Altera, the Arm’s Length Standard, and Customary International Tax Law

Cite: Reuven S. Avi-Yonah, Altera, the Arm’s Length Standard, and Customary International Tax Law, 38 MJILOpinioJuris 1 (2017),



Reuven S. Avi-Yonah*

Irwin I. Cohn Professor of Law, the University of Michigan 


The recent Altera case in the US Tax Court (on appeal to the Ninth Circuit) raises interesting issues in regard to the much-debated topic of whether customary international tax law (CITL) exists. Altera involved the question whether the cost of employee stock options should be included in the pool of costs that must be shared under a cost sharing agreement. In Xilinx, the Ninth Circuit held under a previous version of the regulations that these costs should not be included because unrelated parties operating at arm’s length would not have agreed to include them. Treasury then amended the regulation to state specifically that “all” costs includes the cost of stock options but did not carve out an exception from the arm’s length standard. In Altera, the Tax Court sitting en banc invalidated the new regulation on the ground that it was inconsistent with the arm’s length standard (ALS). This article-in-abstract discusses the implications of Altera for the long-running debate about whether CITL exists and whether it is binding on the United States.

  1. Introduction: Xilinx and Altera

In Xilinx v. Commissioner (125 TC 37 (2005), aff’d, 598 F.3d 1191 (9th Cir. 2010)), the Tax Court held that the cost of employee stock options cannot be included in the pool of costs shared under a cost sharing agreement because unrelated parties would not have agreed to share theses costs and the ALS takes precedence over the cost sharing regulations. A panel of the Ninth Circuit initially reversed, arguing that the more specific cost sharing regulations should override the ALS, and that the omission of stock option costs would undermine Congressional intent to ensure that all costs are included under the “commensurate with income” standard of IRC section 482.[1] A storm of criticism over the court violating the ALS ensued, and the same Ninth Circuit panel vacated its previous ruling and affirmed the Tax Court.

Treasury then amended the cost sharing regulations (Treas. Reg. 1.482-7(d)(2)) to require specifically that the cost of options be included, relying on the commensurate with income standard.[2] However, Treasury did not carve out an exception to the ALS; instead, it stated that such a result “is consistent with the arm’s length standard.”[3] In Altera, 145 TC 3 (2015), a unanimous Tax Court sitting en banc overturned the new regulation because Treasury failed to address evidence that unrelated parties would not have agreed to share the cost of stock options and the ALS (Treas. Reg. 1.482-1) controls.[4]

The issue is on appeal to the Ninth Circuit, and I have joined an amicus brief arguing that Treasury was entitled to rely on the commensurate with income standard: In the absence of true comparables it is impossible to prove that the result is inconsistent with the ALS, and the legislative history of the commensurate with income standard shows that Congress mandated the inclusion of all costs regardless of what unrelated parties would have done.[5] The purpose of this article is to raise the broader question of the implications of the Treasury’s behavior for the question whether CITL exists.


  1. Does CITL Exist?

Customary international law is accepted by the US as a binding part of international law.[6] In order for customary international law to exist, three elements must be fulfilled: (a) the widespread repetition by States of similar international acts over time (State practice); (b) the requirement that the acts must occur out of a sense of obligation (opinio juris); [7] and (c) that the acts are taken by a significant number of States and not rejected by a significant number of States.[8]

In the case of tax law, the first and third elements are relatively easy to prove. There are over 3,000 bilateral tax treaties covering almost every nation on earth, and they all follow similar models (the OECD, UN, and US model treaties). As Ash and Marian have shown recently, about 80% of the words of each tax treaty are identical.[9] Moreover, of all the articles in the treaty, Article 9 (Associated Enterprises), which mandates the ALS, shows the greatest level of identity across treaties.[10]

The harder part is to prove opinio juris, i.e., do states follow the treaties even when they are not legally bound to do so (e.g., in a non-treaty case)? I have previously argued that the behavior of the US in certain cases indicates that it believed itself bound by certain international tax norms even when not legally bound by treaty or otherwise.[11] For example, the invention of the deemed dividend concept for Foreign Personal Holding Companies in 1937 suggests that the US felt itself bound not to tax a foreign corporation on foreign source income even when the corporation is 100% controlled by US residents. The invention of the concept of “tax exempt related party” in the earnings stripping context (IRC section 163(j), 1989) indicates that the US felt itself bound by an obligation to not violate the principle of non-discrimination embodied in Article 24 of the treaties, even though it could have overridden the treaties. And the insistence in the 1988 white paper that profit split, even though it does not rely on comparables for allocating residuals, is an “arm’s length return method” suggests the US felt itself bound by the ALS.[12]

