Research Credits on the Farm

The IRC § 41 credit for increasing research activities (research credit) is available to compensate taxpayers for increases in qualified research. The credit is designed to reward technological innovation, with the goal of stimulating long-term economic growth. The research credit is a business credit under IRC § 38. Congress created the credit in 1981 but has modified it many times since. 

In recent years, promoters have increasingly marketed the availability of the research credit to agricultural producers. It appears that some promises being made to farmers regarding their eligibility for the research credit do not align with the law. Although some farmers will qualify for the research credit, many likely will not, especially for past research. The credit is very technical, requiring qualified research conducted through a process of experimentation grounded in the physical or biological sciences, engineering, or computer science. It is not a vehicle to receive tax benefits for ordinary farming activities. Farmers should generally dismiss claims they are entitled to a large research credit for research they didn’t know they were doing. As one tax court judge recently asked, “Which came first, the research or the research credit study?”[i]  If it’s the latter, the credit will likely be denied. Like other tax benefits, the taxpayer bears the burden to prove they are entitled to the credit. Because credits are a matter of legislative grace, courts generally require taxpayers to clearly establish their entitlement to them.[ii]

Here we review the technical requirements for the research credit, warning farmers to avoid claiming credits that do not fit within these parameters. At the time of this writing, a key decision is pending from the tax court. We will bring you further updates as the law develops.

Research Credit Requirements

To qualify for the research credit, the taxpayer’s research must be “qualified.” This means it must satisfy a four-part statutory test[iii]:

1. The expenditures tied to the research must be eligible to be treated as expenses under IRC § 174A.
2. The research must be undertaken for the purpose of discovering information which is technological in nature.
3. The application of the research must be intended to be useful in the development of a new or improved business component of the taxpayer, and 
4. Substantially all of the activities must constitute elements of a process of experimentation for developing a “a new or improved function, performance, reliability or quality.”

Each of these requirements may be explained further.

1. Expenses under IRC § 174A

Section 174A[iv] requires research activities to be conducted for a taxpayer’s business in the “experimental or laboratory sense.” and intended to “eliminate uncertainty” concerning the development or improvement of a “product.”[v] A “product” includes a pilot model, a process, a formula, an invention, a technique, a patent, or similar property.[vi] Only research activities aimed at resolving “uncertainty” in the development or improvement of a product, and the associated expenditures, qualify for the credit.[vii] Uncertainty exists if the information available to the taxpayer does not establish the capability or method for developing or improving the product or the appropriate design of the product.[viii] Costs paid to construct a product after the elimination of uncertainty are not eligible expenses.[ix]

Several types of expenses are automatically excluded by regulation[x]:

• Ordinary testing or inspection of materials or products for quality control 
• Efficiency surveys
• Management studies
• Consumer surveys
• Advertising or promotions
• The acquisition of another’s patent, model, production or process
• Research in connection with literary, historical, or similar projects

Section 174A applies to research or experimental expenditures only to the extent that the amount of the expenditure is reasonable under the circumstances.[xi]

Expenditures for the acquisition or improvement of land, or for the acquisition or improvement of property which is subject to depreciation are not deductible.[xii]

2. Technological In Nature

Research must be undertaken for the purpose of discovering information that is "technological in nature."[xiii] Information is technological in nature if "the process of experimentation used to discover such information fundamentally relies on principles of the physical or biological sciences, engineering, or computer science."[xiv] This test does not require the taxpayer to rely on new applications of science. Instead, a taxpayer may rely on existing principles of science and engineering to satisfy this requirement.[xv]

Research is undertaken for the purpose of discovering information if it is intended to eliminate uncertainty concerning the development or improvement of a business component. Uncertainty exists if the information available to the taxpayer does not establish the capability or method for developing or improving the business component, or the appropriate design of the business component.[xvi] This requirement can be met even if the taxpayer fails in developing a new or improved business component.[xvii]

3. New or Improved Business Component 

A “business component” is any product, process, technique, formula, or invention which is to be held for sale, lease, or license, or used by the taxpayer in its trade or business.[xviii] Products and processes are treated as separate business components.[xix] Product research is undertaken to improve the quality of the product itself while process research includes searching for ways to produce the same product in a greater quantity or at a lower cost.[xx] If a taxpayer produces a product as part of its trade or business, the taxpayer's search for a way to produce the same product in greater quantity or at lower cost may be qualified research on the production process, but not on the product itself.[xxi] Any process, machinery, or technique for commercial production of a business component shall be treated as a separate business component (and not as part of the business component being produced).[xxii]

After a taxpayer establishes which business component it sought to develop, the business component test requires the taxpayer to show that they intended that the discovered information would be useful in developing a new or improved business component.[xxiii] To be useful within the meaning of this test, the research must provide some level of functional improvement to the taxpayer.[xxiv]

4. Process of Experimentation

The process of experimentation test requires substantially all research activities to constitute elements of a process of experimentation for a qualified purpose.[xxv] This means that 80 percent or more of the taxpayer’s research activities for each business component, measured on a cost or other consistently applied basis, must constitute a process of experimentation.[xxvi]

A process of experimentation is a process designed to evaluate one or more alternatives to achieve a result where the capability or method or appropriate design is uncertain when the taxpayer begins the research activities.[xxvii] While the same “uncertainty” is required for this test as is required for the Section 174A test, the process of experimentation test requires the taxpayer to seek to resolve uncertainty using principles of the physical or biological sciences, engineering, or computer science.[xxviii] A process of experimentation must be an evaluative process and generally should be capable of evaluating more than one alternative.[xxix] The experimentation must be systematic. For example, it may involve modeling, simulation, or a systematic trial and error methodology.[xxx] Informal or ad hoc trial and error is not sufficient. The process of experimentation requires the use of a systematic evaluative process consistent with principles of the scientific method.[xxxi] A taxpayer may undertake a process of experimentation if there is no uncertainty concerning the taxpayer's capability or method of achieving the desired result so long as the appropriate design of the desired result is uncertain.[xxxii]

