Individuals Oppose R&D Non-allocation in FTC Regs

May 1, 2020

The Honorable David J. Kautter
Assistant Secretary (Tax Policy)
Department of the Treasury
1500 Pennsylvania Avenue, N.W.
Washington, DC 20220

The Honorable Charles P. Rettig
Commissioner
Internal Revenue Service
1111 Constitution Avenue, NW
Washington, DC 20044

The Honorable Michael Desmond
Chief Counsel
Internal Revenue Service
1111 Constitution Avenue, NW
Washington, DC 20044

Subject: Comment on Proposed Regulations Relating to the Allocation and Apportionment of Deductions for R&E Expenditures (REG-105495-19)¹

Dear Messrs. Kautter, Rettig, and Desmond:

This letter is in response to the Notice of Proposed Rulemaking under Sections 861 and 904 published in the Federal Register on December 17, 2019 (the “Proposed Regulation”).² The Proposed Regulation would not allocate deductions for research and experimental (R&D) expenditures under Section 174 (“R&D deductions”) to foreign source gross income inclusions under Section 951 (“Subpart F inclusion”) or Section 951A (“GILTI”)³ from a controlled foreign corporation (“CFC”) for the purpose of the foreign tax credit limitation.⁴

The Proposed Regulation's proposed exclusion of Subpart F and GILTI inclusions from direct allocation of R&D is a radical change that is inconsistent with the structure of the statute, four decades of consistent regulatory interpretations of Section 862 (unchanged by the TCJA), and all prior legislative interventions in the allocation of R&D. The Proposed Regulation's preamble justification for this rule lacks foundation and does not provide a reasoned justification for this interpretation. Even if the Proposed Regulation's interpretation could be sustained as a matter of administrative law, which we submit it cannot, this allocation rule is unsound tax policy. The non-allocation of R&D to Subpart F inclusions and GILTI is misguided and should not be adopted.

I. SUMMARY

1. This letter provides a background description of the important role played by the allocation of deductions to foreign source income in applying the foreign tax credit limitation. This background provides context for the succeeding comments, each of which is developed in subsequent parts of this letter. (Part II)

2. The non-allocation of R&D to Subpart F inclusions and GILTI is a radical change from prior interpretation of Section 862(b) as applied to Section 904 limitations applicable to Subpart F income. This interpretation is inconsistent with the statutory structure of Code Sections 904 and 862(b) and the history of interaction between the regulations allocating R&D expense and prior temporary legislative action in section 864(g) (no longer in effect) with respect to those regulations. The regulatory change in longstanding (regulatory) interpretation is unsupported by any indication of legislative intent in the TCJA. (Part III)

3. The justifications in the preamble to the Proposed Regulations (the “Preamble”) for not allocating R&D deductions to Subpart F inclusions or GILTI. These are based on two premises: (i) that “[s]uccessful R&E expenditures ultimately result in the creation of intangible property that will be used to generate income,” and (ii) that such intangible property is fully compensated under section 482 principles without reference to foreign subsidiary earnings included under Subpart F and GILTI. Each of these assertions is demonstrably wrong as a generalization.

4. The proposed treatment of R&D deductions will require the U.S. treasury to subsidize additional foreign taxes paid by a U.S. multinational taxpayer to non-U.S. jurisdictions. There is no evidence this was the Congressional intent in the TCJA and it is not justified as a cost-effective subsidy for domestic R&D. (Part V)

5. The proposed exclusion of Subpart F inclusions and GILTI from allocation should not be adopted. Instead, R&D deductions should be allocated across all categories of income within groupings of 3-digit SIC product categories. The remaining proposed changes to allocation and apportionment of R&D in Proposed Regulation §1.861-17 should be retained. (Part VI)

II. BACKGROUND ON THE FOREIGN TAX CREDIT LIMITATION AND THE GILTI AND GENERAL LIMITATION CATEGORIES

A. Post-TCJA Foreign Tax Credit Limitation Rules for a Domestic Corporate Shareholder in a Controlled Foreign Corporation (“CFC”)

The TCJA adopted a hybrid system of exemption and current taxation, at a reduced rate for operating income, for taxing a domestic corporation on earnings of a CFC.⁵ This section describes the importance of allocating deductions to the operation of the foreign tax credit limitation generally and as applied to the separate foreign tax credit limitation categories for Subpart F inclusions⁶ and GILTI.⁷

The foreign tax credit limitation for a limitation category is determined by multiplying the U.S. taxpayer's total U.S. taxes (before reduction by the foreign tax credit) by a fraction.⁸ The numerator is foreign source taxable income in the relevant category and the denominator is the taxpayer's worldwide taxable income.⁹

The foreign tax credit limitation is applied separately to categories of foreign source income, including passive category income,¹⁰ GILTI (that is not passive category income),¹¹ and general category income (including Subpart F income that is not passive category income).¹² The numerator and denominator of the foreign tax credit limitation fraction each are taxable income (not gross income) amounts and therefore must be reduced by deductions allocable to each income category.¹³

The purpose of the Section 904 limitation on the foreign tax credit is to prevent foreign taxes from being allowed as a credit against U.S. tax on U.S. source taxable income. To calculate the limitation, the fraction is multiplied by the total pre-foreign tax credit U.S. tax that would be owed under U.S. law. Allocating deductions away from foreign source income increases the limitation. The determination of income source and the allocation of deductions to arrive at foreign source taxable income in the numerator of the fraction relies on U.S. tax law for definition. The denominator of the fraction is total U.S. taxable income (from all sources) under U.S. tax law. The Code's foreign tax credit regime only looks to foreign law to ascertain which person is paying the foreign tax and for the amount of the “income, war profits, and excess profits taxes paid or accrued” to the foreign government that is potentially creditable.¹⁴

