In a ruling dated February 6, 2012, the District Court for the Western District of Pennsylvania has denied Bayer’s Amended Motion for a Case Management/Protective Order Based on Statistical Sampling in its R&D credit case. Bayer is one of the largest pharmaceutical companies in the world and has been in a long standing dispute with the IRS over the R&D tax credit dating back to its 1990 year through 2006. Bayer has stated that it has incurred over $6 billion in research costs that fed into its qualified research expenditures for the research credit and has over $175 million in R&D credit at stake, according to the Motion.

This case is still at the discovery stage and both parties have hired statistical sampling experts from the same university – Carnegie Mellon, to try and work out an acceptable sampling plan to use stat sampling to bring to trial a fewer number of R&D items/activities/projects than claimed.

At issue is the business component test. Bayer has stated that is uses a cost center accounting method and has made no attempt to segregate costs by business component for any of its cost centers claimed for the R&D credit over the years, despite the government’s attempt to get them to do so, at least with respect to some cost centers.

The government does not want to be bound by Bayer’s proposed sampling approach to resolve the research credit, stating that this would relieve Bayer from its documentation requirement on the non-sampled items; particularly in identifying the business component test, which is the primary dispute between the parties at this stage of the case.

The government’s position is illustrated in the following quotes from the Motion:
Bayer has not established its entitlement to the claimed ORE credits because it has failed to identify the business components generating OREs during the credit years…Bayer has refused for some unknown reason to identify a single business component supporting its refund claim in the almost two years this civil action has been pending.

In response, Bayer believes it would be unduly burdensome and time consuming to break out their cost center expenditures by business component, stating there are likely over 100,000 business components at issue. Bayer also refused to date to break out any cost centers by business components, even a sampling of the larger ones.

The Court concluded that unless the government and the taxpayer can work out a mutually agreeable sampling plan, Bayer in this Motion cannot force the Court to compel the government to accept Bayer’s statistical sampling proposal.

As a procedural matter, the Court stated:

In the present case the Government disputes Bayer’s ability to meet its burden of identifying the business components for which the QREs claimed for the credit years were incurred, a requirement clearly set forth in Section 41 of the Internal Revenue Code. Until this burden is satisfied, quantifying the amount of QRE credits to which Bayer is entitled is premature.”

The District Court also acknowledged the Cohan rule cited in several recent R&D tax credit cases, which allows the Court to estimate in this case qualified research expenses (QREs), but reasoned that Cohan does not automatically apply; first, the taxpayer must demonstrate that it incurred some QREs before the rule can be applied to estimate the total amount of QREs, that is, “where the taxpayer has clearly shown that he is entitled to some deduction and that uncertainty exists only as to the exact amount thereof.”  Stated another way in this ruling, the Court concluded that Cohan should be applied only in cases where the taxpayer has clearly shown that he is entitled to some deduction and that uncertainty exists only as to the exact amount thereof.”

In sum, Bayer’s Motion was denied, meaning that Bayer’s proposed sample plan cannot be forced upon the government to adopt such a plan without mutual consent.

This decision leaves open the question of what happens next in the Bayer case?  Both parties are left with a significant amount of work ahead, and if all cost centers were brought to trial and no sampling utilized or agreed upon, that trial would take many years and judicial resources to resolve.

On the other hand, there have been a number of research credit cases over the years in which the parties have gotten together to work out either a stat sample or judgment sample plan to review and present at trial much less than 100% of the items claimed, including the Norwest, Union Carbide and Lockheed cases to name a few. Thus, the most likely resolution is the parties will end up working out a sampling plan acceptable to both of the Carnegie Mellon expert statisticians engaged in this case, and may involve some up front work on the part of Bayer demonstrating its ability to show QREs by business component for some selection of cost centers.

Moreover, the discussion in the Bayer decision demonstrates the need to assemble business component data as early as possible in the process, preferably when the work is incurred, to avoid additional hurdles down the road.

Please read the full decision attached below.

Bayer Corp v US