Tax Audit Sample

A tax audit sample is a representative subset of a taxpayer’s qualified research expenses and activities selected by the Ohio Department of Taxation to verify the accuracy of a credit claim without examining every transaction. It allows the Tax Commissioner to extrapolate findings from the sample to determine a final assessment or credit amount for the entire period under review.

The implementation of a tax audit sample within the administrative framework of the Ohio Research and Development (R&D) Investment Tax Credit represents a sophisticated balancing of state revenue preservation and the constitutional mandate for fair tax administration. Historically, the verification of research and development credits was a labor-intensive process, requiring tax examiners to review thousands of individual line-item expenses and verify the specific scientific nature of hundreds of discrete projects. As the Ohio tax system transitioned from the traditional corporate franchise tax to the Commercial Activity Tax (CAT) and the Financial Institutions Tax (FIT), the necessity for more efficient verification methods led to the codification of audit sampling. This process is not merely a statistical convenience but a legally defined mechanism that grants the Ohio Department of Taxation (ODT) the authority to issue assessments for an entire credit period based on the deficiencies found within a limited representative period. The modern landscape of Ohio tax auditing is characterized by a “piggyback” relationship with federal law, specifically Internal Revenue Code (IRC) Section 41, which defines the eligibility of expenses, while the procedural application and jurisdictional situsing of those expenses are strictly governed by the Ohio Revised Code and administrative guidance issued by the Tax Commissioner.

The Statutory Basis for Audit Sampling in the Research and Development Context

The authority of the Tax Commissioner to utilize sampling in the context of the R&D credit is explicitly set forth in two primary sections of the Ohio Revised Code. For taxpayers subject to the Commercial Activity Tax, R.C. 5751.51(E) provides the governing authority, stating that the Tax Commissioner may audit a sample of the taxpayer’s qualified research expenses over a representative period to ascertain the amount of tax credit the taxpayer may claim. For financial institutions, a nearly identical provision exists in R.C. 5726.56(G), which authorizes the use of sampling to determine the amount of credit allowed against the Financial Institutions Tax.

These statutes establish several critical legal requirements that define the boundaries of the audit sample. First, the sample must be “representative,” a term that implies both statistical validity and qualitative similarity to the larger population of expenses. Second, the sample must cover a “representative period,” which typically includes the current tax year for which the credit is claimed and the three preceding taxable years that constitute the “base period” for the incremental calculation of the credit. The legal mechanism of the “representative period” is vital because the Ohio R&D credit is not a flat percentage of all research spending but a 7% credit on the amount by which current-year Ohio qualified research expenses (QREs) exceed the average annual Ohio QREs for the three prior years.

Statutory Element	Legal Requirement and Administrative Application
Representative Sample	Must include a statistically or judgmentally significant subset of projects, cost centers, or employees that accurately reflect the taxpayer’s general R&D activities.
Representative Period	Typically encompasses the credit year and the three-year base period necessary for calculating the incremental growth in research investment.
Good Faith Effort	The Commissioner is mandated to attempt to reach an agreement with the taxpayer regarding the selection and methodology of the sample.
Assessment Authority	Following the sample review, the Commissioner may issue a formal assessment under R.C. 5751.09 or R.C. 5726.20 if the extrapolated credit amount is less than what was claimed.

The “good faith effort” requirement is a unique procedural safeguard. The ODT is legally obliged to negotiate with the taxpayer in selecting the sample, which often involves discussions between the taxpayer’s tax professionals and the Department’s Computer Audit Specialist (CAS). If an agreement is reached, it is generally formalized in a written sampling agreement, which becomes binding on both parties and is difficult for the taxpayer to challenge later. However, the law provides that the Commissioner is not precluded from proceeding with the audit using their own chosen sample if an agreement cannot be made, although this leaves the assessment more vulnerable to challenges at the Ohio Board of Tax Appeals (BTA) if the sample is later found to be biased or unrepresentative.

Administrative Guidance and Local Revenue Office Procedures

Local guidance from the Ohio Department of Taxation further clarifies how these statutes are applied in the field. Information releases and administrative rules emphasize that the R&D credit is a matter of “legislative grace,” and the burden of proof rests entirely on the taxpayer to substantiate their claim. This burden is exacerbated during a sample-based audit, as any documentation failure within the sample is extrapolated to the entire population.

