Writer Sends Copy of Article on Final Research Credit Regs

[2] My note identifies four factors by which the overall effectiveness of the R&E credit should be measured: The size of the economic "spillover effects" generated by the research for which the credit is granted, the effective rate of the credit, the elasticity of research activity with respect to the credit, and the quantity of administrative costs incurred due to the credit. The note then makes some general observations about the potential effects that varying interpretations of qualified research may have on the effectiveness of the credit. First, the adoption of a narrow interpretation of the definition of qualified research may well be justified insofar as it focuses the credit on research projects that generate larger spillover effects, and that have a higher elasticity with respect to the credit. Second, any change in the definition could potentially have an unintended impact on the effective rate of the credit. Third, while the proposed regulations aim to decrease administrative costs incurred by the IRS, some provisions may cause unintended increases in the administrative costs incurred by firms conducting research.

[3] Finally, my note examines some of the specific provisions of the December 1998 proposed regulations. It finds that the existence of a "discovery test" is justified because this test will focus the credit on research that creates larger spillover effects, as well as on more uncertain types of research that will be more responsive to the credit -- goals that are consistent with Congressional intent. It does, however, recommend that the discovery test include the use of existing technology in new and dynamic ways, and that a definition of "existing knowledge" be provided. It recommends that the definition of "process of experimentation" be expanded to include research that is not conducted by a specific process of developing and testing hypotheses. It agrees in principle with the concept of documentation requirements, but finds that the documentation requirements proposed here may incur administrative costs that are too high. It agrees with limiting the credit to research conducted before the beginning of commercial production, but recommends that the list of activities deemed to occur after the beginning of commercial production be more precisely defined. It also concurs with the approach requiring a "high threshold of innovation" for internal-use software projects, since most internal-use software projects will not contribute large spillover effects and will have a low elasticity with respect to the credit.

[4] The final regulations published in January 2001 (T.D. 8930) ("the January regulations") incorporate a number of changes to key provisions that bear on the effectiveness of the credit. Therefore, I offer here some comments on these provisions.

I. DISCOVERY TEST

[5] The January regulations have preserved the concept of a "discovery test." This test is consistent with an effective credit for two reasons: First, the type of research that expands or refines existing knowledge is likely to create significant spillover effects for the economy, while other research is less likely to do so. Second, research that meets this test is more likely than other research to carry a high degree of uncertainty for the firm conducting the research. Thus, it is less likely that such research would have been carried out in absence of the credit, and more likely that the existence of the credit is encouraging the research to be conducted. This will enhance the elasticity of the credit. Therefore, this test should be retained in the final regulations.

[6] One important change to the discovery test in the January regulations is the definition of common knowledge in section 1.41- 4(a)(3)(ii) as "information that should be known to skilled professionals had they performed, before the research in question is undertaken, a reasonable investigation of the existing level of information in the particular field of science or engineering." The proposed regulations had not offered a clear definition of what information would be considered common knowledge, an omission that threatened to impose large administrative costs on both taxpayers and the IRS in the process of proving whether research would qualify as expanding or refining existing knowledge. The current definition is as specific as could reasonably be expected, and is likely to enhance the effectiveness of the credit through reduced administrative costs. It clearly states that trade secrets are not within common knowledge, which is an important distinction as far as the effectiveness of the credit is concerned, because the rediscovery of knowledge held in secret by other taxpayers will produce spillover effects much larger than the rediscovery of knowledge that is already in the public domain. Similarly, section 1.41-4(a)(iv) of the January regulations provides a safe harbor for patented research, which will remove a large number of projects from potential dispute that are likely to have beneficial spillover effects, thus contributing to the positive spillover effects produced by the credit as well as lowering administrative costs.

