Most observers of the tax policy scene would consider the proposal suggested in the title of this article shocking. After all, everybody loves the research credit. Economists who dislike most tax subsidies support the research credit because it is a justifiable policy response to the market's failure to take into account spillover benefits of private research. Politicians like the credit because it is a tax break for deep-pocket beneficiaries. Even voters who abhor "corporate welfare" have a soft spot for government support of technological innovation. Equally favored by Republicans and Democrats, it has almost sacrosanct status on Capitol Hill. And even though Congress plays a phony game of cat-and-mouse with its serial short-term extensions of the credit, it is -- as far as subsidies go -- one of the surest bets in town.
Nevertheless, the credit is crumbling right before our eyes. Recent developments make clearer what we always knew: It is impossible to devise a practical definition of research. This article provides an overview of five recent court decisions. Collectively they demonstrate that the credit is becoming almost impossible to effectively administer.
The current definition of expenditures that qualify for the credit is vague and uncertain. Attempts to apply it to real-world circumstances require in-depth fact-finding, an examination of complex technologies, and evolving techniques for developing those technologies. It also requires the application of a fair amount of subjective judgment. This is not just an administrative problem, it is an economic problem. As this article explains, vagueness undermines the economic incentive effect of the credit.
Sure, it is a political challenge with staggering odds. But given the enormity of our economic problems and the lameness of our legislative process, isn't it time to start thinking outside the box? The credit should be repealed to pay for a 1 percentage point reduction in the corporate tax rate. Alternatively, the credit could be replaced with an incentive whereby a bright line can be drawn around research expenses most likely to result in the creation of socially useful knowledge. One possibility would be a flat credit equal to 25 percent of salaries paid to accredited scientists and engineers.
Part 1. Recent Cases
Over the last 11 months, courts have handed down five decisions concerning the definition of research qualified for the credit. In general, the cases have been resolved in favor of taxpayers. The cases demonstrate the inherent difficulty in determining qualified research and the inability of Congress, Treasury, and the IRS to adequately deal with the issue even after 29 years of experience with the credit.
McFerrin (June 9, 2009)
In United States v. McFerrin et ux., 103 A.F.T.R.2d 2009-2566 (5th Cir. 2009), Doc 2009-13123, 2009 TNT 109-15, Arthur McFerrin, a prominent chemical engineer, and his wife spent years developing new petrochemical products, but they did not document their expenses. The expenses in question were incurred in 1999. In 2003 McFerrin hired Alliantgroup to conduct a study of his 1999 research activities, filed an amended return, and received a tax refund. In 2005 the government, claiming the refund was issued in error, sued in district court and won.
But the Fifth Circuit Court of Appeals overturned that decision. Based on McFerrin's professional reputation and published articles, and the McFerrins' subsequent credible testimony, the court allowed the couple's salaries for time spent on research to qualify for the credit. This is an application of the Cohan rule -- named for the 1930 court ruling that allowed the performer and songwriter of the musical Yankee Doodle Dandy to deduct travel and entertainment expenses for which he kept no records.
The Fifth Circuit's ruling on the Cohan principle is a major blow to the IRS because so many claims for the research credit are made by taxpayers on amended returns. Typically taxpayers are not aware that activities could be eligible for the credit and accordingly don't keep appropriate records for claiming it. (The applicability of the Cohan rule to the research credit was also affirmed in Union Carbide, as described below.)
Practitioners are not paid to concern themselves with policy, but it is important for the rest of us to understand that economics would lead us to a conclusion opposite that reached by the court. The purpose of the credit is to provide incentive for research. If the taxpayer was not aware the credit was available, and if the credit played no role in the financial calculus of the firm when research was being performed, it could not have any incentive effect on that research. Ex post facto awarding of the research credit is just a windfall that may be legally justified but does nothing to promote research.
On economic criteria alone, Congress would be justified in restricting research credits to companies that demonstrate some evidence that they were aware of the credit's applicability to activities before or during the performance of those activities. Credit claims that make their first appearance on amended returns should be automatically disallowed.