Excess foreign taxes in the general limitation category may be carried back one year and carried over ten years.¹⁵ Excess foreign taxes in the GILTI limitation are not allowed to carry back or over to other years and are lost as credits.¹⁶ The absence of carryovers of credits in the GILTI limitation category (after the statutory 20 percent haircut on creditable taxes attributable to GILTI)¹⁷ is the source of much taxpayer angst over excess credits in the GILTI limitation category. This outcome surely was intended (and almost equally as surely, though less directly relevant for statutory analysis, was factored into the revenue estimate provided to Congress). There is no indication in the legislative history to GILTI that Congress expected a radical revision to the rules for allocating R&D expense to foreign tax credit limitation categories.¹⁸

B. Allocation of Deductions for the Numerator of the Foreign Tax Credit Fraction

The foreign tax credit limitation (for each relevant category) restricts the allowance of the credit for foreign taxes to the amount of U.S. tax on the foreign source taxable income in that category as measured under U.S. tax principles. If R&D expenses are under-allocated (or not allocated) to foreign income for determining the foreign source income in the numerator of the foreign tax credit limitation fraction, then the numerator will be inflated and it will appear that a larger share of total U.S. tax is paid on the that category of income than should be attributed to that income. This would allow more foreign taxes to be credited, so U.S. tax is reduced by a credit for the foreign tax. In other words, it would inappropriately subsidize the taxpayer's foreign taxes.¹⁹

Section 862(b), unchanged by the TCJA, defines taxable income from sources without the United States (the term used for the numerator of the foreign tax credit limitation) as follows:

§ 862. Income from sources without the United States

. . .

(b) Taxable income from sources without United States. From the items of gross income specified in subsection (a) there shall be deducted the expenses, losses, and other deductions properly apportioned or allocated thereto, and a ratable part of any expenses, losses, or other deductions which cannot definitely be allocated to some item or class of gross income. The remainder, if any, shall be treated in full as taxable income from sources without the United States.

The numerator of the foreign tax credit limitation fraction accordingly requires reduction of foreign source gross income by allocable deductions, including any deductions of the United States shareholder “properly apportioned or allocated thereto, and a ratable part of any expenses, losses, or other deductions which cannot definitely be allocated to some item or class of gross income.” There is no question that Subpart F income and GILTI inclusions are foreign source gross income. The only question for this comment is whether R&D is “properly apportioned or allocated” to Subpart F income or GILTI from a CFC for purposes of the foreign tax credit limitation.

C. R&D and the Expense Allocation Regulations

The expense allocation regulations for R&D provide special rules for expenses deductible under Section 174.²⁰ Amounts deductible under Section 174 generally include costs incurred in the experimental or laboratory sense related to the development or improvement of a product, including costs intended to discover information that would eliminate uncertainty concerning the development or improvement of a product.

Uncertainty is considered to exist when information available is not sufficient to ascertain the capability or method for developing, improving, and/or appropriately designing the product.²¹ After the elimination of uncertainty regarding the development or improvement of a product, costs incurred are not deductible under Section 174 (and are subject to capitalization).²² R&D expenses deductible under Section 174, however, are not subject to capitalization under Sections 263 or 263A.²³

R&D expense incurred when there is uncertainty as to whether there will be a successful outcome from the research leads to the longstanding approach of the allocation regulations to allocate R&D to broad categories of income. As stated in the current regulations:

“The methods of allocation and apportionment of research and experimental expenditures set forth in this section recognize that [R&D] is an inherently speculative activity, that findings may contribute unexpected benefits, and that the gross income derived from successful [R&D] must bear the cost of unsuccessful [R&D]. Expenditures for [R&D] that a taxpayer deducts under section 174 ordinarily shall be considered deductions that are definitely related to all income reasonably connected with the relevant broad product category (or categories) of the taxpayer and therefore allocable to all items of gross income as a class (including income from sales, royalties, and dividends) related to such product category (or categories).”²⁴

From expense allocation regulations first adopted in 1977 through the current regulations, U.S. shareholder R&D has been allocated to income from the CFC for purposes of determining the numerator of the foreign tax credit limitation. The first sentence of the above-quoted regulation explains why: because of the nature of R&D.

The R&D expenditure in question (i) requires uncertainty when the R&D is performed so it is difficult to associate the expenditure with specific income in the year incurred, (ii) includes unsuccessful as well as successful R&D, and (iii) when successful, may contribute to unexpected categories of income. In this circumstance, it is not possible with a meaningful confidence level to identify the income that benefits from the R&D, that is, the income to which the R&D should be allocated. For these reasons, the regulations have required allocation of R&D deductions to broad categories of income and apportionment of the deductions among all income, domestic and foreign, including sales, royalties and dividends, within that category. The use of sales as a basis for apportionment always has included the sales of a CFC on a look-through basis.²⁵

D. The Proposed Regulation's R&D Allocation Rule

Contrary to four decades of prior regulatory interpretation of the same statutory language, the Proposed Regulations would not allocate any shareholder level R&D deductions to Subpart F inclusions and GILTI for the purpose of determining foreign source taxable income in the numerator of the foreign tax credit limitation.²⁶ For the reasons set out in this letter, this is not a reasoned interpretation of the statute.

The Proposed Regulations accomplish this non-allocation of R&D deductions to a relevant category of foreign income by limiting R&D expense allocation to “gross intangible income,” which is “all gross income earned by a taxpayer that is attributable (in whole or in part) to intangible property . . .”.²⁷ Gross intangible income “includes gross income from sales or leases of products or services derived (in whole or in part) from intangible property, income from sales of intangible property, income from platform contribution transactions described in Regulation §1.482-7(b)(1)(ii), royalty income from the licensing of intangible property, and amounts taken into account under Section 367(d) by reason of a transfer of intangible property. Gross intangible income also includes a distributive share of any amounts described in the previous sentence, but does not include dividends or any amounts included in income under sections 951, 951A, or 1293.” ²⁸ (Emphasis added.)