Audit Methodology and Selection Criteria

The ODT employs several types of sampling methodologies, which are often adapted from general business tax audit principles such as those used for Sales and Use Tax or the Petroleum Activity Tax (PAT). Under O.A.C. 5703-37-02, the Commissioner prescribes that sampling methods may include statistical and non-statistical approaches. Statistical methods utilize random selection and probability theory to project population values with known reliability, whereas non-statistical methods, such as block sampling, involve taking a specific period of time or a specific set of projects and testing 100% of the records within that “block”.

In the specific context of R&D audits, the ODT often uses a stratified judgmental sample. This involves categorizing the total population of QREs into strata based on dollar value or project type. For instance, the auditor may choose to audit 100% of all “high-value” projects (those exceeding a certain dollar threshold) while sampling a smaller percentage of “medium” and “low-value” projects. This stratification is intended to ensure that the largest expenditures are verified in detail while still providing a statistically valid basis for adjusting smaller, more numerous expenses.

Record Retention as a Pillar of Audit Sampling

The efficacy of an audit sample is entirely dependent on the availability of records. HB 33, passed in 2023, codified and expanded the record retention requirements for the R&D credit. Taxpayers must now retain records to substantiate the claim for four years after the later of the return’s due date or the date the return was actually filed. Crucially, this retention requirement extends to the expenses incurred in the three preceding taxable years used in calculating the base amount.

If a taxpayer fails to provide sufficient records for a sampled project, the ODT has the authority to disqualify the entire expense associated with that project. If this occurs within a non-statistical sample, the resulting “Percentage of Error” (POE) can be devastating. For example, if an auditor samples five projects out of 50 and finds that one project lacks sufficient documentation to meet the Four-Part Test, the 20% error rate derived from that sample can be applied to all 50 projects, effectively reducing the taxpayer’s total credit by 20% across the entire period.

The Role of the Computer Audit Specialist (CAS)

A distinctive feature of the ODT audit process is the involvement of a Computer Audit Specialist. The CAS is responsible for analyzing the taxpayer’s electronic records and designing the sampling plan. Under the modern ODT guidance, taxpayers who keep records electronically are required to provide those records to the Department in an electronic format. The CAS uses these records to perform automated tests and select the sample based on parameters discussed with the taxpayer. The involvement of the CAS is intended to introduce a level of objectivity into the sampling process, but it also means that taxpayers must be prepared to provide data in a “sufficiently usable form and detail” to facilitate the audit.

Application of Federal Standards in Ohio Audits

The Ohio R&D tax credit is explicitly defined by reference to the Internal Revenue Code. Both R.C. 5751.51(A) and R.C. 5726.56(A) state that “qualified research expenses” have the same meaning as in Section 41 of the Internal Revenue Code. This “piggyback” structure means that for an activity in an audit sample to qualify for the Ohio credit, it must first satisfy the federal Four-Part Test.

The Four-Part Test and Audit Scrutiny

During the audit of a sample, ODT examiners scrutinize each activity against the following criteria:

1. The Section 174 Test: The expenditures must be eligible for treatment as expenses under IRC § 174, meaning they must be incurred in connection with the taxpayer’s trade or business and represent research and development costs in the experimental or laboratory sense.
2. The Technological Information Test: The research must be undertaken for the purpose of discovering information which is technological in nature. This requires that the process of experimentation fundamentally relies on principles of the physical or biological sciences, engineering, or computer science.
3. The Business Component Test: The taxpayer must intend to use the information discovered to develop a new or improved business component of the taxpayer. This includes products, processes, computer software, techniques, formulas, or inventions to be held for sale, lease, or license, or used in the taxpayer’s trade or business.
4. The Process of Experimentation Test: Substantially all (at least 80%) of the activities must constitute elements of a process of experimentation for a purpose relating to a new or improved function, performance, or reliability or quality. This involves the identification of uncertainty concerning the development or improvement of a business component, the identification of alternatives to eliminate that uncertainty, and the conduct of a process of evaluating those alternatives.

ODT auditors have become increasingly aggressive in their application of the “Process of Experimentation” test, often demanding contemporaneously generated project documentation to prove that a systematic trial-and-error methodology was followed. If a sampled project only contains “validation” or “troubleshooting” activities rather than true experimentation, it will be disqualified during the audit.

Geographical Situsing: The Ohio Incurred Requirement

A critical departure from federal law that is a primary focus of Ohio audit samples is the jurisdictional limitation. While the federal credit applies to research conducted in the United States, the Ohio credit is strictly limited to qualified research expenses “incurred in this state”. During an audit, ODT will use the sample to verify the “nexus” of each expense.