Union Carbide (March 10, 2009)
In Union Carbide Corp. et al. v. Commissioner, T.C. Memo. 2009-50 (Mar. 10, 2009), Doc 2009-5285, 2009 TNT 45-5, the chemical giant made a claim for additional credits arising from 106 projects conducted in the taxpayer's manufacturing facilities in 1994 and 1995. After reviewing five of the largest projects, the court ultimately determined in a 298-page decision that wages generating only $1,045 worth of credit were qualified. Expenses for supplies that would have generated more than $20 million in credits were not deemed qualified. The court considered these supplies related not to a process of experimentation but to ordinary production.
Despite this rebuke to Union Carbide, practitioners have been quick to declare the case a victory for taxpayers. (See, for example, Charles J. Medallis and Jay Hite, "Union Carbide: Good News and Guidance on the Research Credit," Tax Notes, Oct. 12, 2009, p. 213, Doc 2009-20500, or 2009 TNT 197-13.) Among the results of the decision welcomed by practitioners were the affirmation of the Cohan rule, the clarification that the relatively lax 2003 final regulations apply even to claims made before those regulations were issued, and the possibility that costs of experimental trial runs necessary to eliminate uncertainty could qualify for the credit.
FedEx (June 9, 2009)
Another major difficulty is determining which types of software development are eligible for the credit. The Tax Reform Act of 1986 generally denied credits for software developed and used exclusively by a single taxpayer. But the IRS was given regulatory authority to provide exceptions. Concerned about IRS foot-dragging and reticence, the conference report to the Ticket to Work and Work Incentives Improvement Act of 1999 gave the agency a little nudge:
The conferees . . . note the rapid pace of technological advance, especially in service-related industries, and urge the Secretary to consider carefully the comments he has and may receive in promulgating regulations in connection with what constitutes "internal use" with regard to software expenditures. The conferees also observe that software research, that otherwise satisfies the requirements of section 41, which is undertaken to support the provision of a service, should not be deemed "internal use" solely because the business component involves the provision of a service.
In the waning days of the Clinton administration, regulations were issued that included an "internal use software test" and a more generally applicable "discovery test." These 2001 regulations were almost immediately jettisoned by the Bush administration. Regulations proposed in 2001 and finalized at the end of 2003 greatly relaxed the discovery test and "reserved" comment on internal use software. But in another seeming flip-flop, the IRS issued in early 2004 an advance notice of rulemaking on internal use software in which it asserted that until further notice, the tough version of the discovery test in the 2001 regulations would still apply.
Between 1996 and 2001, FedEx Corp. started and then abandoned development of an in-house computer system to manage its huge volume of transactions. The IRS denied $11.6 million of research credits contemporaneously claimed by FedEx on its tax returns.
In FedEx Corporation v. United States, 103 A.F.T.R. 2d 2009-2722 (W.D. Tenn. June 9, 2009), Doc 2009-16668, 2009 TNT 140-9, a federal district court in Tennessee ruled in favor of the taxpayer, whose software development expenses would not qualify under the stringent 2001 version of the discovery test. The court agreed with FedEx that the IRS could not mandate in a February 2004 notice use of the 2001 discovery test that was inconsistent with its December 2003 final regulations.
So the much-hated discovery test was put out to pasture, and as matters stand, the IRS has no internal use software regulations. KPMG said the ruling is "likely to have positive implications for research credits claimed in a number of industry sectors, including banking and other financial services, telecommunications, retail, and manufacturing." (See KPMG's "What's New in Tax," July 31, 2009.)
TG Missouri (November 12, 2009)
By statute, the cost of depreciable property cannot be included as qualified research. Presumably, the policy reasoning behind this rule, originally formulated at the credit's inception in 1981, was that depreciable property was already receiving enough tax incentive from the combination of accelerated depreciation and the investment tax credit. (In 1986 Congress repealed the investment credit but retained the exclusion of depreciable property from the definition of qualified research.)
TG Missouri is an auto parts manufacturer an hour's drive south of St. Louis. It develops and uses production molds to make plastic auto parts like air bags and steering wheels to customers' specifications. In the situations in question, TG Missouri retained physical possession of the molds, even though the customers for whom the molds were created retained legal ownership. (For the opinion in TG Missouri Corp. v. Commissioner, 133 T.C. No. 13 (Nov. 12, 2009), see Doc 2009-24993 or 2009 TNT 217-9 .)
