Contrasting positions demonstrate a philosophical difference between the government and large taxpayers with regard to the research credit.
Research tax credit examinations often involve producing a large volume of documents to substantiate the expenses giving rise to the credit. For large taxpayers, a statistical sampling method could help expedite an IRS examination of the credit. However, in Bayer Corporation and Subsidiaries v. United States, Civil No. 09-351 (W.D. Pa. 2012), a recent District Court opinion focuses on whether the IRS will agree to the company’s use of statistical sampling to expedite an IRS examination of the research tax credit—or if it should impose such a method in pre-trial discovery.
Research Tax Credits and Claims for Refund
Bayer Corporation is well known for a variety of products, from aspirin to plastics to fertilizers, and employs over 20,000 people in the U.S. Research is conducted by Bayer at 49 locations across the country, and qualified research expenses (QREs) were calculated using a cost center approach. Under this approach, QREs are calculated by costs incurred in various research departments, without necessarily tracking direct costs to specific projects. Industries such as the pharmaceuticals industry commonly track research costs this way, as entire departments may be devoted to nothing but research and experimentation. On its originally filed tax return, the company claimed research tax credit for tax years 1990 through 1994.
Bayer originally calculated its research credit by reviewing core research departments within Bayer and determining QREs. When it came to the attention of Bayer’s tax director that other departments within the organization were engaged in research as defined in Section 41(d) of the Internal Revenue Code (“Code”), the company engaged an accounting firm to conduct a research tax credit study for all years that were open under the statute of limitations.
In 1997, Bayer was completing an IRS examination of its 1990-1991 and 1992-1994 cycles. The tax years were still open under the statute of limitations by consent, and Bayer and the IRS had agreed on the allowable amount of research tax credits claimed on Bayer’s originally filed tax returns for those years. In 1998, Bayer filed a claim for a refund based on additional research expenses uncovered in the course of the tax credit study. The IRS denied not only this claim, but also the original tax credits agreed upon in the 1990-1991 and 1992-1994 audit cycles.
The numbers are significant. For tax years 1990 through 1994, the amount of the credits claimed was $80,361,674, a number initially agreed upon by the IRS, but now at issue. Based on the tax credit study, Bayer sought a refund for the amount of $49,236,589. Due to carryovers of the tax credit from years in which Bayer had net operating losses, the initial credit and additional claim for the refund covered tax years 1987 through 2006. In addition, the inclusion of new research departments in calculating the credit caused its base period, from 1984 through 1988, to change, subjecting those years to further analysis. Overall, at issue was $142,921,443 in taxes and interest. Due to amended returns, the carryover and carryback of research tax credits, and potential adjustments to the base period calculation, the controversy covered the years 1984 through 2006.
Since 2007, claims for refunds based on the research tax credit have been deemed Tier I issues by the IRS. All refund claims are coordinated within the IRS, and examinations of refunds based on the research tax credit are often contentious. After denying the refund claim, Bayer’s only recourse was to sue.
Chicken, Egg and Business Component
Pre-trial discovery demonstrated a vast difference between Bayer and the government in their approach to the research tax credit. Bayer kept records on a cost center basis, reflecting the research sites it maintains throughout the country. The government prefers that taxpayers keep track of research costs based on “business components”—for each product, process or invention. But for many organizations, the early stages of research are conducted without assigning the research to a specific product.
The government asked Bayer to “identify and describe each new or improved business component Bayer contends it incurred qualified research expenses to develop during the credit years.” Bayer objected, noting that “during the credit years, Bayer estimates that it developed more than 100,000 business components, which Bayer’s books and records do not, and are not required to, track individually.” In a sense, the parties are begging the question of which came first: the research, which if successful, results in a business component, or a proposed new product at which the research effort is directed? In other words, what comes first, the chicken—the business component, or the egg—the research?
In order to manage the complex litigation, the District Court appointed a Special Master to oversee discovery. The government filed a motion to compel answers to its interrogatories; in response, Bayer filed a motion for a case management/protective order based on statistical sampling. The firm advocated taking a statistical sample of its research expenses from its various research locations in order to limit the production of documents. Bayer argued that agreeing to a statistical sampling method would “enable the parties to conduct all of the necessary discovery and allow the Court to decide the entire case in the next two to three years.”