Intangible property is defined broadly in Section 367(d)(4) and includes intangible property not usually associated with Section 174 R&D deductions.²⁹ The categories of intangible property most often associated with R&D expense are patents, copyrights and trade secrets, but R&D expenditures are not easily categorized or identified with a particular asset.

Relying on comments that “successful R&E expenditures ultimately result in the creation of intangible property that will be used to generate income,” the Preamble states that the Treasury and IRS agree with the comments and propose modifying the regulation to reflect that:

“R&E expenditures that are deductible under section 174 generally give rise to intangible property, and that under the rules in sections 367(d) and 482, the person incurring such R&E expenditures must be compensated properly when such intangible property gives rise to income.”³⁰

The Preamble goes on to state that:

“Gross intangible income . . . does not include dividends or any amounts included under section 951, 951A, or 1293. See proposed §1.861-17(b)(2). As a result, when applying §1.861-17 to section 904 as the operative section, because a U.S. taxpayer's gross intangible income, as defined in the proposed regulations, does not include income assigned to the section 951A category, none of its R&E expenditures are allocated or apportioned to the section 951A category.”³¹

By using the definitional device of “gross intangible income” to exclude “dividends or any amounts included under section 951, 951A, or 1293,” the Proposed Regulation allows U.S. multinationals' R&D deductions that otherwise would reduce foreign source taxable income instead to be shifted to other income for purposes of calculating the foreign tax credit limitation.

As a consequence of non-allocation to Subpart F inclusions and GILTI, foreign source net income will be higher in at least one category where the foreign tax credit limitation often binds, so more foreign tax credits will be allowed. This effectively refunds the foreign tax to the taxpayer thereby lowering the tax cost of foreign investment. The United States will sacrifice its tax revenue to subsidize the corporate tax collections of other jurisdictions and encourage foreign over U.S. investment.

III. THE PROPOSED REGULATION'S EXCLUSION OF SUBPART F INCOME AND GILTI FROM R&D ALLOCATION IS AN UNPRECEDENTED INTERPRETATION OF THE STATUTE

A. Meaning of the Text

The numerator of the foreign tax credit limitation fraction in Section 904 is “taxable income from sources without the United States.” The statute enacted by Congress in Section 862(b) defines the term to mean foreign source gross income reduced by “the expenses, losses, and other deductions properly apportioned or allocated thereto, and a ratable part of any expenses, losses, or other deductions which cannot definitely be allocated to some item or class of gross income.”

R&D deductions have been allocated to foreign income, including Subpart F inclusions, since regulations issued in 1977, except during a legislatively mandated moratorium in 1981 that was extended until the 1986 Act.³² When Congress subsequently legislated temporary R&D allocation rules in 1989, it prescribed allocation of R&D to foreign source income.³³ These rules ceased to have effect after several years, yielding to the regulatory regime of regulations promulgated in 1996.³⁴

The longstanding approach of the regulations applying Section 862(b) to R&D has been to recognize that “research and experimentation is an inherently speculative activity, that findings may contribute unexpected benefits, and that the gross income derived from successful research and experimentation must bear the cost of unsuccessful research and experimentation” and that R&D deductions “ordinarily shall be considered deductions that are definitely related to all income reasonably connected with the relevant broad product category (or categories) of the taxpayer and therefore allocable to all items of gross income as a class (including income from sales, royalties, and dividends) related to such product category (or categories).”³⁵ (Emphasis added.)

The Proposed Regulation would change the R&D expense allocation rule of the current regulations to exclude any possible allocation to income of a CFC received by a United States shareholder as a dividend, Subpart F inclusion, or GILTI. This exclusion is unsupported by the language of the statute or its history. The Proposed Regulation's preamble does not point to any indication of congressional intent in the legislative history of the TCJA to alter longstanding R&D deduction allocation rules that for decades have included allocation to income from CFC earnings repatriated as dividends and Subpart F inclusions.

There is no statutory category of “gross intangible income” as defined and used in the Proposed Regulation,³⁶ but it is clear that “gross intangible income” changes the calculation of foreign source taxable income. There is non-statutory interpretive support for altering the application of Section 862(b) to no longer allocate R&D deductions to foreign source income from CFC earnings. The next section discusses legislative intent and Part IV discusses the Preamble's rationales for its interpretation.

B. Legislative Intent

The Preamble does not cite (and we are unaware of) any legislative history that would justify the Proposed Regulation's approach to the allocation of R&D to Subpart F inclusions and GILTI. The Proposed Regulation nonetheless denies the allocation of R&D to Subpart F inclusions, even though there was no change to the statute affecting Subpart F in relation to the foreign tax credit that would serve as the justification for such a dramatic change in allocation. Not only is there nothing in the statute that anticipates such a change in R&D allocation, but the TCJA's corporate rate reduction increases the likelihood of excess credits for foreign taxes and as a result increases the concern that the United States will lose revenue if R&D deductions are under allocated to foreign source income and firms are allowed additional foreign tax credits.

The only statutorily authorized moratorium on allocation of R&D expense to foreign source income was adopted legislatively in ERTA 1981. That legislative action was temporary to allow and indeed require Treasury to study the issues. Once the full moratorium ended, subsequent temporary legislative adjustments to substantive R&D allocation rules clearly showed an intent to allocate R&D deductions to foreign source income without carve-outs of the nature and breadth in the Proposed Regulation.³⁷

The 1995 final R&D allocation regulations³⁸ have been in place since with only modest changes while there have been repeated legislative changes to the international tax rules of the Code. While legislative re-enactment sometimes has dubious relevance,³⁹ here Congress focused intently on R&D allocation issues over a decade in different administrations prior to the 1995 regulations but since then has not returned to these issues. If Congress had sought to shift R&D deductions away from GILTI and erode the U.S. tax base as the Proposed Regulation would do, it would be reasonable to expect a statutory change.⁴⁰