Applying the Four-Part Test

The four-part qualified research test is applied separately to each business component.[xxxiii] If a component fails any part of the test, the test may be applied to a subset of the product or process. This is called the shrinking-back rule.[xxxiv] If the qualified research tests are not satisfied at the level of the discrete business component, then they are applied at the most significant subset. The application of this rule continues until either a subset satisfies the four-part test, or the most basic element of the business component is reached and fails to satisfy the test.[xxxv]

Qualified Research Expenses

Qualified research expenses (QREs) comprise in-house research expenses and contract research expenses "paid or incurred by the taxpayer during the taxable year in carrying on any trade or business.” [xxxvi]

In-house research expenses are (1) "any wages paid or incurred to an employee for qualified services performed by the employee" (qualified services) and (2) "any amount paid or incurred for supplies used in the conduct of qualified research" (qualified supplies).[xxxvii] Qualified services include either (1) engaging in qualified research or (2) engaging in the direct supervision or direct support of qualified research.[xxxviii] Generally, wages are in-house research expenses if they were paid for qualified services of an employee.[xxxix] Qualified supplies include all tangible property other than land, improvements to land, or depreciable property.[xl] 

Contract research expenses include 65 percent of amounts paid to non-employees to conduct qualified research.[xli] This limit is increased to 75 percent if the payment is made to a qualified research consortium.[xlii] This is a nonprofit (not a foundation) organized and operated primarily to conduct scientific research.[xliii]

Calculating the Credit 

Companies may choose from two options when calculating the research credit: the regular credit[xliv] or the alternative simplified credit[xlv] (ASC). Both formulas are designed to capture only “increasing” research activities.

The calculation for the regular credit is complex. It is equal to 20 percent of the company’s current-year QREs that exceed a “base amount.”[xlvi] The base amount is the product of the taxpayer’s “fixed-base percentage” and its average annual gross receipts for the four years before the credit year.[xlvii] If the company had QREs and gross receipts between 1984 and 1988, the fixed base percentage is the ratio of the company’s aggregate QREs to aggregate gross receipts from that period.[xlviii] The percentage cannot exceed 16 percent.[xlix] If the company began after 1988, the fixed base percentage is established using a formula applying to the company’s first 10 years with QREs and gross receipts.[l] The percentage is capped at 3 percent for the first five of those 10 years.[li] Even so, the company’s base amount cannot be less than 50 percent of current year QREs.[lii] As such, the regular credit rarely exceeds 10 percent.[liii]

 Under the ASC[liv], the credit is equal to 14 percent of the excess of the taxpayer's qualified research expenses (QREs) in the credit year over 50 percent of the average of the taxpayer's QREs from the three preceding years.[lv] If the taxpayer has no qualified research in the three prior years, or if they are unable to prove the qualified research expenditures for those years, the credit cannot exceed 6 percent of the QREs in the credit year.[lvi] 

Recent Agricultural Cases

In two recent tax court cases, a judge applied these rules to agricultural production activities: one case involved poultry production and the other involved crop production. In neither case was the taxpayer a typical farmer. In George v. Commissioner,[lvii] the taxpayer was one of the largest fully integrated poultry processing companies in the United States. In Boswell v. Commissioner,[lviii] the taxpayer was an agricultural company farming over 135,000 acres over two ranches. In addition to commercial crop production, Boswell used its land for agronomic research involving tomato, cotton, and safflower plants. In both cases, the IRS denied a significant portion of the taxpayers’ research credit claims.

George v. Commissioner

On February 3, 2026, the U.S. Tax Court slashed the taxpayer’s alleged QREs by 77 percent, disallowing a significant portion of the research credit claimed for the 2012, 2013, and 2014 tax years.[lix] Although the taxpayer had ample contemporaneous documentation of standard production practices, there was insufficient substantiation supporting most of the claimed research. The court found that some of the research trials were constructed after the fact, fashioned to support a research credit study.[lx]

The taxpayer in George was a large poultry production conglomerate. Although operated through multiple entities, the company was involved in broiler production from the breeding, hatching and growing stages through processing. Although the company contracted with growers to raise their broilers, it remained highly involved in the life cycle of each broiler, which the company continued to own. The company raised large broilers and small broilers, each for a different market. It then processed the broilers for delivery to its customers, such as Kentucky Fried Chicken. Each piece of meat had to fit within a narrow range of sizes to allow the customers to mass produce their end products. The company processed approximately 3.5 million broilers each week. If everything went according to plan, the taxpayer earned approximately one cent of profit per pound of each broiler.  

Small broilers were generally sent for processing at 35 to 37 days old and large broilers were on the farm for around 60 days. During this time, the company monitored the broilers intensely. Although some broilers were raised by outside growers the company sent service technicians and veterinarians to the farms to monitor and support the growers. The technicians regularly sampled broilers for disease, reviewed mortality logs, and administered vaccines and medications. All food for the broilers was supplied by the company feed mill, which provided different recipes for different flocks, corresponding to age and nutritional needs. The feed was the most expensive part of raising broilers.[lxi]

Because of the high-volume, low margin operation, data was crucial to the taxpayers’ operation. Decisions were based upon data collected. Broiler health was paramount. The live production account manager was tasked with collecting, tracking, and analyzing all data associated with the production process. The data collection and recording associated with production, including total costs, shrinkage, and health of the birds, was immense.[lxii] It determined grower compensation and was used by the company and its veterinarians to gauge performance. 