For wage QREs, this means the auditor will verify that the employees performing the research were physically located in Ohio or were “sitused” to Ohio for tax purposes. For supply QREs, the auditor will verify that the supplies were used in Ohio. For contract research expenses, the auditor will determine where the contractor performed the work. The ODT strictly construes the phrase “incurred in this state,” and many taxpayers have seen their credits decimated during audit sampling when it is discovered that a portion of their research team was working remotely from another state or that a specialized testing lab was located outside Ohio.

Legislative Evolution: HB 33 and the Member-by-Member Mandate

The enactment of HB 33 in 2023 introduced a paradigm shift in how the ODT audits R&D credits for combined and consolidated groups. Previously, the credit was often viewed as an aggregate benefit for the entire group. Under the new provisions of R.C. 5751.51(C), each person in the taxpayer’s group must separately calculate the credit using the QREs incurred by that person.

Impact on Audit Sampling Strategy

This legislative change has direct implications for audit sampling. Auditors now select samples on a member-by-member basis. This “siloing” of the credit means that a taxpayer can no longer use the high performance or excellent documentation of one subsidiary to mask the non-compliance or poor documentation of another. If the ODT chooses to audit a sample of entities within a combined group, and a specific member fails the audit, the credit reduction is applied directly to that member’s contribution to the group’s total.

Furthermore, the member-by-member calculation necessitates more granular record-keeping. Each entity must maintain its own base-period average and current-year expenditure data. If an entity leaves the group during the year, its expenses are generally excluded from the aggregate credit calculation as of December 31, a rule that ODT auditors verify through their review of organizational charts and payroll situsing during the audit.

Changes to CAT Exclusion Amounts and Credit Value

The relevance of the R&D credit, and thus the frequency of audits, is influenced by the substantial changes to the CAT exclusion thresholds enacted in HB 33.

Tax Year	CAT Exclusion Amount (Ohio Taxable Gross Receipts)	Impact on Taxpayer Population
Pre-2024	$1,000,000	Most businesses with nexus were required to file and pay the Annual Minimum Tax (AMT).
2024	$3,000,000	Approximately 85% of previous CAT filers are estimated to no longer owe the tax.
2025+	$6,000,000	The exclusion increases further, focusing the CAT solely on larger entities.

As the CAT becomes a tax primarily paid by larger corporations, the ODT’s audit focus has sharpened on these high-value filers. For a company with hundreds of millions in Ohio receipts, the R&D credit remains one of the few significant nonrefundable offsets available to reduce their 0.26% tax liability. Consequently, the ODT has a higher incentive to utilize aggressive sampling methodologies to verify these larger credit claims.

Extrapolation Mechanics and the Mathematics of Assessment

The process of extrapolation is where the results of a limited sample are applied to the broader population of tax data. This is a mathematical exercise that can lead to significant variances between the taxpayer’s claimed credit and the ODT’s final determination.

Calculating the Incremental Credit with Audit Adjustments

The basic formula for the Ohio R&D credit is incremental. If Q_CY represents current year Ohio QREs and Q_B represents the average of the prior three years, the credit is defined as:

Credit = 0.07 × (Q_CY – Q_B)

In an audit, the examiner typically samples Q_CY. If the sample reveals that a portion of the claimed expenses are unqualified, an “Error Rate” (E) is established. The adjusted current year expenses (Q_{CY_Adj}) are then calculated as:

Q_{CY_Adj} = Q_CY × (1 – E)

The adjusted credit (Credit_Adj) is then calculated using the new Q_{CY_Adj}, while the base amount Q_B often remains unchanged unless the auditor also decides to sample and adjust the base-period years.

The Cohan Rule and Estimation in Audits

A common point of contention in Ohio audits is whether the ODT must accept “reasonable estimates” when precise records are missing. Federal jurisprudence, particularly the Cohan v. Commissioner case, allows courts to estimate expenses when it is clear that research occurred but documentation is imperfect. However, the ODT and the Ohio BTA have historically taken a more rigid stance. While some federal courts have allowed “employee recollections” or “data extrapolation” to estimate QREs, the ODT frequently rejects such evidence in favor of contemporaneous documentation.

If an audit sample consists of projects where the only evidence of research is oral testimony from an engineer without supporting lab notebooks or testing data, the ODT will typically assign a 100% error rate to those sampled units. This highlight’s the danger for taxpayers: an ODT auditor is not legally required to “split the difference” or accept a “reasonable estimate” during the sampling process unless the taxpayer can provide “sufficiently usable” records.

Jurisprudential Trends and the Presumptive Validity of Assessments

When the ODT completes an audit using a sample and issues an assessment, the taxpayer enters the realm of administrative litigation. The ODT’s final determinations and subsequent BTA cases establish the “ground rules” for challenging an audit sample.