In a different context involving the DISC and under a pre-Mayo interpretative standard, nonetheless, the U.S. Supreme Court in a 2003 7-2 decision upheld the application of the predecessor to the current R&D allocation regulation to determine DISC combined taxable income.⁴¹ This is not to conclude that the agency here could not change its prior interpretation, but it is to suggest that the agency should exercise an enhanced level of care in doing so where the broad allocation theory of the regulation has satisfied prior Supreme Court scrutiny, there has been prior legislative attention to the very rule in question and multiple subsequent legislative changes to the operative rules of the foreign tax credit limitation including in the TCJA without making any legislative change to the R&D expense allocation rules.⁴²

In response to congressional requests, Treasury has performed two economic analyses of R&D allocation issues, in 1983 and 1995, and neither one recommended zero allocation of R&D deductions to foreign dividends and Subpart F income. There is no indication in the Preamble that any empirical analysis of the effect of the allocation rule issue has been undertaken by professional economists in Treasury's Office of Tax Analysis as has been the case previously. There also is no analysis of the revenue loss from the proposal or how it would affect the Congressional expectations of revenue from the affected TCJA international provisions.⁴³

Other than to benefit the increased number of taxpayers who may have excess foreign tax credits by reason of the reduced effective rates of U.S. corporate tax generally and on GILTI in particular, the only justification for this change in approach, which strains the apparent intent of Section 862(b), is found in Preamble justifications discussed in the next Part IV.

IV. PREAMBLE RATIONALES FOR EXCLUDING CFC INCLUSIONS FROM R&D ALLOCATION ARE NOT CREDIBLE

A. The Proposed Regulation's Preamble Rationales for Not Allocating R&D to Subpart F and GILTI Inclusions

The Proposed Regulation provides its first factual premise in the regulatory text:

“In addition, the method set forth in this section recognizes that successful R&E expenditures ultimately result in the creation of intangible property that will be used to generate income.⁴⁴ (Emphasis added.)

It is learned from the Preamble that, based solely on comments, the exclusion of CFC income from R&D allocation is based on a second factual premise:

“[T]o to the extent a GILTI inclusion is attributable to a CFC's income derived from intangible property developed by the worldwide group, the comments stated that the intangible property must have either been developed by the CFC or a CFC affiliate (in which case the R&E expenditures were not borne by the U.S. taxpayer), or licensed or acquired by the CFC from a U.S. affiliate, which would require that the U.S. affiliate take into account an arm's length royalty, gain on transfer, or a deemed income amount under section 367(d) to which its R&E expenditures should be allocated.”⁴⁵ (Emphasis added.)

In other words, the second premise is that intangible property held by a U.S. taxpayer cannot be used by or rights transferred to a CFC in which that taxpayer is a U.S. shareholder without the use or transfer being compensated by reason of the authorization (but not requirement) for the Government to assess an arm's length consideration under Section 482 or by reason of the requirement of Section 367(d) that a transfer of all substantial rights in a Section 351 or Section 361 exchange be treated as in exchange for a deemed royalty commensurate with the income attributable to the intangible.

The second premise, of complete compensation for intangible property, only is correct if the first premise is understood to mean something possibly different from what it actually says. In order to meet that condition, instead of saying “to generate income,” the first premise must be understood to say “generate income attributable to the intangible property” [sufficient to compensate all R&D].⁴⁶ The italicized language is all that is necessary for this letter's discussion, though the bracketed words are implied as well.

The first and second premises justifying the Proposed Regulation's failure to allocate R&D to Subpart F inclusions and GILTI do not hold up under scrutiny.

• To the extent that R&D fails and does not result in intangible property, the first premise is wrong. As discussed in the next Subpart B, a material portion of R&D fails and does not give rise to compensable intangible property.

• If R&D is successful and does give rise to intangible property held by the U.S. shareholder, and part of the economic return that funds the R&D is from earnings of the CFC without being paid in compensation for the intangible, the second premise is wrong. As discussed in Subpart C of this Part IV, billions of dollars of R&D expenditures are made in relation to intangible property that is licensed royalty-free, including to CFCs. This demonstrates that the second premise is wrong and, together with the fact of unsuccessful R&D, fatally undermines the rationale for the Proposed Regulation.

• The Preamble's second premise is that Section 482 is an accounting rule, instead of a grant of authority governed by an anti-abuse standard. It assumes as a result that a properly filed tax return will include “gross intangible income” that would be allocated under Section 482. This is a fallacy that extends the circumstances in which failure to allocate R&D to Subpart F and GILTI inclusions yields for foreign tax credit limitation purposes an overstated measure of taxable income from outside the United States. This Section 482 fallacy is discussed in Subpart D of this Part IV.

B. Unsuccessful R&D Does Not Result in Intangible Property

The first premise does not take account of the reality that much Section 174 R&D expense is unsuccessful and does not lead to income-generating intangible property. A third party generally does not pay for failed R&D.⁴⁷ Moreover, much R&D is not successful, so the real first premise (that R&D ultimately results in an intangible that generates income) does not apply to a broad range of cases. As one example reported in Genetic Engineering & Biotechnology News:

“In a February 2018 study in the journal Biostatistics, [MIT Professor] Andrew W. Lo . . . and colleagues, last year estimated aggregate clinical trial success rates and durations by indication, using a sample of 406,038 entries of clinical trial data for more than 21,143 compounds from January 1, 2000 to October 31, 2015. The results were sobering: The highest three success rates were 32.6% for clinical studies of ophthalmology drug candidates, 25.5% for cardiovascular drug candidates, and 25.2% for infectious disease products. The lowest percentage came from oncology trials, at just 3.4%.”⁴⁸

Other evidence may be found in a 2017 FDA study of failures of Phase 3 clinical trials following successful Phase 2 trials, where the FDA Report states: “Roughly 9 in 10 drugs/biologics that are tested in humans are never submitted to FDA for approval.”⁴⁹ And further:

