I'm currently teaching a class on U.S. international tax law, which has helped me to reconnect with nitty-gritty details of the existing rules that don't always feature in my (or other people's) analyses of broader conceptual issues in the field. In a class that is coming up soon, we will be discussing the U.S. transfer pricing rules.
From a purely pedagogical perspective, when I read the recent case of Altera Corp. v. Commissioner, decided unanimously by 15 U.S. Tax Court judges on July 27 of this year, I felt like the recipient of a rare gift. Most of the transfer pricing cases are long, fact-specific, and based on past iterations of the transfer pricing regulations that the IRS and Treasury subsequently tried to fix. This tends to leave the cases' continuing precedential value and broader interest far too limited to justify spending a lot of the time on them in a 3-hours-per-week general survey course.
The frequency of changes to the relevant regulations reflects Rule 1 of U.S. transfer pricing litigation, which holds that the government always loses. (This is not quite literally true - but in the few transfer pricing cases that the government won there were generally egregious taxpayer blunders, unlikely to be repeated, such as failing to do anything to establish a proper fig leaf, and/or leaving memos in the files avowing an intention to use bogus transfer prices.)
By such standards, Altera is a dream case for three reasons. First, the regulations under which it was decided remain almost up-to-date. (They were issued in 2003, and subsequently revised in 2011, but not, it appears, relevantly to the main issue in the case.) Second, Altera was decided on summary judgment, so its discussion and analysis almost exclusively pertain to broader legal issues, rather than to narrower factual ones. Third, there could be no better illustration than this case of the farcical nature of U.S. transfer pricing practice, and of the need for it to change.
I view Altera as a farce in three acts (more on this shortly), but this does not count the already farcical set-up. Section 482 of the U.S. Internal Revenue Code authorizes the Commissioner to restate the claimed terms of purported transactions between commonly-owned businesses if "he determines that ... [this] is necessary in order .. clearly to reflect ... income." These words could hardly sound more deferential to the Commissioner's administrative discretion. But unfortunately, the regulations state that "the standard to be applied in every case is that of a taxpayer dealing at arm's length with an uncontrolled taxpayer."
This is far more limiting that "clearly reflecting income," and can be read as suggesting a need to treat actual arm's length transactions between unrelated parties as relevant legal precedents for IRS transfer pricing determinations under section 482. Now, there is a huge literature about all this, which (in my reading, at least) has reached the predominant conclusion that an arm's length approach is completely useless, due to the both theoretical and practical problems that it faces. But for many in the field, adherence to arm's length appears to remain a matter of quasi-religious faith.
Altera itself concerned a regulatory election, under which U.S. taxpayers with foreign affiliates can opt to use an approach called "cost-sharing" for purposes of determining the applicable transfer prices within the group. In the typical case, a U.S. company with U.S. employees who live, say, in California or the Pacific Northwest is creating what it hopes will be valuable intellectual property (IP) that can be profitably exploited worldwide. Cost-sharing is a device that they use to shunt as much of the overall profits as possible to tax haven subsidiaries in, say, the Cayman Islands.
Now, in the real world of transactions between unrelated parties there sometimes are actual "cost-sharing" agreements. For example, two companies with complementary skill sets might agree to collaborate on something that they hope will make them both a lot of money. In such a case, they may agree that the ultimate profit split will be affected by how much $$ each of them has expended in the development process.
Then there is fake cost-sharing between affiliates, the topic of interest here. Back to our U.S. company. It has all the employees and all of the relevant skills for developing particular IP. But it creates a Caymans affiliate that, in substance, contributes nothing to the process. But the Caymans affiliate does indeed observably purport to contribute cash to help pay for developing the IP. Where did it get the cash? Easy, the U.S. parent will typically have given it the cash, in exchange for all of its equity, so that the affiliate could hand the cash right back to the U.S. parent via the pretense of paying for a portion of the development costs. The Caymans affiliate may then get, say, 100% of the upside with regard to profits from selling the IP in all countries outside the U.S.
In short, the typical deal is like (one suspects) almost no actual cost-sharing arrangement in the history of arm's length transactions. One party (the U.S. parent) contributes everything, while the second party (the Caymans sub) contributes nothing, except for giving back cash that the first party had placed in its bank account 5 minutes earlier.
It is already giving these transactions too much credit to say that the Caymans affiliate has effectively gotten the entire foreign "upside" in exchange for nothing. Its getting this upside (and thereby "bearing risk" regarding how great this will actually be) is completely meaningless, given common ownership. But in fact no party to a true arm's length cost-sharing arrangement would get the opportunity to make this sort of a deal. You don't get a piece of the upside without bringing something real to the table. So it's fundamentally ludicrous to look at actual arm's length cost-sharing deals for evidence of how a particular item within the broader deal ought to be treated in related-party arrangements.
Giving away all the foreign upside in exchange for nothing is already a nice feature of supposed cost-sharing arrangements between commonly owned affiliates. But it's better still if, contrary to the supposed logic of section 482 cost-sharing, you can also give disproportionate deductions to the U.S. affiliate. This further increases the proportion of taxable income that can be treated as arising in a tax haven, rather than in the U.S.