[7] The January regulations have also added section 1.41- 4(a)(3)(iii) which provides that research "may employ existing technologies in a particular field and may rely on principles of science or engineering," and still satisfy the discovery test. A key issue raised in litigation has been whether the use of existing technology in new and dynamic ways can satisfy the discovery test. See Norwest v. Commissioner, 110 T.C. 454, 515 (1998); Wamsley, supra, at 173. From the perspective of efficiency, it is desirable to allow such projects to qualify, because they can be the type of innovative projects that are likely to create spillover effects, even if they do not actually employ any new technology. See Wamsley, supra, at 197. The newly added provision is therefore likely to make the credit more effective by including these projects and their spillover effects.

II. PROCESS OF EXPERIMENTATION

[8] The January regulations eliminate the requirement contained in the proposed regulations that research must follow a specific four-step method of developing and testing hypotheses in order to qualify as a process of experimentation. See section 1.41-4(a)(5). This requirement was likely to reduce the amount of positive spillover effects generated by the credit, because many research activities that discover the type of information that produces significant spillover effects are not carried out in a scientific laboratory according to a strict scientific method. See Wamsley, supra, at 198-99. This broadening of the definition of the process of experimentation should enhance the effectiveness of the credit by allowing additional beneficial projects to qualify.

III. DOCUMENTATION REQUIREMENTS

[9] The January regulations also eliminate the requirement that taxpayers record the results of each experiment conducted. A requirement for some level of documentation can be justified as a way to lower the cost of administrating the credit. When insufficient documentation has been kept, IRS auditors incur added expenses in determining whether research projects exhibited the required level of uncertainty. The documentation requirement in the December 1998 regulations posed a problem similar to the requirement of following a specific laboratory process: Many beneficial projects are not conducted according to a strict laboratory method, and do not require the creation of laboratory-style records. Any gain in efficiency from this documentation requirement would likely have been outweighed by the costs that it imposed upon firms in the form of requiring them to create documentation that would otherwise not have been created, so the elimination of this requirement was well-advised.

[10] The new documentation requirement in the January regulations states that the taxpayer, in order to create a rebuttable presumption that the research activity satisfies the discovery test, must create contemporaneous documentation of the research activity and of the taxpayer's belief that this research would exceed, expand, or refine the common knowledge of skilled professionals in the field. See section 1.41-4(a)(3)(v). This will be much easier to meet than the previously proposed standard, as it does not require any specific form of recording experimental results. One can reasonably assume that a firm spending significant dollar amounts on research of a technically uncertain nature will be keeping some sort of records noting that the research is being conducted, and explaining why the research is being conducted. Thus, this documentation requirement is not likely to result in firms creating large numbers of records which they would not have created in absence of a documentation requirement, and so will not cause a large increase in administrative costs. It therefore appears to promote a reasonable balance between the desire to reduce the cost of administering the credit and the need to avoid increasing the administrative costs imposed on firms.

IV. RESEARCH CONDUCTED AFTER COMMERCIAL PRODUCTION

[11] The January regulations retain a list of activities that are deemed to occur after the beginning of commercial production, and are thus excluded from credit eligibility. See section 1.414(c)(2)(ii). Such a blanket disqualification may be advantageous, because activities that are close to the beginning of the production of a profitable product are likely to carry less uncertainty. than activities conducted at an earlier stage, and thus they are activities that would be more likely to be carried out even in absence of the credit. Excluding such activities may therefore increase the elasticity of the credit.

[12] The concern with automatically excluding such items, however, is that some research activities that do exhibit a high level of technical uncertainty may be interpreted to fall under the excluded terms. See Wamsley, supra, at 200-01. In response to similar concerns, the January regulations removed "correcting flaws" in a business component from the list of excluded items. The final regulations should go one step further, and provide definitions for terms such as "trial production runs," "trouble shooting," and "debugging flaws," in order to ensure that these items are not interpreted so broadly as to exclude some projects that have a high level of technical uncertainty.