“To better understand the nature of the evidence obtained from many phase 2 trials and the contributions of phase 3 trials, we identified, based on publicly available information, 22 case studies of drugs, vaccines and medical devices since 1999 in which promising phase 2 clinical trial results were not confirmed in phase 3 clinical testing. Phase 3 studies did not confirm phase 2 findings of effectiveness in 14 cases, safety in 1 case, and both safety and effectiveness in 7 cases. These unexpected results could occur even when the phase 2 study was relatively large and even when the phase 2 trials assessed clinical outcomes. In two cases, the phase 3 studies showed that the experimental product increased the frequency of the problem it was intended to prevent.”⁵⁰

These examples (found with a simple Google search) obviously are anecdotal evidence, but even a layperson is aware that R&D is unsuccessful as well as successful. The Section 174 deduction does not discriminate between the successful and unsuccessful R&D and the deductions for both are required to be allocated between U.S. and foreign source income.

The Preamble posits that these unsuccessful R&D amounts are fully funded by returns to intangible property alone, and not from any other source of return including profits from a CFC in the form of a Subpart F inclusion or GILTI. This is implausible as an economic matter and the Preamble provides no factual foundation for this premise other than the assertion of interested commenters.

In light of the presence, indeed the prevalence in some industries, of unsuccessful R&D, the argument that all R&D would be compensated by returns solely to intangible property is not credible, because arm's length counterparties cannot be assumed to pay for all unsuccessful R&D when they pay for intangible property rights. To fund R&D most businesses will have to rely on margins from their successful products and services as well as returns to intangible property.

The fact of unsuccessful Section 174 R&D expenditures is one reason the first premise is unreliable as a basis for the interpretation in the Proposed Regulation.

C. Successful Does R&D Does Not Always Result in Intangible Property That Is Separately Compensated

The Proposed Regulation Fails to Account for Business Practice in Relation to IP Business Strategies: Business Practice in Relation to Protection of IP Rights

An important rationale for intellectual property legal protection is that knowledge developed from R&D is nonrival?] in consumption, that is, one firm's use of the good does not prevent another from using the same good unless there are market restrictions on use of the knowledge including most usually legal protections.⁵¹ In other words, it is not enough that R&D result in intangible property, such as a copyright, design or process, if that property can be used by another person for free because there is no restriction on its use. In order for the intangible property to generate income in a competitive market, the owner must take the steps necessary to cause it to be legally protected in each market from which the intangible property will be used to generate income.

The first premise, that successful R&D ultimately results in the creation of intangible property “that will be used to generate income,” does not take account of the circumstances in which intangible property from R&D is not separately compensated, including business practice in relation to acquiring and protecting intellectual property rights (“IP”) for intangible property developed by R&D. In practice, IP legal protections for intangible property in any given case can vary from nonexistent to weak to strong.

Businesses calibrate their investment in protecting IP based on a market-by-market cost-benefit analysis. Companies do not seek patent protection for their inventions in every market. They expend only what they consider necessary to protect rights that would be worth asserting or that are necessary for defensive purposes to resist claims of infringement.⁵² Some categories of property may be protected by multiple forms of IP legal protections (e.g., software potentially may be covered by a patent, a copyright and/or a trademark).⁵³ The choice of whether to seek legal protection, which form of protection to seek, and whether to prosecute infringement involves at each step a mix of business and legal considerations and cost-benefit analysis. Each such decision affects whether and the extent to which the intangible property contributes to income generation.

Even when a taxpayer takes appropriate steps to protect IP, the mere claim that a taxpayer possesses an intellectual property right, including a registered patent or trademark, does not establish that the right is legally valid. A recent analysis found that of 2,534 patents that had been fully reviewed by the Patent Trial and Appeal Board, 2,138 or 84% were determined to be “Unpatentable/Cancelled.”⁵⁴ Even where an intellectual property right is valid, there are a series of reasons it may not be enforced.⁵⁵

The Proposed Regulation treats intangible property as though it were a category of asset that taken separately reflects the full value of the underlying R&D. In order to hold value, an intangible property right, which is a creature of law (and in many cases the law or the facts necessary to satisfy the law are uncertain), must be legally protectible. Whether intended or not, a great deal of R&D is performed to generate knowledge that is not legally protected or that would not be if challenged. The Proposed Regulation's premise that R&D is performed to create intangible property income assumes without evidence that intangible property “ultimately” developed from the R&D is legally protected in every jurisdiction in which the CFC earns income. This is without any foundation of evidence in the Proposed Regulation and contrary to routine IP business practices.

To frame the broader point in a way that would seem to be intuitively obvious, firms seek to maximize profits and generally do not care about the legal categorization of the income they earn. Firms do not engage in R&D solely on the basis that it will result in a legally enforceable intellectual property right that taken alone will earn a return to pay for the R&D (though this approach may be followed by some businesses, such as those engaged exclusively in R&D development). A rational business engages in R&D in the expectation that the resulting knowledge will contribute in some manner to overall return in whatever form that return may take (including through the earnings of a foreign subsidiary). That is the premise of all of the R&D allocation regulations since 1977.

For this second reason, the first premise is unreliable.