The Burden of Proof and Presumptive Validity

Under Ohio law, assessments issued by the Tax Commissioner are “presumptively valid.” This principle, articulated in cases such as R.K.E. Trucking, Inc. v. Zaino, means that the burden is not on the ODT to prove the assessment is correct, but on the taxpayer to prove it is erroneous. In the context of an audit sample, this means the taxpayer must demonstrate either that the sampled transactions were actually qualified research or that the sampling methodology itself was so flawed as to be unrepresentative of the population.

In Dean Thomas v. Tax Commissioner, the BTA affirmed that in the absence of complete and accurate records, the Commissioner may use a sampling of business activity for a representative period to verify tax returns. The BTA noted that when a taxpayer provides “summary reports” or “spreadsheets” but fails to provide the underlying “primary records” (like individual guest checks or invoices), they have not met their burden to overcome the presumptive validity of the sample-based assessment.

Challenges to Sampling Methodology

Taxpayers have occasionally succeeded in challenging audit samples by proving that the selected period was not representative. For example, if a company’s R&D cycle is highly seasonal, and the ODT chooses a sample period that only covers the “off-season,” the taxpayer may argue that the results are biased. Similarly, if the ODT uses a “block sample” of one month and extrapolates it to a three-year period, the taxpayer may challenge this as a violation of the statutory requirement for a “representative period”.

However, these challenges are difficult. ODT’s use of statistical sampling, which includes calculations of “confidence levels” and “precision,” is generally considered legally defensible. A statistical sample not only estimates the disallowed deductions but also provides a mathematical assurance that the estimate is within a certain range of the actual population value. If the ODT utilizes a CAS to design a statistically valid sample, the BTA is highly unlikely to invalidate the methodology unless the taxpayer can provide a competing statistical analysis that is more accurate.

Comprehensive Example: The Audit of Aerospace Dynamics Ohio LLC

To demonstrate the practical application of a tax audit sample in the context of the Ohio R&D credit, we analyze a hypothetical but realistic scenario involving a mid-sized aerospace component manufacturer.

The Claim and Initial Audit Trigger

Aerospace Dynamics Ohio LLC (ADO) is a consolidated taxpayer that claimed a $245,000 R&D tax credit on its 2022 CAT return. ADO reported $7,500,000 in Ohio QREs for 2022, with a three-year base period average of $4,000,000.

Financial Metric	Amount
2022 Ohio QREs (Current Year)	$7,500,000
2019-2021 Average (Base Period)	$4,000,000
Incremental Excess	$3,500,000
Credit Claimed (7% of Excess)	$245,000

The ODT initiated a field audit because ADO’s credit claim was significantly higher than its historical average and its credit exceeded its net CAT liability, resulting in a large carryforward.

Developing the Sampling Plan

The ODT examiner and the CAS met with ADO’s tax director to discuss a sampling plan. ADO’s R&D efforts involved 120 different projects across three divisions: Wing Design, Engine Testing, and Navigation Software.

The parties agreed to a stratified judgmental sample over the representative period of 2019-2022. The population was stratified as follows:

Stratum	Description	Total Population	Sample Size	Methodology
Stratum 1	Large Projects (>$500k)	4 projects	4 projects	100% Detailed Review
Stratum 2	Mid-Sized Projects ($50k-$500k)	36 projects	10 projects	Random Selection
Stratum 3	Small Projects (<$50k)	80 projects	0 projects	Error Rate Extrapolation

The Field Audit and Discovery of Errors

The auditor spent two weeks reviewing the 14 sampled projects. During the review, the following discrepancies were identified:

1. Stratum 1 (Large Projects): One project involving “Engine Housing Reinforcement” was found to be a contract for a customer in Indiana. While the research was qualified under IRC § 41, ADO failed to prove that the work was “incurred in this state” because the actual testing was performed at a facility in Fort Wayne, Indiana. The entire $600,000 for this project was disqualified.
2. Stratum 2 (Mid-Sized Projects): Out of the 10 projects sampled (totaling $1,200,000 in QREs), two projects failed the “Process of Experimentation” test. These projects were found to be routine “quality control” and “adaptation” of existing designs for a new client, rather than research intended to eliminate technological uncertainty. These two projects represented $180,000 in unqualified expenses.
3. Base Period Consistency: The auditor also sampled the base-period years and found that ADO had included similar “quality control” activities in its 2019 and 2020 QREs. However, the auditor chose to focus the adjustments on the current year to maximize the assessment potential.

