

40 Id. at 235, discussing the design of a commercial-scale polyethylene production process.

41 Id. at 236.

42 Mayrath, 41 T.C. 582 (1964), aff’d, 357 F.2d 209 (5th Cir. 1966).

43 Congress intended for the R&D credit to be more generous to taxpayers than the Sec. 174 deduction. The Sec. 41 legislative history explicitly addressed the costs of producing an experimental business component and included in the list of supplies that were eligible for the credit “supplies used . . . in the machining by a machinist of a part of an experimental model” (H.R. Rep’t No. 201, 97th Cong., 1st Sess. 118 (1981)). Arguably, after 1981 and the enactment of Sec. 41 the distinction between concept cost and production cost—at least for the experimental part of a business component—has disappeared completely.

44 For example, the author recently received a letter from the LMSB relying on Eustace’s refusal to apply Cohan and citing the Fifth Circuit reversal of McFerrin as a reason to completely disallow an R&D credit.


Gerald Parshall is of counsel with Johanson Berenson LLP in Great Falls, VA. For more about this article, contact Mr. Parshall at

Tax Insider Articles


Business meal deductions after the TCJA

This article discusses the history of the deduction of business meal expenses and the new rules under the TCJA and the regulations and provides a framework for documenting and substantiating the deduction.


Quirks spurred by COVID-19 tax relief

This article discusses some procedural and administrative quirks that have emerged with the new tax legislative, regulatory, and procedural guidance related to COVID-19.