In 2014, the company’s long-time accounting firm proposed to have alliantgroup review the company’s financials to determine if it was eligible for research credits. The company concurred with the recommendation and engaged alliantgroup to perform a research credit study.[lxiii] After extensive interviews and data analysis of past operations, alliantgroup determined that between 2012 and 2014, the company had conducted 7 research trials and was entitled to $4,471,598 in research credits, based upon $62,954,355 in QREs.[lxiv] The QREs were primarily feed expenses (supplies). Although alliantgroup claimed it found qualified activities and related services, it did not calculate qualified service expenses because the process to determine the value of the credit was not worth the effort to allocate the wages.[lxv] Alliantgroup prepared pro forma Forms 6765, which reported the research credits, and the company’s accountants filed amended returns to claim them. 

Upon audit, the IRS found no qualified research, denied the credits and imposed accuracy-related penalties.[lxvi] The taxpayer challenged the IRS’ determination, arguing the research was valid. On review, the court applied the four-part test to each of the seven trials. Although it found some qualified research, it slashed the QREs to $14,315,418 (22.7 percent of the company’s originally claimed QREs). We review each of the seven studies to understand how the court analyzed the credit claims.  

1. Salinomycin (parasite control)[lxvii]

The company sought credits for alleged 2012 research designed to determine whether higher doses of an antibiotic called salinomycin in the feed would better control coccidiosis, a destructive parasite, and improve broiler quality. 

Despite the research credit study’s “promising” narrative, the court found no evidence that the feed recipes provided to the birds during 2012 included altered dosages of salinomycin. In fact, the extensive data collected during that period contained no evidence of any meaningful experimentation or analysis designed to resolve “uncertainty” regarding the use of salinomycin. The court found this was a “clear example” of a research credit study attempting to transform routine production activities and data collection into qualified research.

No QREs were allowed for this activity.

2. HatchPak + Tylan (vaccine + antibiotic program)[lxviii]

The company sought credits for research designed to evaluate whether combining a coccidiosis vaccine (HatchPak) with Tylan (an antibiotic) could control disease and improve its poultry product. The company claimed to conduct this research in 2012 and 2013.

The court agreed, based upon the data collected, that the company had conducted qualified research with respect to this question for the year 2012. The company was able to identify, through contemporaneous documentation, the experimental flocks and the feed costs for those flocks. However, the court disallowed any qualified research expenses for 2013, finding that the uncertainty as to the combined use of these drugs was resolved by the 2012 study. 

The allowed QREs for the 2012 activity were $5,115,281. No QREs were allowed for 2013.

3. Probiotics[lxix] 

The company sought credits for research designed to find alternatives to antibiotics by testing probiotics to improve gut health, feed conversion, and bird mortality. 

The court found that this research passed the four-part test for the 2013 tax year. This was research designed to produce an improved product—birds with better gut health that were less reliant on antibiotics. This research, the court found, also improved the product because consumers considered birds raised without antibiotics to be of superior quality. The experimentation—conducted with test flocks—resolved uncertainty as to whether the use of probiotics could produce a better product. Contemporaneous documentation supported the claims.

The allowed QREs for the 2013 probiotic research trials were $7,280,578.

4. Phytase (enzyme for phosphorus digestion)[lxx]

Broilers need phosphorus for proper bone density. Although corn contains phosphorus, broilers are not generally able to absorb the phosphorus in corn so expensive phosphorus supplements must be added to the feed. The company sought credits for research it purportedly conducted in 2012 to determine whether adjusting levels of the enzyme phytase in the feed could improve phosphorus absorption from the corn and reduce costs. 

Despite another “promising” research credit study narrative, the court found that the company’s extensive feed records did not show that the dosage of phytase had been meaningfully modified during the years of the alleged research. There was no evidence of investigatory activity or a process of experimentation. 

No QREs were allowed for this activity,

5. LT (laryngotracheitis vaccine methods)[lxxi]

Laryngotracheitis (LT) is a highly contagious, often fatal, respiratory infection that can decimate broiler flocks. The company sought credits for research to test different LT vaccination strategies. In 2012 and 2013, the company claimed credits for research to see if the delivery method of the vaccine would increase its effectiveness. For example, the company tried to administer the vaccine in drinking water and by using a laser pointer as opposed to the standard method of spraying the flock. In 2014, the company sought to test whether priming birds with a vector vaccine before immunizing them with the standard vaccine would reduce the side effects of the vaccine and improve the quality of the broilers. 

The court found no evidence that the use of alternative vaccine administration methods was part of an actual research trial. Evidence did not substantiate a timeline for the activities or even pinpoint the year(s) of the activity. Thus, the court found no qualified research for 2012 and 2013, stating this was an “example of the chicken (research credit study) coming before the egg (the research).”[lxxii] The court did find that the 2014 priming activity was qualified research. The experimentation with the priming showed the birds had less severe side effects, leading to a healthier flock. The results were well documented, and preexisting uncertainty regarding this question was resolved.

The allowed QREs for the 2014 priming research totaled $1,521,039.

6. Vaxxitek (IBD vaccine)[lxxiii]

IBD is a disease that destroys the immune system of broilers. The company sought credits for purported company research designed to determine whether a lower than recommended dosage of Vaxxitek, an IBD vaccine, would prevent IBD and reduce costs. This activity was conducted between 2012 and 2014. 

The court found no data in the record to establish that the vaccine was administered in varying dosages to the flocks during the research years. There was also no evidence that the company had set aside an experimental flock or had administered the vaccine differently across different broilers. Because there was no uncertainty regarding whether the vaccine itself was effective, administration of the vaccine itself did not constitute a research activity. 

No QREs were allowed for the vaccine activities. 

  7. Ross 708 (Overall “improved poultry product”)[lxxiv]

The company sought credits for 2014 research designed to determine whether a new genetic line would perform better as a big broiler than the company’s current genetic line. 