The Proposed Regulation Fails to Account for Open Source and Similar Intellectual Property Business Strategies

The first premise also fails to take account of the business reality that an increasingly important subset of modern intellectual property business strategies involves licensing the fruits of the R&D on a royalty-free basis and earning returns through the provision of goods and services. Businesses employing open access IP business models, such as using open source software platforms, are examples. For example, Alphabet Inc.'s Google has invested substantial R&D in developing its Android smartphone operating system,⁵⁶ which is open-source and thus available for free to other businesses (and to Google's own subsidiaries). Google instead generates substantial profits from Android through mobile advertising sales, as well as from its “app” store, Google Play.⁵⁷

Modern IP business strategy includes a spectrum of possible business uses of IP rights. As stated succinctly by John Palfrey in 2012:

“This spectrum starts with full exclusion on one end. That's the sword and shield mode. In the middle is limited exclusion, where you let some people use your intellectual property for some purposes. That's the licensing approach. And then there's open access on the other end of the spectrum, where you permit unlimited use by others.”⁵⁸

Because the case of interest here is where IP is made available by a U.S. shareholder to its controlled foreign corporation, it is unnecessary to consider the strategy of full exclusion.⁵⁹ This leaves the licensing or open access approaches. The first premise of the Proposed Regulation, that all intercompany transfers of intellectual property rights are compensated, conveniently disregards the open access strategy, notwithstanding that use and profitability of that strategy has grown dramatically since Palfrey's primer was published.⁶⁰

The objective of investing in open source software is to expand the user base, to whom additional goods and services can be sold. Making software open-source assures customers and third-party developers that the software will remain available even if one company goes out of business or is forced to end relationships by trade sanctions. Moreover, making software open-source ensures that other developers will contribute improvements to the software. For-profit companies that use an open access strategy by definition are not eleemosynary institutions; they earn their returns other than from directly monetizing intangible property from the R&D performed to enhance open source software.

As another example, consider Red Hat, Inc. (“Red Hat”), acquired by IBM in 2019 at a $32 billion enterprise valuation. Red Hat provides open source software solutions to domestic and international customers.⁶¹ In other words, the software is not subject to a license fee and Red Hat's 10-K indicates that indeed it does not charge a royalty for its patents.⁶² Instead, Red Hat provides products and services compensated by subscription and services revenue.⁶³ In FY 2019, Red Hat reported R&D expense on its income statement equal to approximately 20% of revenue. In other words, Red Hat conducts material amounts of R&D that for this discussion may be assumed to be at least in part for the purpose of creating intangible property (patents and potentially copyrightable software), but under its “Patent Promise” and open-source licensing Red Hat will not charge third parties for this intangible property.

Under this kind of business model, the income earned to support R&D is from providing services to customers, including through foreign subsidiaries. Yet, the income from foreign subsidiaries that is received as Subpart F or GILTI inclusions would be excluded from R&D allocation under the Proposed Regulation.

The implication of an open access business model may be illustrated by modifying the facts of Example 1 in Proposed Regulation §1.861-17(g) to be an open access IP services business model.

The result in Example A clearly mis-allocates R&D exclusively to U.S. source income. This Example A illustrates that the assumption R&D generates an intangible that will be compensated will not be not correct in a range of important cases.

The increasing use of open access IP business strategies is a third reason the Preamble's first premise is not reliable as a basis for the interpretation in the Proposed Regulation.

D. Sections 482 and 367(D) Cannot Be Relied On To Assure That Intangible Property Rights Transferred To a Related Person Are Fully Compensated

1. The Section 482 Fallacy

The reliance on section 482 as a basis to not allocate deductions to CFC income is misplaced. Even where the U.S. shareholder possesses intangible property that generates income, the second premise, that Section 482 will assure an arm's length return to the intangible, is unreliable for reasons relating to the structure and role of Section 482. First, Section 482 is a discretionary enforcement tool for transactions involving persons under common control and not a rule of accounting prescribing a charge for income from a particular asset or class of assets. The objective of the enforcement tool is to ensure clear reflection of income in common control transactions. The policy rationale for Section 482 is to make neutral the decision whether to transact with a related or an unrelated party.⁶⁶

The discretionary enforcement relies on a counterfactual standard: what would unrelated persons have done in the same transaction in similar circumstances. Section 482 does not prescribe a single price for a good or service but seeks equivalence between related and unrelated party transactions by employing a standard that has broad flexibility to encompass different transactions in different industries. The utility of the standard relates to its flexibility across differing factual circumstances, but the confidence level in any result under the standard differs dramatically depending on the nature of the property or service, the circumstances of the transaction and the available data.

This comment will describe three of the material errors that result from the Proposed Regulation's misplaced reliance on Section 482. First, there are circumstances in which R&D will be performed that benefits activity in the same product area conducted by an affiliate, but it is not realistic to expect Section 482 enforcement to impose a compensatory charge. Second, Section 482's clear reflection of income does not distinguish categories of income, including whether income is “gross intangible income,” and if it did it would dramatically expand the number of Section 482 cases. Third, under Section 482 a charge may be required, but the permissible charge will not necessarily capture the full amount of income benefitted by the R&D and there most often is uncertainty and variability as to what constitutes a permissible charge. We take these up in turn.

Limits of Section 482 enforcement. The second premise treats Section 482 as though it were a positive law accounting rule that requires a specific amount of income, based on an arm's length price, to be included on a tax return in relation to any transfer of intangible property or rights under intangible property to a CFC. That is not what Section 482 does nor is it what is intended. Section 482 is a grant of enforcement authority governed by a standard.⁶⁷ The words of the statute are instructive:

“[T]he Secretary may distribute, apportion, or allocate gross income, deductions, credits, or allowances between or among such organizations, trades, or businesses, if he determines that such distribution, apportionment, or allocation is necessary in order to prevent evasion of taxes or clearly to reflect the income of any of such organizations, trades, or businesses.”⁶⁸

It is well understood that Section 482 is an enforcement tool for the government and may not be asserted by the taxpayer.⁶⁹ Moreover, the application of the rule requires that it be asserted and the government's resources are limited.

Consider the example of U.S. R&D performed by a U.S. shareholder of a CFC that contributes to a line of business also conducted by the CFC, but does not result in an independently transferrable intangible.⁷⁰ Under current law residual business intangibles are supposed to be compensated when their value is transferred irrespective of the formality of a legal transfer. That is easily seen in the transfer of assets constituting a business, but is much harder to discern in a context where the residual intangible is adjacent to an independently transferrable intangible licensed to the CFC.⁷¹ There is no meaningful experience applying the concepts of the now expired 2015 Temporary Regulations and the new TCJA rules in this context. It would be difficult, even if appropriate under current law, to charge for that R&D separately or increase an existing royalty rate for such R&D. If the R&D enhances the value of the firm's line of business, there is no reason to believe that the business conducted through a CFC is less benefitted than the same business conducted by the U.S. taxpayer.