V. INTERNAL-USE SOFTWARE

[13] The January regulations modify the definition of internal- use software to provide an exception for software that provides customers with a significant new feature with respect to a non- computer service that is not available from any of the taxpayer's competitors. This exception should operate in a manner analogous to allowing the use of existing technology in new and dynamic ways under the discovery test: If the use for which the software is applied is providing innovative new functionality, then the research is likely to produce significant spillover effects, regardless of whether the underlying technology was highly innovative. Therefore, this change is likely to enhance the effectiveness of the credit.

[14] In addition, the January regulations modify the language of the "high threshold of innovation" test to clarify that no separate requirement of a high threshold of innovation exists separately from the three-part test of innovativeness, significant economic risk, and commercial availability. See section 1.41-4(c)(6)(vi). This is a significant change, which may affect the interpretation of the innovativeness test. In the past, differing interpretations of the innovativeness test have been offered. The Tax Court, in Norwest, makes clear that in determining whether research activity has been intended to produce a reduction in cost or improvement in speed that is "substantial and economically significant," a "high threshold of innovation" is required, because "the extent of the improvements required by Congress with respect to internal-use software is much greater than that required in other fields." 110 T.C. at 498-99. Thus, internal-use software that provides only a modest improvement over existing software will not qualify for the credit. See also WICOR, Inc. v. U.S., 116F. Supp. 2d 1028, 1036-37 (relying on Norwest in holding that "[e]ven though the project resulted in lower operating costs for WICOR, this is not sufficient to satisfy the test for innovativeness, which requires a high degree of innovation, not just a reduction in operating expenses"). The Seventh Circuit, in United Stationers v. United States, 163 F.3d 440 (7th Cir. 1998), applies a somewhat broader standard, holding that internal-use software must be "more clearly innovative than software merely meeting the section 41 requirements in general," but finding that all of the activities under consideration met this standard because they "resulted in improvements in speed or reductions in cost." 163 F.3d at 448. This upheld the district court's holding that all of the activities qualified based on the "plain meaning" of "substantial and economically significant." United Stationers v. United States, 982 F. Supp. 1279, 1287 (N.D. Ill. 1997). Following the Tax Court's narrower interpretation is likely to result in a more effective credit than the adoption of a more permissive standard. Internal-use software research should be held to a requirement of a high threshold of innovation, because most such software would be developed even in absence of the credit (thus its inclusion will decrease the credit's elasticity), and because the product is never disseminated to the general public (thus the spillover effects from such research are likely to be smaller than for other types of research.). See Wamsley, supra, at 192-93, 201-02. Therefore, the effectiveness of the credit would be enhanced by allowing eligibility only for internal-use software that represents an innovative advance in technology, rather than for every project that produces a significant, but incremental, improvement over the software in use by other firms in the industry.

[15] If the words "high threshold of innovation" in the January regulations are read to be merely a label for the three-part test, rather than as an integral part of that test, then the regulations can be read as adopting the United Stationers version of the innovativeness test (or an even more inclusive version of the test), allowing any significant improvement in cost or speed to qualify. Therefore, the final regulations should define the terms "substantial" and "significant" to clarify that a high threshold of innovation is required to satisfy the innovativeness test, rather than significant, but incremental, reductions in cost or improvements in speed.

VI. CONCLUSION

[16] The January regulations contain a number of significant improvements as compared to the regulations proposed in December 1998, and their adoption will enhance the effectiveness of the R&E credit as a policy instrument. Two additional modifications could be made in the final regulations, however, that might further this objective: First, the final regulations should provide definitions for some of the items in the list of activities deemed to occur after the beginning of commercial production, so that they do not exclude research activity that actually does exhibit a high level of technical uncertainty. Second, the three-part "high threshold of innovation" test should define the term "substantial and economically significant" in such a way as to clarify that a high threshold of innovation is required to meet this standard.

[17] Thank you very much for considering these comments. I hope you will find them helpful as you work to finalize the regulations.

Sincerely,