The court found that the activities involved a process of experimentation designed to resolve uncertainty as to whether the Ross 708 line would outperform the Cobb 500 line. The activities used experimentation flocks and control flocks. The employees collected extensive contemporaneous data and compared the quality of the final broilers from each genetic line. The process was an example of using the scientific method—a hypothesis was formed and tested and the results analyzed. 

The allowed QREs from this trial totaled $398,520.

Court’s Credit Calculation

In total, the court found that the company proved QREs totaling $14,315,418[lxxv] (22.7 percent of those originally claimed). The court then turned to the credit calculation. Because the company used the ASC method, it was required to establish QREs for the three prior tax year (base years) to calculate its credit. The court found that the company failed to substantiate the base year QREs.[lxxvi] Instead, the company’s advisors had merely estimated the QREs for the base years. They looked at the research years, calculated a percentage of feed expense tied to the research, as compared to overall cost, and applied that same percentage to the base years. The court rejected this approach, finding that the company failed to provide any basis for a reasonable estimate for qualified research during the three base years. The court stated that because the “research credit study grossly overvalued the amount of QREs for the research year,”[lxxvii] it was inappropriate to rely on that data to estimate research expenses from prior years. 

The court explained that if a taxpayer provides enough evidence to support an informed estimate, a court can—but is not required to—apply the Cohan rule[lxxviii] to estimate QREs. But here, the court found no such informed estimate. Instead, the company merely concluded that its perpetual striving to improve performance involved qualified research and relied on vague documentation and employee recollection to support that. The court said this is not acceptable. The court explained, for example, that a few words in a board report are insufficient to show that activities were qualified research. Qualified research requires a process of experimentation relying on the scientific method. Because the company failed to substantiate research activities during the three base years, the limit of IRC § 41(c)(5)(b) applied, and research credits could not exceed 6 percent of the QREs proven for each taxable year (2012, 2013, and 2014).[lxxix]  

Accuracy-Related Penalties

Because the taxpayer “grossly overvalued” the research credits to which they were entitled in 2012 and 2014, the IRS argued that the company should be liable for the 20 percent accuracy-related penalty.[lxxx] This penalty may be imposed where negligence or disregard of rules or regulations leads to a substantial understatement of income tax.  The penalty does not apply if it is shown that there was a reasonable cause for the taxpayer’s position and that the taxpayer acted in good faith. A taxpayer’s reliance on professional advice may meet the reasonable cause and good faith standard if (1) the adviser was competent and had sufficient expertise to justify reliance, (2) the taxpayer provided accurate information to the adviser, and (3) the taxpayer relief in good faith on the adviser’s judgment.[lxxxi]

Here, the court found that the company reasonably relied in good faith on the advice of alliantgroup. As such, the court did not impose accuracy related penalties. This decision was not a general endorsement of the work of alliantgroup. Instead, it was a recognition that at the time of the engagement, alliantgroup had more than 12 years of experience conducting tax credit studies for clients and it employed skilled tax attorneys and policy experts. Its “thousands of employees,” for example, included a former IRS commissioner and former tax counsel to the U.S. Senate Finance Committee. The court also noted that the services of alliantgroup were suggested by the company’s long-time trusted accounting firm. Alliantgroup “diligently” requested and reviewed documents and met with employees to determine the company’s eligibility for the credit. As such, the court found that the company acted in good faith in relying upon the advice of alliantgroup for determining its eligibility for this complex credit.

The IRS pointed out that in Betz v. Commissioner, T.C. Memo 2023-84, at *113, n. 53, the tax court stated that “any apparent reliance by [the taxpayers] on [a]lliantgroup with respect to claiming the research credits was inconsistent with ordinary business care and prudence and thus that [the taxpayers] failed to establish reasonable care for their underpayments of tax.”[lxxxii]

Here, the court clarified that the court has “not established a hardline rule that reliance on alliantgroup can never be reasonable or in good faith.”[lxxxiii] The court noted that in Suder v. Commissioner, T.C. Memo. 2014-201, at *77-78, the court found good faith reliance on alliantgroup. The court then noted that Betz was a more recent case than Suder and that Suder was the only decided case at the time the company in George relied on alliantgroup’s advice. 

Lessons from George

Seeking Research Credits for Past Research Is a Challenge

The taxpayer here was a huge poultry conglomerate processing 3.5 million heads of broilers a week. The company was innovative and progressive. It was always seeking to improve the quality of its poultry and its bottom line. It regularly employed new vaccines or methods of raising healthy birds. It regularly adjusted feed additives and supplements. It regularly collected extensive data associated with its many production activities. Even so, the court found there was insufficient data to prove that most of the company’s research efforts employed the scientific method. The research credit study narrative was persuasive, but the contemporaneous documentation, in most cases, did not support the research. Most so-called experiments had no test flock or control group, and the data, in most cases, did not reveal a process of scientific experimentation to resolve uncertainty. The court made much of the fact that this taxpayer’s research study looked backwards to retrofit already-conducted research into a research credit paradigm. “Which came first, the research or the research credit study?” Where the court found contemporaneous records documenting scientific research, it allowed the QREs. But where the contemporaneous data was lacking, it denied them. Ultimately, the court determined that the company “grossly overvalued” its QREs.

Lesson: Most agricultural operations are not as complex as the George conglomerate. Most do not conduct the extensive trial and error George employed to maintain the health of its millions of birds and increase the cost-effectiveness of raising them. Despite the sophistication of the George operation and the troves of data accompanying its processes, the company was unable to secure most of its claimed research credits. As the George court explained, the research credit requires a documented process of scientific experimentation designed to resolve uncertainty. Once the uncertainty has been resolved, activities related to that uncertainty no longer qualify as research. It is not enough to write a compelling research credit study narrative explaining how this research occurred. Contemporaneous records and data must show exactly how the taxpayer sought to resolve uncertainty using the scientific method. Collecting this proof after the fact will be an insurmountable challenge for most farm operations. Farms that want to claim the research credit should investigate its requirements before conducting the research.

