Navigating tax changes with statistical sampling

By Hamid Ashtiani; Michael Del Medico, CPA; and Wendy Rotz, PStat, Washington, D.C.

Editor: Greg A. Fairbanks, J.D., LL.M.

Continual shifts in the tax landscape provide challenges for taxpayers. However, these challenges create space for innovative solutions and opportunities, along with tax planning considerations. Statistical sampling aids tax departments in cutting costs and gaining efficiencies by estimating reliable values for tax returns. This item discusses issues taxpayers face and describes how they may meet challenges by applying sampling and estimation.

Common threads run through the many applications of statistical sampling in tax. Generally, when a taxpayer uses statistical sampling, a list of records requires facts-and-circumstances determinations to correctly assign costs to "buckets." Sampling is advantageous when completing the entire list is an onerous task, either because the list is voluminous or even bucketing the costs for a single item on the list takes a great deal of effort. The list is sampled, tax determinations are made with respect to the sampled items to bucket the costs in the sample, and these sample results are used to extrapolate quantities for the entire listing.

Taxpayers' use of statistical sampling has been accepted by the IRS for decades. According to Rev. Proc. 2011-42, statistical sampling and estimation is appropriate when (1) there is no better probative information and (2) the time and cost of analyzing large volumes of data are excessive.

The key point to remember is this: If it can be listed, it can be sampled.

So, when a tax department finds itself making burdensome determinations on a list of records, it may be time to consider a sampling approach.

This covers a wide range of tax applications — many of which tax departments should be considering, especially in light of recent tax law changes. The tax issues range from inventory-related, such as for uniform capitalization (UNICAP), to computing credits and deductions, such as those for research and development (R&D), business-related meals and similar expenses, and transportation and parking expenses. They include depreciation of tangible property with cost-segregation studies and expensing repairs. Applications even include international tax issues, such as the Sec. 965 transition tax, global intangible low-taxed income (GILTI), and the foreign-derived intangible income (FDII) deduction. These are discussed below.


The UNICAP rules under Sec. 263A require taxpayers generally to capitalize direct and indirect costs that are allocable to property produced or acquired for resale, unless the taxpayer has aggregate average annual gross receipts of $25 million or less ($26 million for 2019) (Sec. 263A(i)(1)). The UNICAP rules can be tedious to apply, especially for moderate- to large-size companies, as it generally requires businesses to capitalize additional costs beyond those capitalized in their books and records. Total Sec. 263A costs are generally the costs capitalized under Sec. 471 plus additional Sec. 263A costs.

Historically, most companies were able to easily identify their Sec. 471 costs because they would typically use their book method. However, with final regulations issued in 2018 (T.D. 9843), taxpayers now generally must start with their book method, but they must adjust it to treat all direct costs as Sec. 471 costs, regardless of whether the direct costs are capitalized for book purposes (Regs. Sec. 1.263A-1(d)(2)(iii)). As a result, many taxpayers may be required to adjust their Sec. 471 costs for items such as trade discounts and freight-in, which might not be included in their book method. Developing a tax method for identifying and allocating these costs can be burdensome, particularly because the IRS does not allow lump-sum allocations of Sec. 471 costs.

Determining the amount of additional Sec. 263A costs that must be capitalized in addition to the Sec. 471 costs has always been a significant challenge and, under the final regulations, it still is. Between the new definition of Sec. 471 costs and additional Sec. 263A costs, there are many inclusions and exclusions to consider.

There are also new opportunities, though. Since the IRS issued the final regulations, companies have been reassessing their UNICAP calculations for more tax-favorable approaches. For example, many taxpayers are teasing out their labor costs for "pick and pack" order-fulfillment services that do not require capitalization. Others are looking at implementing more favorable ways of determining "mixed-service costs," such as human resources or IT, that are partially allocable to capitalizable activities and partially allocable to noncapitalizable activities. For companies with many cost centers, determining the amount of costs allocable to capitalizable vs. noncapitalizable buckets can involve a great deal of effort.

Statistical sampling can reduce the burden of bucketing these costs. A typical sampling solution begins by listing the company's cost centers. A sample is drawn from the list. The additional Sec. 263A costs are determined only for the sampled records and then extrapolated to the entire listing to estimate additional Sec. 263A costs for the whole company. There are variations of this approach. For example, sometimes the indirect costs and/or the mixed-service costs or even the Sec. 471 costs require extrapolation. In the case of trade or other discounts, a list of vendors and their credits is compiled and sampled. The sampled vendor credits are classified as either qualifying or nonqualifying as trade or other discounts, and the bucketed sample is used to estimate qualifying discounts for the entire company, thereby reducing the cost of merchandise acquired for resale.

Also note that determining a company's R&D expenditures (see below) impacts the UNICAP calculation because these costs do not require capitalization.