Statistical Sampling
In cases with large volumes of data, statistical sampling has been utilized in the past by the IRS and taxpayers. For example, in examinations of the deduction for domestic production activities, statistical sampling is permitted “if the taxpayer can demonstrate a compelling reason for its use.” The Internal Revenue Manual provides: “Statistical sampling techniques are valuable examination tools where effective use of resources makes it uneconomical to audit voluminous accounting data.” The IRS’ Audit Techniques Guide permits statistical sampling “in a research credit case whenever excessive amounts of time or resources are anticipated in examining all of the taxpayer’s expenses or projects.” Courts have also accepted sampling in situations where the taxpayer and the IRS agree to the sample size and methodology.
Bayer’s research credit issue appeared to be an excellent candidate for statistical sampling. The firm estimated it had spent more than six billion dollars in research expenses in 1,300 cost centers at its 49 research sites. The court entertained a proposed sampling of expenses based on four major research sites, but the government renewed its request for Bayer to identify all of its “business components” at the four sites before it would consider statistical sampling. Bayer argued that it was not worthwhile to identify “business components” until a methodology for statistical sampling could be agreed upon and asked the Special Master to impose sampling as a protective measure to reduce burdensome discovery. Because the parties could not reach an agreement, the Special Master asked the District Court for guidance.
Bayer’s Request Denied
Abruptly, the court refused to grant Bayer’s request for protective relief, to order statistical sampling on a limited basis. Instead, it ordered that Bayer respond to the government’s request to identify “business components” as a first step in the discovery process. In its 49-page opinion, the court based its decision on three points. First, Treasury regulations require that Bayer maintain records “to substantiate that the expenditures claimed are eligible for the credit.” Allowing Bayer to limit its production of documents to a small sample would relieve the firm from these obligations. Second, the court was not persuaded that the time and expense of document retrieval was “undue,” requiring a protective order. Much of the expense of gathering information was related to Bayer’s outdated computer systems. Finally, the court concluded that the government’s interests in determining Bayer’s “business components” outweighed any burden of discovery imposed on Bayer. Therefore, it refused to impose statistical sampling to limit pre-trial discovery.
Impact on the Research Tax Credit
The Bayer case involves solely a request to limit pre-trial discovery by using statistical sampling to narrow the scope of evidence. However, the contrasting positions of the parties demonstrate a philosophical difference between the government and taxpayers with regard to the research credit, and serves as a warning signal to taxpayers claiming this credit.
The government’s focus on defining a business component often overrides the methodology used to compute the credit. It is hard to argue that Bayer is not engaged in research and development at some level, but the government’s insistence on identifying “business components” was vastly different than the cost center approach. Based on the court’s opinion, Bayer is likely to have to reorder its research expenditures to conform to the government’s position.
While the opinion is not dispositive of the entire litigation—the court only addressed Bayer’s request for statistical sampling—it does provide some caution for taxpayers pursuing the research tax credit or any refund claim, such as:
Use of statistical sampling. Statistical sampling is a valid method of handling issues with large quantities of data to be analyzed. In certain situations, the IRS has allowed statistical sampling, provided it is agreed upon and allows a level of confidence in the sample size and analysis. In situations where sampling may be useful, taxpayers should reach an agreement at the IRS examination level. Although it is unclear why sampling was not done in the initial examination of Bayer, asking the court to impose statistical sampling was a tall order.
Refund claims. Research tax credits claimed on an original return have not been subject to the same level of contentiousness as claims for refunds. On a refund claim, when the parties cannot agree, the taxpayer’s remedy is to sue for a refund in U.S. District Court or Court of Claims. Litigation in those courts is handled by the Justice Department, whose attorneys are aggressive in pursuing trial tactics but have less interest in substantive tax issues. Consequently, a case handled by the Justice Department is usually subject to contentious litigation procedures and is less likely to be settled. For taxpayers filing a refund claim, all administrative remedies, including appeals and mediation, should be pursued before turning the issue over to litigation.
Documentation. A perennial disagreement between the IRS and taxpayers is the level of documentation required to substantiate the research tax credit. The IRS has proposed strict requirements, but Treasury regulations pertaining to the credit provide only a loose requirement to maintain records in accordance with general record keeping in section 6001. In identifying business components and expenditures used in computing the credit, it is important to preserve substantiation early in the process in the event of a subsequent examination.
The Bayer litigation will continue, as the court’s opinion only addresses the course of pre-trial discovery and not the amount of its research tax credit. However, the court’s opinion demonstrates the contentious nature of refund claims in general and the need for contemporary documentation. iBi