The company claimed these expenses as qualified research expenses on its 1998 and 1999 tax returns, even though they were incurred to create depreciable property. In 2006 the IRS audited the return and denied the credit. Before the Tax Court, the company argued that the expenses were qualified because they were not depreciable in the hands of the taxpayer claiming the credit. The court agreed. According to Dean Zerbe, national director of Alliantgroup in Washington, "this decision may have a wide-ranging impact on all industries in determining supply costs that are eligible for the R&D tax credit." (See PlasticsNews.com, Dec. 8, 2009.)
Whatever the legal validity of the decision, there is no economic policy for making a distinction on ownership. The ruling opens the door to more tax-motivated mischief that only diverts private business from maximizing pretax profits. It obviously creates an incentive for ownership changes that are not necessarily the most advantageous from a normal business perspective. For no good reason, companies that retain title to molds are put at a competitive disadvantage to those that transfer ownership to customers.
Trinity (January 29, 2010)
In Trinity Industries Inc. v. United States, No. 3:06-cv-00726 (Jan. 29, 2010), Doc 2010-2781, 2010 TNT 25-16, the U.S. District Court in Dallas considered the case of a shipbuilder's expenses incurred for the design and construction of "first in class" ships. Expenses relating to six of these prototypes were examined. The court rejected the government's argument that in designing new ships, Trinity was simply integrating existing components. Although the court considered all six projects to contain elements of qualified research, it only considered two to be significantly innovative, allowing substantially all (80 percent, according to regulations) of Trinity's expenses to be classified as qualified research and therefore all project expenses to be qualified. The court believed it had to reject all the expenses of the other projects because it lacked sufficiently detailed information to isolate the qualified expenses. "Overall, the District Court's ruling is good news for taxpayers who have been wary of the IRS' heavy handed administration of the R&D credit," said Zerbe. (See "Another Taxpayer R&D Victory at the Courthouse!" Tax Alert, Alliantgroup, Feb. 4, 2010.)
As busy as the courts have been recently in addressing the definition of qualified research, their decisions address only some of the issues facing taxpayers and the IRS. It seems to be a never-ending story. In 1986 Congress narrowed the definition of qualified research because, as the Joint Committee on Taxation explained, "The Congress believed that the definition has been applied too broadly in practice, and some taxpayers have claimed the credit for virtually any expenses relating to product development." ("General Explanation of the Tax Reform Act of 1986," JCS-10-87, May 4, 1987). In the past, the IRS has tried to deny credits by industry -- for example, to defense contractors under fixed price contracts and for generic drug makers. (For a practitioner's view of the issues that arise in the face of "noticeable hostility" from the IRS about the research credit, see David L. Click, "Zeal and Activity in the Arena of the Research Tax Credit," Tax Notes, Dec. 15, 2008, p. 1305, Doc 2008-23915, or 2008 TNT 242-48.)
A recent report from the Government Accountability Office also discusses the problems of defining qualified research ("Tax Policy: The Research Tax Credit's Design and Administration Can Be Improved," GAO-10-136, Nov. 6, 2009, Doc 2009-26846, 2009 TNT 234-22 ; see especially Appendix IV). In addition to the issues with defining qualified research already described here, the GAO reported on the controversy between taxpayers and the IRS when determining research support and supervisory activities that qualify for the credit. The report also describes the difficulties in determining the point in time when commercial production commences (after which related research expenditures no longer qualify).
Figure 1. Annualized Estimated Revenue Cost of Research
Credit in Successive Extenders Legislation
Source: Underlying estimates are from the Joint Committee on Taxation. Adjustments and details are described in "The Permanently Temporary Retroactive Resarch Credit," Tax Notes, Jan. 18, 2010, p. 295, Doc 2010-716, or 2010 TNT 11-5.
Rising Revenue Effect
Until the recent recession, the revenue cost of the credit has grown far more rapidly than economic growth. This remarkable growth is shown in Figure 1. Based on these court decisions, we can expect that trend to continue.
Part 2. Economics: The Inefficiency of Inaccuracy
The research credit's reason to exist is to provide an economic incentive. Unfortunately, the practical difficulties in determining which expenditures should qualify for the credit undermine its effectiveness.
