Federal research tax credits under Internal Revenue Code Section 41 require evidence of both qualifying activities and eligible expenditure. This study examines Martin v. Commissioner alongside Little Sandy Coal, Betz, and Phoenix Design Group, distinguishing the actual holdings from broader practical lessons. Martin concerns net operating loss (NOL) carryforwards, not research credits; it did not establish a new Section 41 doctrine or abolish reasonable estimates.
The Substantiation Doctrine and the Martin Precedent
In Martin v. Commissioner, T.C. Memo. 2021-35, Linda and John Martin claimed NOL carryforwards from a 1990s racing business on their 2009 and 2010 returns. Judge Holmes found that they had not established the losses and their availability in those years. Settlements concerning 2007–2008 and 2011–2012 did not determine the disputed deductions. The court also addressed business deductions and unreported income; it did not decide whether any activity qualified under Section 41.
The NOL analysis required proof of the original loss, applicable carryback treatment, intervening use, and expiration. Some losses at issue were subject to a 15-year carryforward, so equating their treatment with a universal 20-year NOL rule is incorrect. The decision supports a practical analogy for research-credit carryovers: preserve the evidence needed to establish their origin and remaining balance. It does not create perpetual audit jurisdiction or override limitation periods and binding determinations.
| Feature of Martin v. Commissioner (2021-35) | Detail and Judicial Finding | R&D Tax Credit Equivalent / Implication |
|---|---|---|
| Primary Issue | Substantiation and availability of NOL carryforwards from a racing business. | A practical analogy for supporting research-credit carryovers, not a Section 41 holding. |
| Taxpayer Argument | Settlements of other tax years should prevent another challenge. | Prior audit acceptance alone does not guarantee a later credit claim. |
| Court Ruling | The settlements did not preclude the disputed NOL determination. | Evaluate the legal effect and scope of any earlier agreement. |
| Recordkeeping Standard | Establish the losses and the amount still available. | Retain qualification, calculation, and carryover records. |
| Technological Excuse | Insufficient historical evidence did not establish entitlement. | Maintain accessible records; Martin did not mandate digitization or indefinite retention. |
Martin and Research Credit Case Law
These cases share concerns about proof, but it is misleading to describe the research-credit decisions as descendants of a new “Martin Principle.” Section 41 and its regulations independently govern research eligibility, expense categories, exclusions, and substantiation. A persuasive claim connects the technical work actually performed with the expenses included in the calculation.
Phoenix Design Group, Inc. v. Commissioner, T.C. Memo. 2024-113, examined engineering work on mechanical, electrical, plumbing, and fire-protection systems. The court found that the projects tried failed the uncertainty and experimentation requirements. Its analysis does not establish that engineering calculations are always ineligible or that a skilled engineer can never face qualifying design uncertainty.
The Financial Issues in Phoenix Design Group
The amounts below distinguish credits generated from amounts used in later years. The stated penalties are the assessed amounts in the case summary; they should not be recomputed as 20% of the credit-used column. A Section 6662 penalty applies to the relevant underpayment, which can differ from a credit adjustment. The decision addressed three selected projects under a litigation agreement, so it should not be presented as an individual adjudication of every project in the firm’s portfolio.
| Tax Year | R&D Credit Generated | R&D Credit Utilized | Accuracy Penalty (20%) |
|---|---|---|---|
| 2013 | $138,205 | — | — |
| 2014 | $158,031 | — | — |
| 2015 | $138,334 | $55,504 | $11,101 |
| 2016 | $128,228 | $47,811 | $9,021 |
| 2017 | — | $219,177 | $43,835 |
| 2018 | — | $68,147 | $13,629 |
| 2019 | — | $71,102 | $14,220 |
The table summarizes the disputed amounts, not a standalone final tax computation. The case illustrates why descriptions of design stages and completed systems need supporting evidence of the uncertainty and evaluative work involved.
Applying the Four-Part Test
Qualification is assessed separately for each business component. A business component includes a product, process, computer software, technique, formula, or invention held for sale, lease, or license, or used in the taxpayer’s trade or business. Sophistication, novelty, or commercial success alone does not establish eligibility.