Base Year QREs Must be Proven

Here, the taxpayer sought to calculate the research credit using the ASC method. As explained above, the ASC calculation is 14 percent of the current QREs over 50 percent of the QREs from the prior three years. The prior year QREs are called the base year QREs. If the taxpayer cannot prove the base year QREs, the credit cannot exceed six percent of the current year QREs. In other words, if a taxpayer proves $100,000 of current-year QREs but cannot substantiate base-year QREs, the maximum ASC credit may be limited to $6,000.Here the company sought to estimate the base year QREs using the Cohan rule. While the court said it was possible for a court to allow an estimation, that was only possible where there was a reasonable basis for such an estimate. In other words, the company must offer evidence to support an informed estimate. Here, the company used current year QREs to estimate QREs from the past. This was unacceptable. Actual data supporting base year QREs must be submitted. 

Lesson: If an agricultural producer has not been engaging in documented research, they should not attempt to create it for the purpose of supporting base year QREs. The IRS and the courts will analyze the data to ensure it supports the claim. Without contemporaneous documentation, a taxpayer using the ASC method will be limited to a maximum credit of 6 percent of QREs. 

Choose Advisers Carefully

In this case, the company relied on the advice of alliantgroup in claiming the research credit for the three tax years at issue. Although most credits were denied for two of those years, the court did not impose an accuracy related penalty because it found the company relied, in good faith, on experienced research credit advisers. The court noted that the only case law discussing reasonable reliance on alliantgroup at the time found the taxpayer had reasonably relied upon the advisers. The company here could have relied on that opinion in selecting alliantgroup for this work. Additionally, the taxpayer’s long-time accounting advisers had recommended alliantgroup for the research study. 

Lesson: Agricultural producers should carefully screen advisers engaged to help secure research credits. This includes alliantgroup. The court noted that a case published after this company’s research study found that “any apparent reliance by [the taxpayers] on [a]lliantgroup with respect to claiming the research credits was inconsistent with ordinary business care and prudence and thus that [the taxpayers] failed to establish reasonable care for their underpayments of tax.” Because of the complexity of the research credit, taxpayers are not expected to understand all the intricacies of its application. Even so, good faith reliance requires understanding the basics of the credit, seeking advisers who are competent and knowledgeable in applying the law, and being on guard against overaggressive tactics or promises that seem too good to be true. After George, good faith reliance will be hard to establish for research credit claims arising from research reconstructed after the fact without contemporaneous documentation supporting a process of experimentation. Without good faith, the penalty is 20 percent of the tax underpayment. In other words, the taxpayer pays the underpayment, with interest, plus an additional 20 percent as a penalty.

Boswell v. Commissioner

The Boswell case is ongoing. We await a trial court ruling that should clarify the distinction between research intended to produce a better crop (product research) and research intended merely to produce the same crop more cheaply or at a higher yield (process research). 

The taxpayer in Boswell operated a large farming business in the central California valley. During the tax years at issue, the company grew cotton, tomatoes, alfalfa, hay, and safflower, among other crops from which it produced and sold products including cotton lint and tomato paste.

The taxpayer conducted 33 research trials in 2013 and 22 research trials in 2014 on about 7 percent of its farmland. For purposes of this research, it grew crops on control plots, using the standard, nonexperimental method it used on its other acres. On the test plots, the researchers tried an experimental method, which they evaluated by comparing the output of the test plot to output from the control plot. The company claimed $17,062,147 in QREs in 2014, yielding a claimed research credit of $1,735,424.[lxxxiv] The company included all costs of cultivating and evaluating the research acre crops as QREs. This included supplies such as seed and fertilizer. On audit, the IRS reduced the taxpayer’s research credit to $117,722 after finding only $885,132 of QREs. Finding that the taxpayer’s research was “process” research, the IRS restricted the QREs to those incurred solely because of the research activities. Costs for supplies, such as seed and fertilizer, that would have been used to grow the crops in the absence of research activities were excluded from the QRE calculation.[lxxxv]  

Both parties sought summary judgment, and the court issued an order denying the motions on July 12, 2022. 

Court’s Summary Judgment Analysis  

The court first reviewed the law, stating that QREs arise from “qualified research,” which is research on a “business component of a taxpayer” that satisfies the criteria of IRC § 41(d)(1).[lxxxvi] The business component, the court continued, is “any product, process, technique, formula, or invention,” which is held for sale or use by the taxpayer in its trade or business.[lxxxvii] The court then explained that the statute “directs us to treat the product a taxpayer produces as a separate business component from its production process”:

Special Rule for Production Processes.--Any plant process, machinery, or technique for commercial production of a business component shall be treated as a separate business component (and not as part of the business component being produced).[lxxxviii]

Based on this distinction, the court said that if a taxpayer tests an experimental production process designed to boost yield or cut costs without altering the product itself, QREs do not include costs the taxpayer would have incurred to create the same product using the standard method. Drawing from the tax court’s memorandum opinion in Union Carbide[lxxxix], the court illustrated with an example:

[I]magine that a taxpayer tests two experimental production processes designed to improve on its standard process for producing Product X. In Test A, the taxpayer evaluates an experimental process designed to produce an improved product, Product X+. Test B, on the other hand, should yield the same Product X in greater quantity or at lower cost than the standard process. If Test B involves qualified research at all, the taxpayer conducts such research on the production process alone. The Court must allocate the taxpayer's expenditures on Test B between product and process and exclude the former from QREs.[xc]

The company argued that the statute makes no such distinction. It contended that even if the research was “process” research, the statute does not restrict QREs based upon that category. Rather, the company noted that IRC § 41(b) defines QREs to include “any wages” or “any amount paid or incurred by the taxpayer for supplies used in the conduct of qualified research.”[xci] 

The court rejected the company’s argument based upon the mandate to avoid reading the Code in a way that makes any language superfluous.[xcii] The court saw no reason for the distinction between product and process in § 41(d) if the purpose was not to restrict the scope of the QREs where the taxpayer conducts qualified research on one business component but not the other. 