Extrapolation of Results

The auditor then calculated the adjustments for the entire 2022 claim.

• Stratum 1 Adjustment: Direct reduction of $600,000.
• Stratum 2 Adjustment: The “Percentage of Error” (POE) for the sampled mid-sized projects was $180,000 / $1,200,000 = 15%.
• The 15% POE was then applied to the entire Stratum 2 population of $3,000,000, resulting in a total Stratum 2 disallowance of $450,000.
• Stratum 3 Adjustment: Per the sampling agreement, the auditor applied the 15% POE to the $900,000 in small projects, resulting in a $135,000 disallowance.

Total Disallowed QREs: $600,000 + $450,000 + $135,000 = $1,185,000.

Recalculated Credit and Assessment

The auditor recomputed the credit based on the “True” Ohio QREs discovered during the sample-based audit:

Category	Claimed	Audit Adjusted
Current Year QREs	$7,500,000	$6,315,000
Base Period Average	$4,000,000	$4,000,000
Excess QREs	$3,500,000	$2,315,000
Allowable Credit (7%)	$245,000	$162,050

The ODT issued a Notice of Assessment for the credit difference of $82,950. In addition to the tax, the ODT applied a 15% penalty ($12,442.50) and pre-assessment interest. ADO had 60 days to file a Petition for Reassessment to object to either the disqualification of the “Engine Housing” project or the 15% POE applied to the other strata.

Procedural Rights and Strategies for Managing Audit Risk

The use of sampling in an audit is not a one-way street. Taxpayers have specific rights and can employ strategies to mitigate the impact of an ODT examination.

Negotiating the Sampling Agreement

The statutory requirement for a “good faith effort” to reach an agreement on the sample selection is the taxpayer’s strongest leverage point. Taxpayers should never agree to a sample without first performing their own internal “pre-audit.” If a taxpayer identifies a specific project with weak documentation, they should argue that it is an outlier and should not be included in a representative sample, or they should propose a stratification that isolates that project from the general population.

Furthermore, taxpayers should advocate for statistical sampling over judgmental “block” sampling if they believe their overall compliance is high. Statistical sampling is often more favorable because it includes “precision” adjustments; if the auditor’s sample size is small and the variance is high, the “confidence bound” may actually work in the taxpayer’s favor to reduce the extrapolated assessment.

Managing the “Ohio Incurred” Verification

To prevent “low-hanging fruit” adjustments during an audit sample, companies must maintain rigorous payroll situsing data. Under R.C. 5751.033 and O.A.C. 5703-29-22, the ODT focus is on where the benefit of the service is received or where the employee is based. With the rise of remote work, companies often neglect to update their R&D credit software to exclude wages for employees who moved out of state but remain on the Ohio payroll. Discovering even one such employee in an audit sample can lead to an extrapolation that disqualifies a percentage of the entire group’s R&D wages.

Utilizing Voluntary Disclosure Agreements (VDA)

For taxpayers who discover significant errors in their R&D credit claims before an audit notice is received, the ODT offers a Voluntary Disclosure Agreement program. By coming forward voluntarily, a taxpayer can limit the look-back period (typically to three years) and avoid the 15% or 35% penalties associated with an audit assessment. However, once an auditor contacts the taxpayer to initiate an audit or an “investigation,” the VDA option is no longer available.

Future Outlook and the Impact of Continuing Legislative Changes

The administration of the Ohio R&D credit and its associated audit methodologies will continue to evolve as the CAT exclusion thresholds rise. As fewer small businesses pay the CAT, the ODT’s R&D audit resources will likely consolidate into a specialized unit focused on the “mega-filers”—those with billions in receipts who generate the majority of R&D credits.

Furthermore, the member-by-member calculation requirement of HB 33 is likely to lead to more litigation at the BTA. We can expect future cases to focus on how to allocate shared “overhead” research costs among group members and whether a sample of one member’s activities can be used to influence the audit of an affiliate. As things stand, the ODT’s “aggressive audit policy” is now firmly supported by the explicit sampling and record retention language of the Revised Code, leaving taxpayers with little room for error in their credit substantiation.

The meaning of a tax audit sample in the context of the Ohio R&D credit is thus one of high stakes. It represents the point where scientific innovation meets rigorous administrative scrutiny. For the taxpayer, the sample is the “stress test” for their entire R&D documentation system. For the state, it is the primary mechanism for ensuring that the “legislative grace” of the credit is only extended to those who can affirmatively prove, through contemporaneous and sitused records, that they have truly invested in the future of Ohio’s technological economy.

Final Thoughts