As the dust settles from the law known as the Tax Cuts and Jobs Act (TCJA), P.L. 115-97, taxpayers who might not have claimed R&D in the past, as a research tax credit (RTC) under Sec. 41 and/or expense under Sec. 174, might now consider it, and not just due to UNICAP. First, more taxpayers may be eligible than before because the TCJA eliminated the alternative minimum tax for corporations — which removed a hurdle that once prevented some taxpayers from taking advantage of credits. Second, more taxpayers may need credits and deductions than before because the TCJA limits deductions for net operating losses to 80% of taxable income — driving some taxpayers to seek new sources of tax savings, such as the RTC, to offset their regular tax liability. Third, the reduced corporate tax rate indirectly allows more R&D credit to be claimed when the reduced credit under Sec. 280C(c) is elected. There is now only a 21% reduction in credit when applying Sec. 280C(c), while it was a 35% reduction in the past because of the then-higher top corporate income tax rate. The result is that taxpayers receive a larger credit for the same amount of qualifying research expenditures (QREs).

Gathering the audit-ready documentation required to support R&D activity to the satisfaction of an IRS exam team is key to a successful study, and it does not need to be an onerous task. Statistical sampling can reduce that effort. In common approaches, wage, supply, and/or contract expenditures are listed by project, department, employee, or task. The expenditure listing is sampled. For each sampled record, the expenditures are bucketed into QREs and non-QREs. The qualifying activity supporting the QRE determinations is thoroughly documented. The sampled QREs are then used to estimate total QREs for the entire listing.

Meals and other business expenses

The TCJA amended Sec. 274 by changing the rules for deducting meals, entertainment, and travel expenses. Entertainment is not deductible for expenses incurred after 2017 (Sec. 274(a)(1)). Because Sec. 274 divides expenses into many categories, companies may find it difficult to determine which expenses are fully deductible, partially deductible, or nondeductible. Companies with thousands of expenses may find their only feasible means of taking a meals or other allowable deduction is with a sampling approach. To do this, the expenses are listed, a sample is drawn from the list, and the expenses in the sample are bucketed into fully deductible, partially deductible, or nondeductible categories. These sample results are then used to extrapolate the deductible amounts for the entire listing.


The TCJA also changed rules for deducting parking and other qualified transportation fringe benefit expenses. Companies are now generally required to carve out the portion of their parking expense that is for employee use, as this expense is no longer deductible (Sec. 274(a)(4)). For companies that lease or own their own parking lots, this means determining the portion of their lots that is either reserved for employees or nonreserved but primarily used by employees.

Sampling can help ease this task. A list of parking lots is compiled, and the list is sampled. The portion of spaces for employee use is determined in the sampled lots. This portion per lot is multiplied by the expense of the lot to bucket each sampled lot's parking expenses into employee and nonemployee use. The sample findings are then extrapolated to the whole list to estimate the amount of parking expenses that is nondeductible.

Tangible property

No discussion of sampling applications in tax would be complete without mentioning tangible property and cost segregation. Cost segregation increases near-term cash flow by accelerating the return on capital from investment in property, thus reducing income taxes. Cash-strapped companies will often consider cost segregation in lieu of borrowing cash.

It works by examining the acquisition or construction costs and then separating out personal property and land developments, which can be placed into shorter depreciation recovery periods. With little extra effort, expensing repair costs and identifying dispositions can be done in conjunction with a cost-segregation analysis for added value to a study.

The IRS Cost Segregation Audit Techniques Guide specifically addresses sampling and estimation techniques. Retail chains, real estate investment trusts, and banks with multiple properties especially find a sampling approach advantageous.

A typical study will list locations, projects, or assets and the cost basis associated with each record. A sample is drawn from the list. Then, just for the sampled items, tax professionals determine how much of those costs should be expensed as repairs, or placed into five-, seven-, 15-, 27.5-, or 39-year depreciation categories, or other recovery period as appropriate. Then from the sample results, a statistician extrapolates how much belongs in each category for the entire listing. The estimates can even be made down to the asset level for depreciation calculations and uploading of a new fixed-asset listing after the study's completion.

Transition tax under Sec. 965

The TCJA imposed a mandatory one-time tax on previously untaxed foreign earnings of controlled foreign corporations (CFCs) to both transition to a new "quasi-territorial" tax regime and to raise revenue to pay for various tax cuts included in the act. The transition tax was structured as a mandatory inclusion of earnings under Sec. 965 with a corresponding deduction to create a reduced rate of tax. By now, taxpayers should have included their Sec. 965 transition tax liability on either their 2017 or 2018 income tax return. The IRS has instituted an audit campaign targeting the computation of the Sec. 965 tax liability, and as a result, the provision may remain relevant for certain taxpayers.

An IRS examination of the transition-tax liability computation may involve scrutiny of the computation of accumulated post-1986 earnings and profits (E&P) and foreign income taxes claimed, covering up to 31 tax years for each deferred foreign income corporation and E&P deficit foreign corporation of a U.S. multinational corporation. Substantiating these amounts could be a burdensome task for a taxpayer, especially if a company did not properly adjust its E&P under certain accounting standards, e.g., Regs. Sec. 1.964-1(c)(6).