The court found that § 41(d)(2)(C) applies to any research trials that focus on the production process and not the product. Even so, the court denied summary judgment to both parties, finding that it was not clear from the record which of the company’s research trials focused on the production process and which ones focused on improving its products. The court explained that one example of process research from the motion papers might be the company’s safflower experiment. Here, the taxpayer tested a way to reduce production costs by withholding nitrogen from the crop, apparently without altering the product at all.[xciii] The court noted, however, that it required further factual development to determine whether the safflower experiment or any other 54 research trials fit the paradigm of process research (as exemplified by Test B in the example above).[xciv] The court stated that the company bore the burden of establishing any disallowed QREs associated with these trials. 

A trial was held, beginning November 17, 2025, with each party presenting experts on the nature of the research conducted by the company. Opening post-trial briefs were filed March 9, 2026, and answering briefs were filed April 23, 2026. The parties have stipulated that the taxpayer’s 55 research trials meet the four-part test for qualifying research. The question before the court is the amount of allowable QREs. 

Post-Trial Arguments

Taxpayer

The taxpayer is arguing post-trial that it is entitled to the full amount of its claimed credit, regardless of the nature of its research or business component designation. The company primarily argues:

• Expenses incurred researching process business components are QREs by the plain text of Section 41. 
• Union Carbide was wrongly decided. IRC § 41(d)(2)(C) does not relate to QREs but to whether the activities constitute qualified research. This determination, the company argued, is made separately for product and process based upon the four-part test. 
• In affirming the tax court’s Union Carbide decision, the Second Circuit Court of Appeals deferred to the agency’s interpretation under Chevron. Because of Loper Bright, this interpretation cannot stand.
• The evidence establishes that the company’s plants, as products, were the business components in the research trials. They were products because they were “used by the taxpayer in a trade or business of the taxpayer.”[xcv] The plants, if not considered products, are “pilot models” for which all expenses are QREs.[xcvi]

IRS

The IRS is arguing post-trial that 47 of the company’s 55 research trials were geared toward improving its plant growing processes, not toward improving the products it sold to its customers. Thus, the IRS argues that QREs do not include costs the taxpayer would have incurred to raise the crops using the standard method. Specifically, the IRS primarily argues:

• Where a research trial is determined to be concerned with increasing yield for the company, it is a process business component and the costs associated with that research should be treated as process-related business expenses. 
• Many of the agronomy reports associated with the research trials specifically stated that the hypothesis of the trial was to grow the same product in greater quantity or at lower cost. This, the IRS urges, is process research.
• The company’s plants were not products because the company did not sell them to customers. The products were the produce or grain or fiber grown on the plants. Even if the plants were products, the research was not intended to improve the quality of the plants. 
• The research trials were not pilot models, which are under IRS regulations, a “representation or model of a product.” The research trials did not constitute a representation or model of the physical product sold to customers. They were agronomic experiments designed to reduce costs or improve yield. Producing more of a product does not increase its quality.  

Takeaways from Boswell

• Even where activities satisfy the four-part test, not all associated expenses necessarily qualify as QREs.
• Under the IRS view adopted in Union Carbide, research aimed at producing the same crop more cheaply or efficiently is generally treated as process research, while research aimed at improving the characteristics or quality of the crop itself is treated as product research. If the tax court does not change its agreement with Union Carbide, crop research intended to increase yields or reduce costs will be considered process research and only costs in excess of the expenses that would have been incurred to raise the crops using standard methods will be includible as QREs. 
• As also made clear in George, detailed contemporaneous documentation is crucial. Agronomy reports, research hypotheses, and detailed notes may become critical evidence in determining whether expenses qualify for the research credit. In this case, the IRS relied heavily on the taxpayer’s own descriptions of the agronomic research in determining it was process research. 
• Advisers should expect increased scrutiny of agricultural research credit claims. The scale of the adjustment in Boswell—from a claimed credit of more than $1.7 million to an IRS-allowed amount of roughly $118,000—demonstrates the IRS’ willingness to aggressively challenge large agricultural research credit claims. 
• The case highlights continuing uncertainty as to how IRC § 41 applies to crop production. Farmers and their advisers must monitor this case closely and be cautious about relying heavily on research credit strategies tied to ordinary crop production experimentation.

Examples

Livestock Production Example One

A swine producer works with a research credit consultant to design a research project to determine whether a new feed additive can improve gut health in weaned pigs while reducing the incidence of respiratory disease and the need for antibiotic treatment. The producer establishes control groups and test groups, develops written hypotheses, systematically tests alternative feed and barn-environment combinations, gathers health and growth-performance data, analyzes the results, and refines the protocols over multiple trial cycles.

This research likely qualifies for the research credit because it appears to meet the four-part statutory test. Because the project seeks to improve animal health outcomes and reduce disease incidence, the research is directed toward improving the quality of the livestock product itself. As such, the direct research-related costs associated with the research groups (including supply and wage expenses) should be QREs.  

Livestock Production Example Two

A cattle feeder routinely changes feed rations throughout the year to improve average daily gain and manage feed costs. A consultant reviews past feed records and says the producer qualifies for the research credit because trying different feed mixes over time was qualified research. 

The producer’s activity does not likely qualify for the research credit because the “research” was created after the fact. The producer did not identify uncertainty and did not create a research hypothesis and scientific experimentation process to address the uncertainty. Routine operational adjustments, ordinary management decisions, or incremental production optimization generally do not by themselves establish the required scientific uncertainty. Here, the producer did not establish test and control groups and did not contemporaneously document the process of experimentation. Rather, this activity resembles ordinary production optimization, not scientific research qualifying for the research credit. The George court denied the research credit for this type of activity.