As another example, Section 482 will not effectively police the boundary of a return to an intangible owned by a U.S. shareholder if its CFC uses the intangible to earn Subpart F income. Where a taxpayer makes IP available to its CFC without a charge and the CFC uses the intangible to earn Subpart F income, whether by on-licensing the transferred right for a royalty that is foreign personal holding income⁷² or by performing services that are foreign base company services, the income from the intangible right would be included in the income of the U.S. shareholder as a result of the Subpart F inclusion.

Accordingly, the second premise disregards the reality, well known to the Treasury and I.R.S. (as well as multinational taxpayers), that a taxpayer only is tested under Section 482 if challenged, that challenges to transfer pricing are resource intensive for the government, that probability of a successful challenge by the government is uncertain, and that the limited number of cases brought have a broad range of outcomes.⁷³ Importantly, there is no bar to filing a return that would be adjusted under Section 482 if examined and litigated to final conclusion. For these reasons, there is little downside risk for a taxpayer that takes a transfer pricing position on a tax return that is defensible but nonetheless is subject to challenge.⁷⁴

Section 482 allocations do not distinguish between categories of income. The Section 482 regulations include a set-off rule. Even if an adjustment to a taxpayer's income from a CFC is warranted for undercharging for intangible property, under the regulations, no allocation is required if in the same year the taxpayer overcharged the CFC in another transaction, without regard to whether the other transaction is in the same product category as that which the intangible benefits.⁷⁵ In this case, absent the allocation (and none would be required), the arbitrary limitation of the R&D allocation to gross intangible income would result in zero R&D allocation. Again, the Proposed Regulation's reliance on Section 482 is misplaced.

Returning to the example where a royalty should be charged that would have reduced Subpart F income, it is noteworthy that changing the categorization of the income from a Subpart F inclusion to a royalty would not change the foreign tax credit limitation category of the income. Strangely, under the Proposed Regulation's arbitrary gross intangible income approach, R&D would not be allocated to the Subpart F income but would be allocated if it were a royalty.⁷⁶

To be effective, the Proposed Regulation's R&D allocation rule would have to create a new category of potential Section 482 cases, recasting Subpart F income and GILTI as royalties in order to protect the foreign tax credit limitation by requiring an appropriate R&D expense allocation.⁷⁷ It simply is not credible that, assuming it spotted the issue, the IRS would use its scarce resources to adjust the intangible return portion of the Subpart F income inclusion to be a royalty (thereby reducing the Subpart F inclusion by the same amount without changing the overall amount or source of U.S. taxable income).

The uncertainty of application of Section 482. In many cases it will be unclear as a factual matter whether even successful R&D should be the subject of a charge under Section 482. It may be incurred in a year before the R&D project is found to be successful. If it is successful, it may not have been identified with intangible property. It may be basic R&D and not used directly in the company's single line of business, but it may add to the company's background knowledge or enhance its reputation in its industry. In this case, there is no reason to believe a priori that these deductions would not benefit the same line of business conducted in a CFC (e.g., in Juárez, Mexico) to the same extent as that conducted directly by the taxpayer (e.g., in El Paso, Texas).

In cases where the R&D can be associated with an intangible, such that a Section 482 charge would be clear, cases illustrate the very wide range of potential results. For example, in the recent Amazon case, the Ninth Circuit upheld the Tax Court's determination of a “buy-in” for three categories of intangible property of $779 million compared with the taxpayer's position of $254.5 million. The accepted adjustment is 306% of the taxpayer's position. Absent the IRS's examination of the issue, gross intangible income of the U.S. group would have been half a billion dollars less.

For all of the preceding reasons, the second premise and its reliance on Section 482 to assure that all R&D that results in intangible property that generates income and benefits a CFC is fully compensated, is unreliable.

Further, the Preamble would rely on Section 367(d) to ensure proper compensation for transfers of intangible property in an outbound Section 351 or 361 exchange. Section 351 and 361 involve transfers of property. As noted previously, R&D expenditures deducted under Section 174 are not capitalized to the cost of any property under Sections 263 or 263A. Accordingly, it is a factual question in each case as to whether the income in question fully compensates the portion of R&D appropriately associated with the intangible asset. A Section 367(d) royalty would be required raising the same issues in determining the appropriate royalty as previously discussed in relation to Section 482. Section 367(d), which applies to a limited category of cases, offers no panacea as an assurance that all R&D will be separately compensated.

2. Turning the Statutory Rationale for GILTI (and Subpart F) on Its Head

Best practices in tax system design counsel building redundant protections around tax boundaries otherwise protected only by provisions that are standard-based and difficult to administer and enforce. In the case of Section 482 and the boundary between a domestic corporation and a low-taxed CFC, that is one of the reasons for statutory adoption of GILTI.⁷⁸

It is disingenuous to rely on a second premise that Section 482 addresses intangible transfer pricing issues when Congress included the GILTI minimum tax in the TCJA as a result of concern about profit shifting in the context of an exemption or territorial system, a concern that would not exist if transfer pricing “worked” to identify and fully compensate such transfers. If Congress believed that to be the case, the GILTI regime would be unnecessary. Congress intended Subpart F and the GILTI minimum tax as rule-based backstops to catch abuses that Section 482's standards-based approach fails to stop. The Proposed Regulation would turn this structure on its head, relying on Section 482 to correct misallocations of R&D deductions for Subpart F and GILTI purposes.

