The Research Credit: Using
Statistical Sampling
TaxDrone.AI – R&D Tax Credit. Reinvented.
Introduction
The Sec. 41 research credit rewards businesses for increasing research activities, but substantiating claims can be complex. In particular, taxpayers must demonstrate that each business component meets the four-part test for qualified research activities.
To ease the documentation burden, many taxpayers turn to statistical sampling, which allows them to estimate qualified research expenses (QREs) while limiting the scope of record review. However, as clarified in IRS Field Attorney Advice (FAA) 20212501F, statistical sampling does not replace or relax the requirement to analyze the four-part test at the business-component level.
Business Components and the Four-Part Test
Under Sec. 41(d), taxpayers must evaluate research activities separately for each business component,
defined as any product, process, software, technique, formula, or invention held for sale, lease,
license, or internal use.
Sec. 174 Test
Costs must qualify as research or experimental expenditures aimed at eliminating uncertainty in product development.
Technological Information Test
Research must rely on principles of science, engineering, or computer science.
Business Component Test
The activity must relate to developing a new or improved business component.
Process of Experimentation Test
Substantially all activities (≥80%) must involve a process of experimentation for function, performance, reliability, or quality.
"Shrink-Back Rule"
If the component as a whole fails the test, the “shrink-back rule” applies, requiring analysis of the next most significant
subset (e.g., from engine to carburetor).
Substantiation
Requirements
IRS regulations (Regs. Sec. 1.41-4(d)) require taxpayers to keep usable, detailed records supporting research credit claims. Courts have consistently upheld this standard at the business-component level (e.g., Union Carbide, Suder, Siemer Milling, Max).
These cases emphasize two key points:
- The burden of proof rests on the taxpayer.
- Each business component must independently meet the four-part test, regardless of sampling methods.
Why Statistical Sampling
Helps
While sampling does not eliminate the four-part test, it offers practical advantages:
Efficiency
Reduces the number of records requiring full review.
IRS Acceptance
Rev. Proc. 2011-42 outlines permissible methods, and the IRS often encourages sampling for large, complex claims.
Audit Readiness
By limiting scope, sampling helps taxpayers and the IRS streamline examinations and information document requests (IDRs).
How Sampling Works
A typical sampling process includes:
Population Definition
Identify all business components with qualified activity/QREs.
Stratification
Group components by type/size.
Random Sampling
Select items from each stratum.
Analysis
Apply the four-part test to each sample.
Estimation
Extrapolate total QREs using statistical methods, ensuring confidence and precision meet IRS standards.
Choosing the Sampling Unit
Selecting the right sampling unit is critical for compliance and efficiency
Business Component
Preferred, as it aligns directly with IRS requirements.
Projects
Often overlap with business components and are straightforward units.
Employees
Practical if time-tracking is limited, but requires establishing nexus between employee time and qualifying business components.
Cost Centers/Departments
Can be efficient if historical data supports qualifying percentages, but still requires business-component-level substantiation.
Two-stage approaches (e.g., sampling employees, then sampling their business components) can reduce
documentation burdens while preserving validity.
Sampling Unit Best Practices
Align with IRS Guidance
Follow Rev. Proc. 2011-42 to ensure compliance.
Maintain Business-Component Focus
Always substantiate the four-part test at this level, regardless of sampling method.
Work with Experts
Engage statisticians familiar with IRS expectations to ensure confidence and precision standards are met.