Crop Farming Example One

A tomato producer designs a research project to determine whether a new growing technique can improve the quality and shelf life of tomatoes sold to processors. The producer documents uncertainty about the proposed growing method, establishes control plots and experimental plots, tests several irrigation and nutrient combinations, measures firmness and spoilage rates, and evaluates and documents results through repeated experimentation.

This activity, which appears to meet the four-part statutory test, would likely qualify for the research credit. Uncertainty exists as to a proposed growing method. The experimentation is systematic and documented. The project involves technological and biological sciences. Control plots and test plots are used to evaluate alternatives. The research is planned in advance, not reconstructed after the fact. The trials and results are well documented. This is product research designed to improve the quality of the tomatoes sold to customers. As such, the costs of the research, including the costs of raising the tomatoes in the experimental plots, should be QREs.  

Crop Farming Example Two

A corn farmer experiments with reducing nitrogen applications to lower fertilizer costs while attempting to maintain the same corn yield and quality. The farmer does not establish test plots or control plots in the current year, but he compares current yields to yields from the prior year. A year later, a consultant tells the farmer he should qualify for the research credit and offers to complete a study.

This activity should not qualify for the research credit because it violates George: research reconstructed after the fact without contemporaneous evidence of a documented process of experimentation is unlikely to satisfy the statutory requirements. Here there is no evidence that the farmer undertook a scientific process of experimentation to address the uncertainty of maintaining yield with reduced nitrogen. The farmer has little contemporaneous documentation and no evidence of trial plots and control plots separate from the production acres. Although a consultant may seek to create uncertainty, a hypothesis, and documentation of a scientific process after the fact in a well-written research study narrative, there must be contemporaneous documentation and records to support those claims. On audit, the IRS will request contemporaneous study notes and documentation, evidence demonstrating uncertainty, actual scientific experimentation (i.e. that test plots were treated differently and monitored separately from control plots), and proof that varying amounts of nitrogen were applied to each plot. 

Crop Farming Example Three

A corn farmer experiments with reducing nitrogen applications to lower fertilizer costs while attempting to maintain the same corn yield and quality. The farmer identifies uncertainty and designs a process of experimentation to resolve the uncertainty. The farmer establishes test plots or control plots in the research year and compares yields across plots. All stages of the experiment are contemporaneously documented, including the amounts of nitrogen applied to each plot, the yields of the plots, and all other factors that may have impacted the yields. The farmer wants to claim the research credit for this work.

Unlike the farmer in Crop Farming Example Two, this farmer appears to have engaged in a process of experimentation to resolve uncertainty. It is likely this research meets the four-part test. 

Even so, the credit available to this producer may be minimal. This appears to be process research conducted to reduce costs, not product research designed to improve the quality of the corn. The George and Boswell courts, relying on Union Carbide, both explained that QREs for this type of research do not include ordinary production expenses that would have occurred without the research.  Unless the Boswell court decides that Union Carbide was wrongly decided, this producer could not include as QREs ordinary production expenses that would have occurred without the research. In other words, the standard production costs, even for the experimental plots, would not be QREs.

Crop Farming Example Four

A row crop farmer raising soybeans is contacted by a consultant specializing in research credits. The consultant tells the farmer most agricultural producers are eligible for the credit because they are always doing research. The consultant requests five years of Schedule Fs and contacts the farmer several weeks later with a draft report. The narrative describes a process of research the farmer has conducted over the past several years to “test new soybean varieties” to see how they perform on the farmer’s fields. The report reviews the four-part statutory test and makes numerous claims of a systematic process of experimentation conducted by the farmer. It states that data was collected and documented and that the experiment relied on genetics and soil science to resolve the uncertainty. 

 The report then claims as QREs all seed and fertilizer costs reported on the Schedule F for the “research years.” Pro forma Forms 6765 are provided. The farmer is instructed to have their tax preparer file amended returns for the research years using the forms. 

This farmer does not appear to qualify for the research credit. The George court clearly teaches that research that qualifies for the research credit is not accidental. It is deliberately planned and executed with a specific research question in mind. There must be scientific uncertainty. Here, the farmer was engaging in ordinary farming and only thought about “research” when the research credit consultant contacted her. On audit, the IRS would request contemporaneous documentation of the process of experimentation that occurred. This might include test plot maps, trial notes, statistical analysis, control group comparisons, dated protocols, and the collected data. Without such evidence, the farmer could be liable for an accuracy-related penalty. 

Another problem with this “research” is that true seed variety research is conducted by the seed companies themselves. Farmers generally purchase seed that has been well tested for disease and drought resistance, increased yield, etc. It is unclear whether a farmer could establish that scientific uncertainty exists regarding these questions. Even if these obstacles were overcome and the farmer conducted a valid research trial, the farmer’s goal appears to be increased yield, not higher quality soybeans. The soybeans were not sold in a specialty market at a premium price. The research benefit to the farmer is more acres per bushel. This would appear to be process research, not product research. Following the Union Carbide position, which the IRS takes, the standard costs for growing the soybeans would not be QREs.

[i] George v. Commissioner, T.C. Memo. 2026-10, *2 (Feb. 3, 2026).

[ii] Helvering v. Northwest Steel Rolling Mills, Inc., 311 U.S. 46 (1940).

[iii] IRC § 41(d)(1).

[iv] The One Big Beautiful Bill Act created IRC § 174A after distinguishing the law for foreign research expenses from that applying to domestic research expenses. This discussion applies only to domestic research.

[v] Treas. Reg. § 1.174-2(a)(1) 

[vi] Treas. Reg. § 1.174-2(a)(3)

[vii] Treas. Reg. § 1.174-2(a)(1).