Let's take the simplest example. A company engages in two activities, A and B. Activity A is research providing spillover benefits and therefore has economic justification for a subsidy. B provides no spillover benefit, and there is no reason to subsidize it. Ideally, expenditure qualified for the research credit (QRE) should equal A. Unfortunately, vexing practical problems with the definition of QRE prevent this. The IRS can only draw a bright line that it hopes approximates the ideal. The bright line may be too restrictive or too generous, as illustrated in Figure 2.
To fully appreciate the problem, we must now introduce the economic concepts of "complementarity" and "substitutability." Perfect complements are things that are used in fixed proportions. An example would be a left shoe and a right shoe. Except in a few unfortunate circumstances, there is only value when they are paired. This has implications for taxes and subsidies. Subjecting right shoes to a $1 tax is the same as taxing the pair at $1.
At the other end of the spectrum are perfect substitutes. These would be items like a screwdriver with a red handle and a screwdriver with a green handle. For most people there is absolutely no difference. If the government levies a tax of $1 on red-handled screwdrivers, there would be a massive switch to the untaxed green.
Returning to the example in Figure 2, let's assume A and B are perfect complements. This limits (but as we shall see, does not eliminate) the economic problems that arise from an inaccurate definition of research.
In Case I, the bright-line definition unduly reduces the tax benefit from the credit. For the company, the tight definition is equivalent to a low rate of credit on all of activity A. (If the resulting effective rate is deemed too low, the problem is readily fixed by an upward adjustment of the statutory rate.) In effect, the legal definition of qualified research is serving as good proxy for the research we want to subsidize.
Figure 2. Bright Lines Used to
Define Qualified Research
Conversely, in Case II, in which there is a liberal definition of qualified expenses, tax benefits are undeservedly extended to B, but the effect is the same as an increase in the credit rate on A.
Unfortunately, the situation deteriorates if there is substitutability within and across categories A and B. In Case I with restrictive definition, the company will do more research that earns the credit and less research that doesn't, even though all of its research deserves a subsidy. This is a tax-induced distortion of research practices that reduces the productivity of that research.
In Case II with the generous definition, the company now properly gets a subsidy for all the right kinds of research, but in response to the credit, the company does more B activities that qualify for the credit and less of those B activities that do not. The company moves away from its most efficient production mix in order to maximize credits, and the government is subsidizing activity with no spillover benefits.
Interfirm, Interindustry Effects
This is just the beginning of the story. Even in the less problematic case of complete complementarity, inefficiency can still result. This occurs simply because companies do their research differently from each other. Figure 3 illustrates the point.
Figure 3. Unjustified Unequal Treatment of
Five Companies With
Different Approaches to Research
With the bright line drawn tightly (Case I), only firms 1, 3, and 4 are eligible for credits. Firm 2 is undeservedly shut out. Firm 5 is rightly excluded. From the point of view of competition, firms 1 and 3 are on equal footing. Firms 2, 4, and 5 are at a competitive disadvantage, but that is entirely justified. So what's the problem?
It is more intuitive if we think about each row as representing an industry. Industries 1 and 3 are being shortchanged given the social benefits they produce. To promote efficiency, we want to expand these industries. The economically correct subsidies would make the credit proportional to activity A. From the viewpoint of overall economic efficiency, industries 1 and 3 suffer (and their share of economy shrinks for the less efficient industries 2, 4, and 5).
With a generous bright-line definition (Case II), all industries doing good research (activity A) get credit for all that research. But now industries 1, 2, 4, and 5 get excess subsidies for no good reason. (Industry 5 gets a credit but does not conduct any good research.) Industries 1, 4, and 5 are given a competitive advantage over 2. And all are better off than industry 3, which just does good research. Industries 1, 4, and 5 expand at the expense of 2. All of these shifts in industry size are tax-induced distortions that reduce productivity and damage growth.
If we move to the more complicated (but realistic) case in which not only do firms research differently, but there is also substitutability, the inefficiencies are even larger. Firms are getting uneven subsidies that plow furrows through the free-market level playing field and are not justified by spillover benefits their research activities produce. On top of that, inside their walls, they are inefficiently disrupting the mix of their research and nonresearch activities.