Taxpayers have assiduously pursued these opportunities. Earlier versions of the cost-sharing regulations had proven ripely exploitable, forcing the IRS to revise them, but it had also taken two beatings in prior litigation concerning the earlier regulations.
So the farce really started long before Altera. In Altera itself the legal issue was relatively narrow, although (as we will see) its implications are considerably broader. Taxpayers evidently saw how they could take advantage of the fact that common practice in the IP industry involves giving the "talent" incentive compensation such as stock options. Thus, if the engineering team contributes to hitting a home run, such that the value of the company's stock skyrockets, the members of the team get to see the value of their compensation go up accordingly.
A trick that taxpayers came up with was to argue that, under the cost-sharing regulations, this incentive compensation - often a huge piece of the overall development costs - should be excluded from those that the Caymans affiliate needs to "share." This wouldn't matter economically, but it would permit the U.S. parent's taxable income to be lower, and the foreign affiliates' share to be higher, than if such costs were included in the cost-sharing formula.
Even before the final version of the 2003 cost-sharing regulations came out, it was clear that the IRS would require including incentive compensation in the costs to be "shared." So the regs would have to be invalidated in this regard, if the above plan was to work. Taxpayers therefore set up a farce that played out in the following 3 stages:
(1) The industry and its friends flooded the notice-and-comment process that gave rise to the 2003 regulations with extensive information documenting that true arm's length cost-sharing deals NEVER require the parties to share the cost of each other's incentive compensation. They also explained why this was so. For example, it would give unrelated parties odd incentives, e.g., to try to drive down each other's stock price so that the costs one had to share would be lower. Needless to say, these considerations don't actually apply to related party deals, where there is only one affiliated group, playing on both sides, and thus there is no possible concern about thus harming deal concord.
The taxpayers of course did not have to show (as would have been impossible) that arm's length parties would ever make a deal in which one side provides all the value, and the other gets a huge piece of the upside despite adding nothing that the other side needed. For good measure, the taxpayers proffered statements by reputable leading experts, saying, for example that there is no economic cost to a corporation or its shareholders of providing stock-based compensation. If this is true, I would like to offer $5 per firm for options just like those that high-end IP firms grant to their star employees.
(2) As no doubt was expected, the Treasury stuck to its guns in the final regulations. It kept the requirement that incentive compensation be included in cost-sharing. In two important respects - each no doubt anticipated by the strategists on the other side - the way in which this was done placed the regulations in legal peril. First, the transfer pricing regulations as a whole continued to say that the standard in all cases is that of arm's length transactions between unrelated parties. There was no separate reliance on clear reflection of income. Second, the preamble to the regulations offered conclusory statements to the effect that the Treasury was simply unpersuaded by the evidence that taxpayers had offered in step (1) of the farce. The preamble did not carefully explain, for example, why the facts evinced concerning true arm's length deals had little bearing here, given other differences between the two settings. What made this unsurprising was common practice by the Treasury. Preambles generally are not written as litigation documents - although, after Altera, they probably will be - because the Treasury evidently believes (or has believed) that its seemingly broad administrative discretion makes this unnecessary.
(3) The final stage of the farce took place before the Tax Court in Altera. The taxpayer's litigators successfully peddled a dramatic story of stubborn regulatory high-handedness. In fact, what the Treasury had been guilty of was indifference to evidence that was logically irrelevant. But 15 Tax Court judges bought the story sufficiently to be unmoved even by the IRS argument that cost-sharing's elective character as a taxpayer method should make full adherence to "arm's length" unnecessary here, even if it is required elsewhere under the transfer pricing regulations.
Altera likely has broader implications for the tax regulatory process. Taxpayers will regularly flood the notice-and-comment process with evidence and arguments that the Treasury has now learned it will need to rebut expressly and extensively, such as in preambles to final regulations. The point need not be to persuade the Treasury - just to delay it and raise the legal risks it faces. The preambles, or other published support for final regulatory pronouncements, will need to be written as litigating documents, in cases where a serious and well-funded legal challenge can be anticipated.
In addition, the transfer pricing regulations generally (i.e., not just in cost-sharing) are likely to be subject to multiple challenges. These regs have developed over the years to have an ever more "formulary" character. They set forth multiple approaches that look, say, at the profit split or rates of return being claimed by the different members of a commonly owned group. In many of these cases, taxpayers may be able to adduce evidence that, in arm's length deals of a seemingly (but not actually) similar character, particular aspects of a given formula are not in fact taken into account. So the farce of existing transfer pricing practice has a good chance of getting a lot worse.
How the Treasury should respond to this is not entirely clear. But one thing they certainly should do is delete, as soon as possible, the statement in the regulations that arm's length, rather than clear reflection of income, applies "in every case."
There are also arguably broader implications for the ongoing BEPS process. Obviously, Altera is not a relevant precedent outside the United States. But it shows what can happen if one doggedly tries to apply arm's length "evidence" and reasoning outside their actual realm of economic meaningfulness and relevance.