Research Expenditures and Technical Uncertainty
For the historical years addressed in these cases, Section 41 referred to Section 174. Following the 2025 legislation, the current statutory cross-reference is to domestic research or experimental expenditures under Section 174A. The uncertainty inquiry concerns whether available information establishes the capability, method, or appropriate design for developing or improving the component. Uncertainty about price, customer preferences, or scheduling alone is insufficient.
In Betz v. Commissioner, T.C. Memo. 2023-84, shareholders of Catalytic Products International claimed credits associated with air-pollution-control systems. The court found shortcomings in evidence of qualifying investigation and in support for claimed expenses. Describing a difficulty and the eventual solution does not, by itself, show what research was performed between those points. This is an application of research-credit requirements, not a rule created by Martin.
The Technological in Nature Test
The research must fundamentally rely on physical or biological science, engineering, or computer science. An employee’s professional title is not a substitute for evidence of the work. Equally, research need not advance knowledge throughout an entire industry. The original claim that Phoenix Design Group rejected the technological nature of the engineering work misstates the central issues; the disputed tests were uncertainty and experimentation.
The Business Component and Permitted Purpose Requirements
A qualifying purpose concerns improved function, performance, reliability, or quality. A tangible finished product is not the only possible component: a process or technique may also qualify. A design service must still be analyzed under the statutory definition and all remaining requirements. Changes directed merely to style, taste, cosmetic appearance, or seasonal factors do not satisfy the permitted-purpose rule.
The Process of Experimentation Test and the 80% Rule
Substantially all means at least 80% of the relevant research activities, measured on a cost or other consistently applied reasonable basis, must constitute elements of experimentation. The analysis identifies uncertainty, alternatives, and an evaluative process, which may involve modeling, simulation, or systematic trial and error. This is not an 80%-new-product rule, nor is it the separate employee wage-allocation rule.
In Little Sandy Coal Co. v. Commissioner, 62 F.4th 287 (7th Cir. 2023), the credit involved 11 vessels, with two treated as representative at trial. The Seventh Circuit affirmed disallowance because the taxpayer lacked a principled basis for establishing the experimentation fraction. It rejected arbitrary allocations and reliance on the vessels’ novelty. It also rejected categorical exclusion of direct support and supervision from the numerator; inclusion depends on the activities proved. The decision does not prohibit all estimates or require a particular timekeeping product.
Where an entire business component fails, the regulatory shrinking-back rule can require applying the tests to the most significant qualifying subset of its elements. It does not excuse missing evidence or automatically qualify a selected subcomponent.
Pilot Models and Funded Research
A pilot model is produced to evaluate and resolve uncertainty. Under Treasury Regulation Section 1.174-2, its ultimate sale or use does not automatically prevent research-expense treatment. Pilot-model status also does not automatically turn all construction costs into Section 41 qualified research expenses: the expense categories, four-part test, and exclusions still apply.
Betz did not establish that customer-delivered products can never be pilot models. The problem was failure to establish qualifying investigative production and adequate expense support on the facts. Routine fabrication after uncertainty has been resolved, acceptance testing, and quality-control work require separate analysis; they do not qualify merely because the finished system is customized.
The funded-research exclusion requires analysis of economic risk and substantial rights. Review the complete contract, the applicable law, and the payment conditions. A fixed price, milestone invoice, intellectual-property clause, or professional-liability obligation is not independently decisive. Retaining general experience alone is not necessarily retaining substantial rights in the results.
The source also discusses Smith v. Commissioner and System Technologies, Inc. v. Commissioner. The cited procedural outcomes concerned denial of IRS summary-judgment motions on funding issues, allowing factual disputes to proceed. They were not final determinations that all claimed research and expenses qualified. Contract milestones and local-law rights may affect that analysis, but they do not provide an automatic safe harbor.