Altera provides an interesting case study that further shows that at least in the US the ALS is considered binding even in a non-treaty situation, i.e., that it is part of CITL.[13]


  1. Altera, CITL and the ALS

Altera involved a cost sharing agreement between Altera, Inc. and its Cayman Islands subsidiary, and the US does not have a tax treaty with the Cayman Islands. Xilinx, on the other hand, involved a cost sharing agreement with Xilinx’s subsidiary in Ireland, so that the US-Ireland tax treaty applied. Thus, when confronted with Xilinx, Treasury could have distinguished it as a treaty case and amended the 1.482-7 cost sharing regulation any way it wanted for non-treaty cases.[14] Specifically, Treasury could have relied on the legislative history of the commensurate with income standard of IRC 482. Congress stated for a cost-sharing agreement to satisfy the commensurate with income requirement, “the income allocated among the parties” should “reasonably reflect the actual economic activity undertaken by each,” meaning that “the cost-sharer would be expected to bear its portion of all research and development costs.” H.R. Rep. No. 99-841, at II-638 (1986) (Conf. Rep., emphasis added).

Moreover, Congress also stated that this result should govern regardless of what unrelated parties would have done at arm’s length, stating that Treasury would not be required to focus on “industry norms or other unrelated party transactions” if they would not exist in a particular context (like “related party intangibles transfers”). H.R. Rep. No. 99-426, at 425 (1985). Congress explained that such transactions rarely if ever occur between unrelated parties:

A fundamental problem is the fact that the relationship between related parties is different from that of unrelated parties…. The problems are particularly acute in the case of transfers of high-profit potential intangibles…. Industry norms for transfers to unrelated parties of less profitable intangibles frequently are not realistic comparables in these cases. Transfers between related parties do not involve the same risks as transfers to unrelated parties.[15]

Thus, Treasury could have just stated that the inclusion of the cost of stock options in the cost sharing pool between related parties is an exception to the ALS because there are no realistic comparables and precisely because unrelated parties would not have agreed to share of cost of stock options since the value of the options depends on the performance of an entity that by definition they do not control (i.e., an unrelated party). The Xilinx outcome, after all, warned the Treasury not to address the ALS risks–precisely the result reached by the Tax Court in Altera.

Instead, Treasury in a non-treaty context, in which it was not bound by the ALS (which is for domestic law purposes only a regulatory requirement, since it is not in IRC 482), chose to stick with the ALS and risk the consequences. Altera shows that it was (arguably) wrong to do so. But the strong implication is that Treasury believes itself bound by the ALS even when there is no formal legal requirement to be so bound (i.e., where there is no treaty), and that is the essence of opinio juris.[16] Thus, I believe Treasury’s behavior indicates that it believes the ALS is part of CITL, and hence that CITL exists. This, of course, is a very controversial conclusion, since many commentators doubt the existence of an international tax regime, and a fortiori of CITL.[17]


  1. Implications: Should Barclays be Revisited?

If CITL exists, and if the ALS is part of it, then, absent a specific contrary statutory provision, it binds the US as a matter of US law.[18] Most importantly, unlike tax treaties, which in the US do not apply to the states, CITL would apply to state laws.[19] And in that case, perhaps the US Supreme Court was wrong in its conclusion in Barclays that formulary apportionment is consistent with the foreign Commerce Clause.[20] If the states are bound to follow the ALS in dealing with foreign parties, perhaps the water’s edge approach (applying the ALS to multinationals operating outside the US while using formulary apportionment for domestic operations) that the states generally adopted despite winning in Barclays is not just politically prudent but a legal requirement as well.

I have long advocated the overthrow of the ALS, and continue to believe that–at least where comparables cannot be found–some other solution (like a formula) is necessary.[21] But until the OECD is persuaded otherwise, the ALS is part of CITL, and the Xilinx/Altera saga (however the Ninth Circuit decides) shows that this is true from a US perspective as well. 



* Irwin I. Cohn Professor of Law, the University of Michigan. Thank you to Steve Ratner, Kristina Daugirdas Monica Hakimi and Julian Mortenson for comments on an earlier version.

[1] On the original Ninth Circuit opinion see Reuven S. Avi-Yonah, Xilinx and the Arm’s-Length Standard, 123 Tax Notes (June 8, 2009); 54 Tax Notes Int’l 859 (June 8, 2009).