Inefficiency of Claims on Amended Returns
Where economic incentives are concerned, justice and fairness must be carefully interpreted or incentive effects will be subverted. A prominent example of this subversion is the widespread use of research credit studies and, based on these studies, the filing of amended returns. As it explained in April 2007, the IRS has made research credit claims on amended returns a top audit priority:
The volume of R&E Credit claims that have been filed, along with the compliance audit resources required to examine these claims, has continued to rise to a level of high strategic importance to LMSB, as a Tier I issue. A growing number of these claims, both formal and informal, are based on marketed tax products supported by studies prepared by the major accounting and boutique firms. Typically these studies are marketed on a contingent fee basis. [ "Industry Director Directive #1 on Research & Experimentation (R&E) Credit Claims," LMSB-4-0307-025, April 4, 2007, Doc 2007-8754 , 2007 TNT 66-55.]
Amended returns claiming the research credit raise serious questions about the incentive effect. In most circumstances, it is sensible to allow taxpayers to amend their tax returns. The only concerns are the extra administrative cost and the problem of aging information. These practical problems are the reason for time constraints on amended returns. After all, taxpayers should not be denied deductions for, say, medical expenses or taxes paid, just because they were unaware or made some sort of mistake at the time of the original filing.
For tax subsidies intended to provide incentives, however, we have to look through a different lens. From an economic perspective, there is something intrinsically damning about filing a late claim. For an incentive to have a positive impact on research to which it applies, the taxpayer must know about the credit before or during the time that research is conducted. If there is a cost benefit analysis, the credit must be part of that analysis. Taxpayers who file claims years after the fact, after being informed by outside consultants that they might be eligible for the credit, are not provided an incentive. They are receiving a windfall. Their representatives have every right to seek this windfall, but as a matter of policy -- remember, the reason the credit exists is to provide an incentive -- credit claims on amended returns simply should not be granted.
Inefficiency From Uncertainties
Up to this point we have not addressed the issue of uncertainty. Investors prefer certainty. Any investment subsidy with uncertain benefits will have a diluted incentive effect. Businesses potentially eligible for the research credit face many different types of uncertainty.
First, there is the normal uncertainty faced by any taxpayer whose investment incentives do not come in the form of a refundable credit. If there is insufficient tax liability (think General Motors), the credit's incentive effect will be diminished (if the credit must be carried forward) or eliminated (if carryforwards expire).
But the research credit is plagued with other elements of uncertainty that other tax incentives are able to avoid or, at least, experience with less severity.
There is the uncertainty about whether the perpetually temporary credit will be extended. Although the credit has been unavailable for only one year in its 29 years of existence, tight budgets could increase worries. (For a timeline of the credit's history, see Tax Notes, Jan. 18, 2010, p. 295, Doc 2010-716, or 2010 TNT 11-5.) And then there is uncertainty not just about the existence of the credit, but also about exactly which form it will take. The credit expired again on December 31, 2009. Congress is currently searching for funds for the credit's 14th extension since 1981.
There is uncertainty that arises from the incremental nature of the credit. In general, research credits are only available for expenditures above a base period amount. It may be difficult for companies to do long-term forecasts of whether they will be above or below the base amount and, if so, by how much.
Finally, as emphasized earlier, there is uncertainty about exactly which expenditures may or may not qualify for the credit. This diminished incentive effect is in addition to inefficiency resulting from the inaccurate definition of research pictured in figures 1 and 2.
Part 3. Windfall
Let's suppose all the uncertainty and definitional problems discussed previously are magically eliminated. There is still one huge economic issue hanging over the credit that is not resolved. Economists do not know if, once the incentive is delivered, the taxpayer in fact responds to it. A few studies have been conducted, but as with most economic measurement, the results are inconclusive. And there is always the possibility that any measured increases in research are just paper changes -- whereby activities previously not identified as research are now labeled research solely for the purpose of obtaining the credit. As a general rule, we know that to be eligible for subsidies, businesses are much faster at adjusting paperwork than in making real changes in their activities. The former is attractive to business because it only requires work by accountants and lawyers, with no disruption of business practices.
When economic evidence is ambiguous, we must fall back on common sense and anecdotes to help make judgments. Is it unreasonable to believe that most companies plan their research not taking the credit into consideration and then look at any credits received as gravy? At his nomination hearing to become the 72nd Treasury secretary, Paul O'Neill, a former CEO at two Fortune 500 companies, told senators: "I never made an investment decision based on the tax code." And in response to further questioning, he elaborated:
Maybe I should say more directly to you, if you are giving money away, I will take it. If you want to give me inducements for something I am going to do anyway, I will take it. But good business people do not do things for inducements.