Due to the relative proximity of the release of the proposed regulations under Sec. 965 and the extended due date of 2017 tax returns for calendar-year taxpayers, it is very likely that errors will be uncovered in the computations of Sec. 965 liabilities. Multinational corporations should have calculated their E&P, cash balances, and creditable foreign income taxes for each entity. However, to save time and effort, some taxpayers may have used shortcuts that could be challenged by the IRS during exam. This, in combination with the vast number of "entity-years" (i.e., number of entities multiplied by number of tax years), could result in large, unreserved adjustments.

Statistical sampling is an effective means for a company to assess any potential tax liability or savings, estimate reserves based on sound figures, or file an amended return using IRS-approved methods. To accomplish this, the hundreds or thousands of entity-years are listed, and a random selection of entity-years is drawn from the listing. Only the sampled records require substantiation. If corrections are warranted, the E&P, cash, and creditable foreign income taxes can be redetermined just for the sample records. The sample results can then be used either to support the originally filed returns or to extrapolate corrected values of E&Ps and foreign taxes to recompute the transition tax that may either go on an amended return or be used for reserves.

Global intangible low-taxed income

The GILTI regime was enacted as part of the TCJA and is designed to act as a global minimum tax. To accomplish this, earnings of CFCs that are deemed in excess of a "routine return" on their fixed assets are subject to current taxation. For these purposes, foreign earnings are computed using U.S. taxable income concepts, referred to as "tested income."

The computation of taxable income for each CFC may prove difficult for large U.S. multinationals. The challenges of burdensome facts-and-circumstances determinations discussed abovefaced by domestic taxpayers when computing taxable income also are faced by multinationals when computing the tested income of a CFC. For purposes of GILTI, taxpayers may consider using the aforementioned statistical sampling techniques, as applicable. For example, sampling methods for estimating UNICAP and meals and other deductions may be applied to a tested-income computation.

Foreign-derived intangible income

The TCJA provides an incentive for domestic C corporations that earn FDII, which is designed to counterbalance the GILTI regime. Generally, this provision provides a deduction of 37.5% for certain foreign-derived deduction-eligible income for tax years beginning after Dec. 31, 2017, resulting in a 13.125% effective tax rate on FDII (21.875% for tax years beginning after Dec. 31, 2025).

The determination of whether income qualifies for the special deduction is generally made by analyzing whether that income is derived in connection with the sale of property to a foreign person for a "foreign use," or if the income results from a service to any person, or with respect to any property, outside the United States. Varying documentation requirements must be met with respect to sales and services income to substantiate that the income qualifies.

Sales of certain property, due to their fungible nature, may be difficult (or impossible) to trace to their location of use or consumption. As such, producing adequate documentation to demonstrate that the foreign-use requirement has been met in this context may provide a challenge. This scenario provides another opportunity for taxpayers to use statistical sampling to solve a TCJA-related problem. The proposed regulations under Sec. 250 provide a de minimis rule that provides that a seller may treat an entire "fungible mass" of multiple items of property as for a foreign use if it can use market data (including statistical sampling) to establish that 90% or more is for a foreign use (Prop. Regs. Sec. 1.250(b)-4(d)(3)(iii)). To apply a statistical approach, sales are listed together with their amounts. The list is sampled, and the sales in the sample are bucketed into foreign or domestic use. These amounts are then extrapolated up to the entire listing to estimate the portion qualifying as foreign use.

As noted above, the IRS went so far as to suggest using statistical sampling in such determinations for a fungible mass of goods. It is important to note that the use of statistical sampling is not limited to a fungible mass and may be used in a number of other contexts with respect to FDII that are outside of the scope of this discussion. Taxpayers performing a burdensome task for a list of records should consider a sampling approach.

Lightening the burden

Sampling is an efficient tool to master today's challenges in tax. Additionally, sampling has advantages in an IRS examination. Not only does the IRS accept sampling, but as long as taxpayers follow valid statistical approaches, the IRS will review the taxpayer's sample selections rather than draw its own sample in an audit. So an added bonus to sampling is that taxpayers know which items will be reviewed in case of an IRS audit. It is easier to prepare solid, thorough documentation for a sample than for an entire list. Therefore, audit risk can actually be reduced through sampling.

It is highly recommended that taxpayers involve a statistician familiar both with statistical estimation in tax and with IRS preferences. While Rev. Proc. 2011-42 does list commonly accepted methods, the IRS accepts more advanced techniques that can be more efficient in some settings. Pitfalls for the unwary, however, may lead an IRS agent to disallow a sample or may result in less advantageous estimates. So it is best to involve statisticians knowledgeable of sampling in these tax settings.

Applications of sampling are not restricted to those in this discussion. When taxpayers are confronted with a burdensome task, making facts-and-circumstances determinations, and bucketing amounts for a list of records, just remember: If it can be listed, it can be sampled.


Greg Fairbanks, J.D., LL.M., is a tax managing director with Grant Thornton LLP in Washington.

For additional information about these items, contact Mr. Fairbanks at 202-521-1503 or

Contributors are members of or associated with Grant Thornton LLP.

Tax Insider Articles


Business meal deductions after the TCJA

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


Quirks spurred by COVID-19 tax relief

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