TaxProf Blog op-ed: Altera Redux, by Reuven Avi-Yonah (Michigan):
On June 7, 2019, the Ninth Circuit Court of Appeals released its long awaited opinion in Altera. Like its predecessor, the new panel chosen after Judge Reinhardt’s death reversed the Tax Court and held that the regulation requiring multinationals that enter into a qualified cost sharing agreement (QCSA) to share the cost of employee stock options is valid.
The issue in Altera has been litigated repeatedly. In Xilinx, the Tax Court held that a previous regulation requiring that all costs be shared including the cost of stock options was invalid because it conflicted with the governing arm’s length standard (ALS). On appeal, the Ninth Circuit first upheld the regulation, but then reversed itself over a strong dissent from Chief Judge Reinhardt.
Altera involved the same issue under a post-Xilinx version of the regulation, which explicitly required the inclusion of stock options in the costs to be shared under a QCSA and stated that this result was consistent with the ALS. The Tax Court, sitting en banc, held this regulation to be invalid under the APA because Treasury ignored comments that showed that there were no arm’s length comparables to sharing such costs. The Ninth Circuit reversed in a 2 to 1 decision joined by judge Reinhardt. This opinion was withdrawn after judge Reinhardt’s death, and a new panel was assembled, which has now rendered its opinion reversing the Tax Court and upholding the regulation.
Given the amounts of money at stake (much more than the $80 million litigated in Altera), it is to be expected that Altera (now owned by Intel) will appeal to the Ninth Circuit en banc and/or to the Supreme Court. It is therefore worth revisiting some of the underlying issues.
The basic problem in Altera is the same problem that has bedeviled courts ever since the ALS was incorporated into the regulations under section 482 in 1934: how to decide what unrelated parties would have done at arm’s length when comparables cannot be found? Until 1968 courts, including the Ninth Circuit, gave various answers to this question that rested on a variety of standards, such as what seemed fair or reasonable. In 1962 the House responded to the resulting uncertainty by passing a bill requiring formulary apportionment when comparables could not be found. The Senate, however, rejected this proposal and instead Congress mandated Treasury to issue regulations applying the ALS.
The result was the 1968 regulations under section 482 that created the three traditional methods of applying the ALS where comparables could be found, namely Comparable Uncontrolled Price (CUP), Cost Plus and Resale Price. The 1968 regulations also permitted fourth methods where comparables could not be found, without defining what these methods were.
In the period between 1968 and the promulgation of the current 482 regulations in 1994-5, numerous cases were decided, and in many of them either inappropriate comparables were used or the courts resorted to fair and reasonable results in the absence of comparables. In 1986 Congress acted again by adding to section 482 the “commensurate with income” standard in cases of transfers of intangibles and by mandating a new set of regulations. The new regulations added two methods, CPM and Profit Split, that relied on very loose standards of comparability and in the case of Profit Split abandoned comparables altogether in assigning the residual profit.
Cost sharing existed since the 1960s, but became important after the commensurate with income language was added to the statute because it enabled multinationals to shift profits from intangibles out of the US without triggering a super royalty under the commensurate with income language. In the 1990s, stock-based compensation became important, and that led to the current litigation about whether the cost of stock options should be included in the costs that have to be shared under a QCSA.
Fundamentally, the problem lies with the ALS itself. The reason that courts have had trouble applying the ALS is that in most cases comparables cannot be found. The reason for the absence of comparables is that multinationals exist in order to internalize costs that would be borne if the related parties were unrelated to each other, and because of this they can drive unrelated parties out of the relevant market because of their superior cost structure. That is why the GAO found in 1981 that the vast majority of transfer pricing cases under the 1968 regulations were decided using fourth methods and not the three methods based on comparables.
This issue becomes even more acute when we consider the specific question raised by Altera. The reason that unrelated parties do not share the costs of stock options is that the value of the option depends on the performance of the stock and an unrelated party will not agree to share in the cost of options when it cannot influence the value of the underlying stock (precisely because it is stock of an unrelated party). When the parties are related, the stock of the parent reflects in part the value of the subsidiary, and therefore the subsidiary is willing to share in the cost of stock options that are based on the value of the combined group.