Part 4. Inherent Imprecision
An economic definition of qualified research might go something like this: business activity that creates spillover benefits beyond those provided to the performer of the activity in the form of knowledge that increases the productive capacity of the economy. There are many problems associated with translating this idea into an operational definition, but the critical one is identifying activities that create knowledge. We are all comfortable with the image of white-frocked scientists in laboratories conducting experiments to create new products. Even in this easiest of cases, issues arise -- for example, which support and managerial staff qualify? Still, these issues are relatively easy to deal with in traditional corporate structures of manufacturing companies, where designated research divisions are dedicated solely to product development.
The first problem is that traditional organizations are becoming increasingly scarce. Business management structures have become much more fluid over the last few decades. There is less hierarchy -- especially in cutting-edge, high-tech companies. Companies strive for continuous interaction between marketers and product developers and between line workers and process developers. In any good company, most employees are encouraged -- and sometimes even required -- to innovate. It is integral and indistinguishable from their more routine activities. But when the traditional indicators of research are not present -- there are no formally trained scientists, no lab coats, no clearly designated research groups -- matters get extremely dicey.
Add to this that manufacturing is a shrinking share of the economy. We make fewer things and provide more services. Devising innovative ways of providing services is an extremely important driver of U.S. productivity growth. But again we are in a situation with much innovation but none of the traditional indicators.
We can very easily get into the awkward position in which maybe all or maybe none of a company's activities could reasonably be considered research. For example, is a company that produces custom-made components for its business clients doing research? Another reasonable answer is that it is all a matter of degree. At best, administrators and judges can only make informed guesses based on legal criteria that are often only crude approximations of our economic definition.
These are hardly academic issues. Probably the most significant scientific development of the last half-century is the Internet. Certainly we would want any incentive for research to foster developments like that. But beyond the scientists who are involved in the development of personal computers, servers, fiber optic cable, and core software coding, who else should be considered conducting research? There are myriad applications. Are they nothing more than electronic versions of existing procedures, or are they new knowledge? Detailed facts-and-circumstances determinations are required to even begin making judgments.
Or let's take another not-so-obscure example: the development of consumer electronics like the iPod, iPhone, and now, the iPad. A case certainly can be made that much of the magic performed by Steve Jobs could easily be considered mere design (and therefore ineligible under the regulations). But streamlining and simplifying the use of computer and communications technology for billions of users is at the heart of maintaining our technological revolution. How can we draw a line through highly integrated, innovative companies like Apple and put qualified activities on one side and nonqualified activities on the other? Why would we want to?
Part 5. Tax Reform
The research credit, now 29 years old, and the investment tax credit, which in its most familiar form lived for 24 years -- from 1962 to 1986 -- share one important characteristic. Over time, taxpayers become more adept at expanding the definition of expenditures qualified for the credit. The general rule for the old investment credit was that it applied to equipment and not to structures. But the credit became an inducement to recharacterize investments that essentially were structures as equipment. Now the research credit provides an inducement to recharacterize nonresearch expenditures as research.
The revenue saved from repeal of the investment tax credit was the primary source of funds for reduction of the corporate tax rate from 46 percent to 34 percent in the revenue-neutral Tax Reform Act of 1986. The research credit is proportionately much smaller than the old investment credit, but it could be used to fund a reduction in the statutory corporate tax rate from 35 percent to 34 percent. The benefits of the research credit, as shown in this article, are questionable. Rate reduction -- the essence of tax reform -- is widely perceived as essential to U.S. competitiveness.
About Tax Analysts
Tax Analysts is an influential provider of tax news and analysis for the global community. Over 150,000 tax professionals in law and accounting firms, corporations, and government agencies rely on Tax Analysts' federal, state, and international content daily. Key products include Tax Notes, Tax Notes Today, State Tax Notes, State Tax Today, Tax Notes International, and Worldwide Tax Daily. Founded in 1970 as a nonprofit organization, Tax Analysts has the industry's largest tax-dedicated correspondent staff, with more than 250 domestic and international correspondents. For more information, visit our home page.
For reprint permission or other information, contact firstname.lastname@example.org