The second premise is utterly unreliable as a basis for the interpretation in the Proposed Regulation. The broad allocation of deductions to foreign income under Section 862, and allocation of R&D to income in broad product categories and apportionment based on sales that look through CFCs explicitly does not rely on the vagaries of Section 482 application and enforcement. The Proposed Regulation's segregation of gross intangible income in order to exclude allocation of R&D expense to CFCs' Subpart F inclusions and GILTI is inconsistent with this statutory design and all prior regulatory interpretations.

V. IS THERE A SOUND POLICY JUSTIFICATION FOR THE PROPOSED REGULATION'S FAILURE TO ALLOCATE R&D TO INCOME FROM CFCS?

The proposed exclusion of Subpart F inclusions and GILTI from direct allocation of R&D is an arbitrary and unprecedented interpretation of the statute. For the reason set forth above, the Preamble's justification for this interpretation is deeply flawed and does not provide a reasoned justification for the rule.

The question may be asked whether the rule can otherwise be justified on policy grounds. The Preamble suggests, without providing any empirical support, that failure to adopt the rule would cause R&D to be located outside the United States. Any allocation to foreign income of a U.S. shareholder deduction for an expense that is not deducted in the foreign country (e.g., stewardship and R&D) and that reduces allowable foreign tax credits increases the marginal U.S. tax on that activity compared to no allocation. Whether that is good policy depends on whether any R&D that would relocate as a result of the allocation (after accounting in the baseline for the effects of higher tax benefits for deductions in other countries and lower tax rates on certain IP income) would result in a U.S. detriment that exceeds the benefit to the United States of having the additional tax revenue from R&D that stays in the U.S. (whether it is used for social services including pandemic relief, defense, deficit reduction or debt reduction). This is an empirical question not addressed in the Preamble.

The responsiveness of the location of R&D depends on factors and elasticities that are not analyzed in the Preamble. Indeed, there is little prior evidence on whether R&D allocation would drive location decisions,⁷⁹ and we are unaware of any empirical analysis of the issue under the new law. If, nonetheless, a rationale for the proposed rule is to serve as an incentive for U.S.-located R&D, it would be a poorly designed incentive.

• The subsidy would primarily benefit R&D of multinationals, not R&D of small and medium-size businesses and startups. Yet there is reason to believe that R&D at small and startup businesses is more productive.⁸⁰

• The subsidy for CFC's taxes paid to foreign governments would support foreign rather than U.S. investment, and would be directed at R&D whose benefits are in whole or in part for non-U.S. business. This support for foreign over U.S. investment has not been proven justified by the marginal additional spillover benefits from increased U.S. R&D.

• The subsidy arises from reimbursing foreign taxes on U.S. income contrary to the purpose of the foreign tax credit limitation. While we do not disfavor paying taxes to foreign governments, it is longstanding U.S. policy that they should not be reimbursed from U.S. taxes on U.S. income.

• To the extent that the recipients of the subsidy's benefit are a U.S. multinational's public shareholders,⁸¹ the preponderance of these beneficiaries who are U.S. resident would be high income, wealthy and most often both. The subsidy would worsen rather than alleviate U.S. economic inequality. Moreover, based on the best available data of who owns U.S. equities, as much as 25% to 35% of the beneficiaries would not be U.S. residents.⁸²

There is no principled and empirically supported policy rationale for the complete exclusion of CFC earnings from R&D expense allocation. It goes without saying that the global COVID-19 pandemic provides no cover for the Proposed Regulation's exclusion of CFC earnings from R&D allocation. Federal revenue could be better used for public purposes than for subsidizing foreign investment of U.S. multinationals.

VI. CONCLUSION

The Proposed Regulation's failure to allocate R&D expense to Subpart F income and GILTI is a radical change to the longstanding regulatory interpretation of “taxable income from sources without the United States.” There is no indication in the legislative history that this change was sought by Congress and there is a long contrary history indicating Congressional blessing of or acquiescence to the approach of the currently effective regulations.

The justifications provided by the Proposed Regulation for a non-allocation rule do not stand up to scrutiny. The assertion that R&D ultimately results in intangible property disregards the substantial expenditures made for unsuccessful R&D that does not result in intangible property. The assertion that R&D resulting in intangible property will be compensated by returns to the intangible property disregards the substantial R&D performed to develop or support intangible property licensed under open license or royalty-free business models. It also disregards the difficulty of associating R&D with intangible property developed long after the R&D. Finally, the assumption that Section 482 transfer pricing authority granted to the government will assure that all returns from intangible property developed from R&D will be paid by CFCs flies in the face of the practical realities of enforcing transfer pricing, belies the Congressional purpose for adopting the GILTI rule to restrict profits shifting notwithstanding the availability of Section 482, and verges on the fantastic as a reasoned justification for the rule.

The effect of the Proposed Regulation's R&D allocation rule will be to overstate taxable income from sources without the United States in the numerator of the foreign tax credit fraction and thereby to reimburse taxpayers for foreign taxes. This will benefit the foreign countries whose taxes would be reimbursed by the United States.

The Proposed Regulation's exclusion of Subpart F inclusions and GILTI inclusions from R&D expense allocation should not be adopted. Instead, R&D deductions should be allocated across all categories of income within groupings of 3-digit SIC product categories and otherwise follow Regulation §1.861-17 as it would be modified by the Proposed Regulation.

Thank you for your consideration of this comment. We would be pleased to answer any questions you or your colleagues might have.

Sincerely,

Stephen E. Shay
Reuven Avi-Yonah
Patrick Driessen
J. Clifton Fleming, Jr.
Robert J. Peroni

Cc:
L. G. “Chip” Harter, Deputy Assistant Secretary (International Tax Affairs), Department of Treasury
Douglas Poms, International Tax Counsel, Office of International Tax Counsel,
William M. Paul, Deputy Chief Counsel (Technical), Internal Revenue Service
Peter Blessing, Associate Chief Counsel (International), Internal Revenue Service