| Case | Business Component | Primary Failure Reason | Judicial Takeaway |
|---|---|---|---|
| Little Sandy Coal | 11 marine vessels; two representative vessels at trial. | Insufficient proof of the experimentation fraction. | Use supported, component-specific allocations; novelty alone is insufficient. |
| Betz | Air-pollution-control systems. | Deficiencies in qualification, pilot-model evidence, wage support, and rights for certain projects. | A sale does not automatically disqualify a pilot model; establish research and expenses. |
| Phoenix Design Group | MEPF engineering projects. | Failure to establish uncertainty and experimentation for the projects tried. | Explain the investigative work, not only engineering complexity. |
| Martin (2021-35) | Racing-business NOL carryforwards. | Failure to establish the losses and their continued availability. | Other-year settlements did not resolve the disputed carryforwards. |
Practical Implications for Research Credit Compliance
Supported Estimates and Reliable Records
The practical distinction is between an estimate grounded in evidence and an unsupported percentage. Preserve project notes, design revisions, test results, source-control history, payroll data, invoices, and credible explanations of employee responsibilities. Time records are useful, but these decisions did not enact a universal daily-timesheet requirement. Retrospective interviews can help explain records; they should not replace proof that qualifying work occurred.
Carryforward Records and Research-Expense Treatment
Unused general business credits generally have a one-year carryback and 20-year carryforward under Section 39, subject to applicable restrictions. Keep generation-year calculations, relevant base-period evidence, and a reconciliation of use, adjustments, and remaining balances. Record retention should cover every period in which the information remains material to tax administration, including applicable examination periods after use. Booking an asset does not itself extend an assessment deadline.
The source’s claim that research expenditures now invariably require five-year amortization is outdated. For tax years beginning in 2022–2024, the enacted rules generally required five-year amortization for domestic research and 15 years for foreign research. Section 174A now generally permits current deduction of domestic research expenditure for tax years beginning after 2024, with an elective capitalization alternative. Foreign research generally remains subject to 15-year amortization. Transition provisions address certain previously capitalized domestic costs.
Credit eligibility and deduction treatment are distinct inquiries, and Section 280C coordinates their benefits. A disallowed credit does not, by itself, establish the correct deduction or capitalization treatment. Apply the law and elections for the actual expenditure year rather than assuming an automatic “double loss.”
The Evolution of Form 6765
The IRS’s December 2025 Instructions for Form 6765 state that Section G is optional for tax years beginning before 2026 and required for years beginning after 2025, subject to the specified exceptions. Amended research-credit claims have additional information requirements. The source’s prediction that project detail would become mandatory starting in 2024 is therefore obsolete. The filing requirements should not be described as a proven direct consequence of Martin or any single judicial decision.
| Change in Practice | Before Modern Cases (Martin/Little Sandy) | After Modern Cases (Current Standard) |
|---|---|---|
| Documentation Timing | Retrospective interviews could support a study, but evidence was still required. | Capture evidence as work occurs and use interviews to explain it. |
| Allocation Method | Estimates required a factual basis; no general exemption from substantiation existed. | Use a consistent, supported allocation method; arbitrary percentages remain vulnerable. |
| Audit Defense | Prior acceptance did not automatically bind other tax years. | Preserve the scope of agreements and evidence for each claimed carryover. |
| Project Selection | The statutory research requirements already applied. | Identify uncertainty and evaluative work for each business component. |
| Contract Review | Economic risk and substantial rights were already relevant. | Examine operative contract terms and applicable law together. |
The table describes practical improvements, not a newly enacted before-and-after legal standard.
Strategic Recommendations for Corporate Taxpayers
Maintain an accessible archive that connects each claim to technical and financial evidence. Assign responsibility for retaining it through staff departures, acquisitions, and system migrations. Digital storage can reduce loss and retrieval problems, but no case discussed here imposes a universal obligation to digitize every historical document.
Document the specific uncertainty, alternatives considered, evaluation performed, and results. Successful and unsuccessful investigations can both support a claim; failure is not a statutory prerequisite. Distinguish qualifying research from production, routine testing, administration, and other excluded activities, and connect the resulting allocation to payroll and other claimed expenses.
Review customer and contractor agreements before relying on the research credit. Record who bears the relevant economic risk and who may use the results. Professional advice should evaluate the actual terms and conduct of the parties, including governing law; drafting a label such as “research” cannot change the substance of an arrangement.
Consider the shrinking-back rule and use a defensible calculation method. Accuracy-related penalties require a separate analysis, including reasonable cause and good faith where available. Hiring an adviser alone does not establish a defense, and the taxpayer remains responsible for providing complete, accurate facts and reviewing the claim.