[viii]Id.

[ix] Treas. Reg. § 1.174-2(a)(2).

[x] Treas. Reg. § 1.174-2(a)(6).

[xi] Treas. Reg. § 1.174-2(a)(9).

[xii] Treas. Reg. § 1.174-2(a)(10).

[xiii] IRC § 41(d)(1)(B)(i).

[xiv] Treas. Reg. § 1.41-4(a)(4).

[xv] Id.

[xvi] Treas. Reg. § 1.41-4(a)(3).

[xvii]Id.

[xviii] IRC § 41(2)(B).

[xix]IRC §41(d)(2)(C); Treas. Reg. § 1.41-4(b)(1) (last sentence). 

[xx] .  See Union Carbide Corp. & Subsidiaries v. Commissioner, T.C. Memo 2009-50, [slip op.] at 275-78, aff'd, 697 F.3d 104 (2d Cir. 2012). See George v. Commissioner, T.C. Memo. 2026-10 at 42-43. 

[xxi] Id; cited by Boswell v. Commissioner, 2022 U.S. Tax Ct. LEXIS 57, *7-11 (citing Treas. Reg. §1.41-4(b)(1). IRC § 41(d)(2)(C)).

[xxii] IRC 41(d)(2)(C).

[xxiii] IRC § 41(d)(1)(B)(ii).

[xxiv] Norwest Corp. & Subs. v. Commissioner, 110 T.C. 454, 495 (1998).

[xxv] IRC § 41(d)(1)(C).

[xxvi] Treas. Reg. §1.41-4(a)(6).

[xxvii] Treas. Reg. §1.41-4(a)(5).

[xxviii]Id.

[xxix]Id.

[xxx] Id.

[xxxi]Union Carbide, T.C. Memo 2009-50 at 47. Treas. Reg. §1.41-4(a)(5)(i).

[xxxii]Id.

[xxxiii] IRC § 41(d)(2).

[xxxiv] Treas. Reg. § 1.41-4(b)(2)

[xxxv]Id. (The shrinking-back rule is not itself applied as a reason to exclude research activities from credit eligibility).

[xxxvi] IRC § 41(b)(1).

[xxxvii] IRC § 41(b)(2)(A)(i) and (ii).

[xxxviii] IRC § 41(b)(2)(B).

[xxxix]IRC § 41(b)(2)(A)(i); Treas. Reg. § 1.41-2(d)(1).  

[xl] IRC § 41(b)(2)(C).

[xli] IRC § 41(b)(3)(A).

[xlii] IRC § 41(b)(3)(C)(i).

[xliii] IRC § 41(b)(3)(C)(ii).

[xliv] IRC § 41(a).

[xlv] IRC § 41(c)(4).

[xlvi] IRC § 41(a)(1).

[xlvii]IRC § 41(c).

[xlviii] IRC § 41(c)(3).

[xlix] IRC § 41(c)(3)(C).

[l] IRC § 41(c)(3)(B).

[li] Id.

[lii] IRC § 41(c)(2).

[liii]See Congressional Research Service, The Federal Research and Development (R&D) Tax Credit, R48848 (Feb. 6, 2026), p. 4.

[liv] IRC § 41(c)(4),

[lv] IRC § 41(c)(4)(A).

[lvi]IRC § 41(c)(4)(B); Treas. Reg. § 1.41-9(c)(1). 

[lvii]George v. Comm’r, T.C. Memo 2026-10 (Feb. 3, 2026).

[lviii]J.G. Boswell Co. v. Comm’r, T.C. Memo. 2022-127.

[lix] George, at 36, 86.

[lx] Id., at 2, 86.

[lxi] Id. at 9.

[lxii]Id. at 10-12.

[lxiii]Id. at 33.

[lxiv]Id. at 34-35.

[lxv] Id. at 34.

[lxvi] In 2023, the parties entered into a stipulation accepting IRS’ rejection of $842,907 in research credits carried forward from 2013 and 2014. The stipulation included a $168,581 accuracy-related penalty. The parties agreed to be bound by that stipulation, which was not part of this case. This explains the fact that this case addressed only $3,628,691 of the company’s claimed $4,471,598.

[lxvii]George, at 51-52.

[lxviii]Id. at 52-62.

[lxix] Id. at 62-68.

[lxx] Id. at 68-69.

[lxxi]Id. at 69-74.

[lxxii]Id. at 69-70.

[lxxiii] Id. at 74-75.

[lxxiv]Id. at 75-79.

[lxxv]Id. at 79.

[lxxvi] Id. at 80.

[lxxvii] Id.

[lxxviii]Cohan v. Commissioner, 39 F.2d 540 (2d Cir. 1930); Union Carbide Corp., TC Memo at 294-295.

[lxxix]George, at 81.

[lxxx] IRC 6662(a). The penalty was sought for tax years 2014 and 2016, the years the company applied the credits.

[lxxxi]George, at 82.

[lxxxii]Id. at 85.

[lxxxiii]Id.

[lxxxiv]J.G. Boswell, at 4.

[lxxxv] Id. 

[lxxxvi] IRC § 41(d)(2)(A); ); J.G. Boswell at 5-6.

[lxxxvii] IRC §41(d)(2)(B).

[lxxxviii] IRC § 41(d)(2)(C).

[lxxxix] Union Carbide, T.C. Memo 2009-50, [slip op.] at 275-78, aff'd, 697 F.3d 104 (2d Cir. 2012).

[xc]  J.G. Boswell, at 6.

[xci] Id. at 8-9.

[xcii]Id. at 9.

[xciii]Id. at 11.

[xciv]Id. at 15.

[xcv] IRC §41(d)(2)(B).

[xcvi] Treas. Reg. §§ 1.174-2(a)(1)-(4).