[2] Treas. Reg. § 1.482-7(d)(3) (2015) (“For purposes of this section, IDCs mean all costs, in cash or in kind (including stock-based compensation, as described in paragraph (d)(3) of this section.”).

[3] Treas. Reg. § 1.482-7(d) (2015); see also Compensatory Stock Options Under Section 482, 68 Fed. Reg. 51171-02 (Aug. 26, 2003) (“Treasury and the IRS continue to believe that requiring stock-based compensation to be taken into account for purposes of QCSAs is consistent with the legislative intent underlying section 482 and with the arm’s length standard (and therefore with the obligations of the United States under its income tax treaties and with the OECD transfer pricing guidelines). The legislative history of the Tax Reform Act of 1986 expressed Congress’s intent to respect cost sharing arrangements as consistent with the commensurate with income standard, and therefore consistent with the arm’s length standard, if and to the extent that the participants’ shares of income “reasonably reflect the actual economic activity undertaken by each.” See H.R. Conf. Rep. No. 99-481, at II-638 (1986). The regulations relating to QCSAs implement that legislative intent by using costs incurred by each controlled participant with respect to the intangible development as a proxy for actual economic activity undertaken by each, and by requiring each controlled participant to share these costs in proportion to its anticipated economic benefit from intangibles developed pursuant to the arrangement. In order for the costs incurred by a participant to reasonably reflect its actual economic activity, the costs must be determined on a comprehensive basis. Therefore, in order for a QCSA to reach an arm’s length result consistent with legislative intent, the QCSA must reflect all relevant costs, including such critical elements of cost as the cost of compensating employees for providing services related to the development of the intangibles pursuant to the QCSA. Treasury and the IRS do not believe that there is any basis for distinguishing between stock-based compensation and other forms of compensation in this context.”) (emphasis added).

[4] Treas. Reg. § 1.482-1(b)(1) (2015) (“In determining the true taxable income of a controlled taxpayer, the standard to be applied in every case is that of a taxpayer dealing at arm’s length with an uncontrolled taxpayer. A controlled transaction meets the arm’s length standard if the results of the transaction are consistent with the results that would have been realized if uncontrolled taxpayers had engaged in the same transaction under the same circumstances (arm’s length result). However, because identical transactions can rarely be located, whether a transaction produces an arm’s length result generally will be determined by reference to the results of comparable transactions under comparable circumstances. See § 1.482-1(d)(2) (Standard of comparability).”) (emphasis added).

[5] Breif of Amici Curiae Anne Alstott, Reuven Avi-Yonah, Lily Batchelder, Joshua Blank, Noel Cunningham, Victor Fleischer, Ari Glogower, David Kamin, Mitchell Kane, Sally Katzen, Edward Kleinbard, Michael Knoll, Rebecca Kysar, Zachary Liscow, Daniel Shaviro, John Steines, David Super, Clint Wallace, and George Yin in Support of Appellant, Altera Corporation and Subsidiaries, v. Commissioner of Internal Revenue, 2016 WL 3767014.

[6] The Paquete Habana, 175 U.S. 677, 700 (1900) (“International law is part of our law, and must be ascertained and administered by the courts of justice of appropriate jurisdiction as often as questions of right depending upon it are duly presented for their determination. For this purpose, where there is no treaty and no controlling executive or legislative act or judicial decision, resort must be had to the customs and usages of civilized nations…”).

[7] Anthea Elizabeth Roberts, Traditional and Modern Approaches to Customary International Law: A Reconciliation, 95 Am. J. Int’l L. 757, 757-58 (2001).

[8] Restatement (Third) of Foreign Relations Law § 102(2), n. 2 (1977).

[9] Elliott Ash & Omri Marian, The Making of International Tax Law: Empirical Evidence from Natural Language Processing (Nov. 9, 2017) (unpublished paper) (presented at National Tax Association)

[10] Id.

[11] See Reuven S. Avi-Yonah, International Tax as International Law, 57 Tax L. Rev. 483, 498-501 (2004).

[12] On the white paper, see Reuven S. Avi-Yonah, The Rise and Fall of Arm’s Length: A Study in the Evolution of U.S. International Taxation, 15 Va. Tax Rev. 89, 93–94 (1995).

[13] For an early argument that the ALS is part of CITL see Chantal Thomas, Customary International Law and State Taxation of Corporate Income: The Case for the Separate Accounting Method, 14 Berkeley J. Int’l L. 99 (1996).