That fact is the fundamental reason why the majority is right and the dissent wrong in Altera. As the majority points out, by the time the commensurate with income standard was added to the Code in 1986, Congress was well aware of the problem of finding comparables, and therefore instructed Treasury not to focus on comparables but rather on arm’s length results. The key point is that where there are no comparables, any result is an arms length result, because the opposite cannot be proven. This is why judge O’Malley is wrong when she says that “the absence of evidence is not evidence of absence”. In the case of stock options, the absence of evidence that unrelated parties would have shared them is in fact evidence of absence, because unrelated parties cannot be expected to share such costs precisely because they are unrelated.
Given this reality, to expect Treasury to agree that the cost of stock options should not be shared is tantamount to saying that where there are no comparables, the related parties can agree to any result they like regardless of the impact on tax revenues. That is not what Congress intended when it added the commensurate with income language to section 482. As the majority states, Congress clearly intended that where there are no comparables Treasury should require results that are compatible with the underlying clear reflection of income language which is the actual statutory standard of 482 (since the words arm’s length do not appear in the statute). Such results are also compatible with the ALS, since in the absence of comparables it is impossible to prove the opposite.
Hopefully, the majority opinion is the last word on this issue. But in case it does get relitigated, courts should bear in mind that where there are no comparables by definition, the ALS must be read to encompass any reasonable result. In the case of stock options, unrelated parties are not comparable with related parties, therefore comparables can never be found, and therefore the only way to satisfy the ALS is to require reasonable results, as courts have done repeatedly since the ALS was first promulgated in 1934.
Currently, there are signs that the OECD may be considering abandoning the ALS, and India in particular is ready to adopt full-fledged formulary apportionment. For those who believe that the traditional ALS has value where comparables can be found, I would recommend not being too rigid in requiring that the only way to establish arm’s length results is by using comparables. If they insist on this outcome, as the dissent in Altera seems to do, they may find themselves in a world with no ALS at all.
 Altera Corp. vs. Comm’r, Case: 16-70496, 06/07/2019, ID: 11323039, DktEntry: 153-1 (9th Cir. 2019).
 Ibid. For the previous opinion see Altera Corp. v. Commissioner, No. 16-70496, 16-70497 (9th Cir. 2018).
 Xilinx, Inc. and Consolidated Subsidiaries, Petitioner, v. Commissioner of Internal Revenue, Respondent, 125 T.C. 37 (2005).
 Xilinx, Inc. v. Comm’r, 598 F.3d 1191 (9th Cir. 2010).
 26 C.F.R. § 1.482-7A(d)(2).
 Altera Corp. & Subsidiaries v. Comm’r, 145 T.C. 91 (2015).
 Altera Corp. v. Commissioner, No. 16-70496, 16-70497 (9th Cir. 2018).
 For my previous writing on this issue see Avi-Yonah, Altera, the Arm’s Length Standard, and Customary International Tax Law, 39 MJIL Opinio Juris 1 (2017); Avi-Yonah, Xilinx Revisited, 57 Tax Notes Int’l 1141 (Mar. 29, 2010); Avi-Yonah, Xilinx and the Arm’s Length Standard, 123 Tax Notes (June 8, 2009); 54 Tax Notes Int’l 859 (June 8, 2009).
 For the history see Avi-Yonah, The Rise and Fall of Arm's Length: A Study in the Evolution of U.S. International Taxation, 15 Virginia Tax Rev. 89 (1995), cited in Altera, 11-12.
 Avi-Yonah, Rise and Fall, supra.
 See Avi-Yonah, Testimony on Profit Shifting by US Multinationals, Senate Permanent Subcommittee on Investigations (Sept. 20, 2012).
 U.S. GEN. ACCOUNTING OFFICE., GGD-81-81, IRS COULD BETTER PROTECT U.S. TAX INTERESTS IN DETERMINING THE INCOME OF MULTINATIONAL CORPORATIONS (1981).
 Altera, 75.
 Avi-Yonah and Kir, India’s New Profit Attribution Proposal And the Arm’s-Length Standard, Tax Notes Int’l (June 17, 2019).