[14] See Rufus Rhoades, The Even More Curious Case of Xilinx, Inc. v. Commissioner and the Future of Transfer Pricing, Lexisnexis Legal Newsroom (March 31, 2010), (“The two regulations are certainly irreconcilable. Given that, one does wonder why the government did not meet that conflict between the two regulations head on and state that the “all cost” regulation is a divergence from the general arm’s length standard, but that Treasury is authorized to take that action. That position may not have carried the day, but it would have been far more difficult for the panel to hold that Treasury is not authorized to make small in-roads on the arm’s length principle than to try to convince the court that an obvious conflict really did not exist.”).

[15] H.R. Rep. No. 99-426, at 424-25 (1986).

[16] Compare the reference to the OECD Transfer Pricing Guidelines in the preamble, see Preamble to Treas. Reg. 1.482-7(d), with the cost sharing regulation invalidated in Altera, see Altera Corp. and Subsidiaries v. C.I.R., 145 T.C. 91 (2015). The TPGs are not formally binding on the US or any other country unless they are part of CITL.

[17] For the skeptical view see, e.g., H. David Rosenbloom, International Tax Arbitrage and the “International Tax System”, 53 Tax L. Rev. 137, 137—8, 166 (2000); Michael J. Graetz, Taxing International Income: Inadequate Principles, Outdated Concepts, and Unsatisfactory Policy, 54 Tax Law Review 261 (2001); Julie Roin, Taxation Without Coordination, 1 J. Legal Stud. (2002),; Mitchell A. Kane, Strategy and Cooperation in National Responses to International Tax Arbitrage, 53 Emory L.J. 89 (2004); Adam H. Rosenzweig, Harnessing the Costs of International Tax Arbitrage, 26 Va. Tax Rev. 555 (2007). For the contrary position see, e.g., Reuven S. Avi-Yonah, Commentary on Rosenbloom, 53 Tax L. Rev. 167, 169 (2000) (“The claim that an international tax regime exists, however, rests mainly on the bilateral tax treaty network, which, as Rosenbloom states, is “a triumph of international law… I would argue that the network of 1,500 or more bilateral tax treaties that are largely similar in policy, and even in language, constitutes an international tax regime, which has definable principles that underlie it and are common to the treaties”); Yariv Brauner, An International Tax Regime in Crystallization, 56 Tax L. Rev. 259, 259 n.2 (2013).

[18] The Paquete Habana, 175 U.S. 677, 700 (1900).

[19] See Thomas, supra note 15, at 111–2 n. 80 (for the majority view). Contra Phillip R. Trimble, A Revisionist View of Customary International Law, 33 UCLA L. Rev. 665 (1986); Phillip R. Trimble, Foreign Policy, Frustrated-Dames & Moore, Claims Court Jurisdiction and a New Raid on the Treasury, 84 Colum. L. Rev. 317 (1984); Fritz W. Scharpf, Judicial Review and the Political Question: A Functional Analysis, 75 YALE L.J. 517 (1966) (for a view that customary law is not binding on the states).

[20] Barclays Bank PLC v. Franchise Tax Bd. of Cal., 512 U.S. 298 (1994). In Container Corp. of America v. Franchise Tax Bd., 463 U.S. 159 (1983), the Court held that world-wide formulary apportionment for a US-based multinational did not violate the dormant Commerce Clause. In Barclays, the Court extended this holding to foreign-based multinationals, holding that the foreign Commerce Clause’s “one voice” requirement did not bar California from imposing worldwide formulary apportionment on a foreign-based multinational. The Bush Administration had sided with Barclays, but the Clinton administration sided with California, so that the 1992 election result may have determined the outcome. However, the Clinton Administration then persuaded California and other states to switch to the water’s edge approach under which the ALS governs both US and foreign based multinationals outside the US, thereby providing more evidence that the ALS may be part of CITL (since after the Barclays result there was no legal impediment for the states to continue applying formulary apportionment on a worldwide basis since they are not bound by tax treaties) see Barclays Bank PLC, 512 U.S. 298 (1994) at 328 n.30.

[21] See Reuven S. Avi-Yonah, Between Formulary Apportionment and the OECD Guidelines: A Proposal for Reconciliation, 2 World Tax J. 3 (2010).


Online Production Team:


Managing Editor Online

Lakshmi Gopal



Nir Fishbien

Lukas Kutilek

Chul Hun Lee

Gianluca Mazzoni

Shay Moyal


Online Editor

Ava Morgenstern



Leave a comment

Your email address will not be published. Required fields are marked *