The Altera Oral Argument

We welcome back guest bloggers Professor Susan C. Morse from University of Texas School of Law and my colleague Senior Lecturer on Law Stephen E. Shay from Harvard.  Professors Morse and Shay, build on their post last week to fill us in on what happened before the 9th Circuit in Altera. Keith

At the Ninth Circuit on Wednesday October 11, government counsel carefully threaded the needle of statutory and regulatory interpretation in Altera, a case about transfer pricing and administrative law. Taxpayer counsel appeared to overreach. It refused to concede that Treasury has any authority to regulate the pricing of intercompany intellectual property sharing under qualified cost sharing arrangements (QCSAs) unless the guidance proceeds from the starting data point of unrelated party dealings, otherwise known as comparability analysis.

read more...

The panel included Ninth Circuit Chief Judge Sidney Thomas, Ninth Circuit Judge Stephen Reinhardt, and D.C. Circuit Judge Kathleen O’Malley, sitting by designation. Reinhardt joined the first Ninth Circuit Xilinx decision overturning the Tax Court decision, which interpreted the prior cost sharing regulation to allow the IRS to include stock option costs in the pool of shared costs. After a rehearing, Reinhardt dissented in the superceding Ninth Circuit Xilinx opinion that upheld the Tax Court. In Xilinx, he would have allowed the government to require inclusion of stock option costs in a cost-sharing pool even under earlier regulations that did not explicitly address stock options. The final regulations at issue in Altera, the current case, plainly say that stock option costs must be included in a QCSA cost pool, to the disadvantage of U.S. multinational groups which as a result may take fewer tax deductions resulting from the exercise of stock options. Billions of dollars of tax revenue are at stake in Altera.

The oral argument featured three important threads: The imposition of an administrative law framework with a Chevron starting point; the argument that “arm’s length” is not synonymous with “comparability analysis”; and the idea that the second sentence of section 482, which refers to “commensurate with income” payment for intellectual property “transfers”, specifically envisioned transfer pricing not tethered to unrelated party data points.

Judge O’Malley, who brought seven years’ worth of D.C. Circuit administrative law experience    to the hearing, repeatedly insisted on a textbook administrative law analysis. She asked both parties whether there is statutory authority for these regulations under Chevron. Yes, replied the government. Chief Judge Thomas asked whether the government has the statutory authority to “eliminate” comparability analysis altogether, for all transactions. No, replied the government, here trying to thread the needle. The statute does not say “arm’s length,” let alone comparability. Both are described in regulations. But there is “too much history.”

Well, then if the government cannot erase the arm’s length standard, how can it write regulations that set aside unrelated party data, like the agreements taxpayers point to under which unrelated parties develop technology together without mentioning stock options? Judge Reinhardt suggested that the validity of the regulation had to do with its subject: the sharing of intangible assets. Perhaps comparability analysis is not relevant for transactions involving intangibles in particular, he suggested. Agree, with respect to cost-sharing arrangements, replied the government.

But why doesn’t the departure from comparability analysis for intangibles violate the arm’s length standard? In response to prompts from the panel, the government agreed that arm’s length and comparability do not “go hand in hand” and are “not synonymous.” There are several “means to [the] end” of an arm’s length result. In the case of QCSAs, unrelated party data is “inherently not comparable” and cannot support clear reflection of income.

Taxpayer counsel, in contrast, contended that “it has to be an empirical analysis” and that “you have to take comparables as far as they will go,” and appeared to argue that this approach was required by the statute itself. “What if [the comparables] don’t go anywhere?” asked Chief Judge Thomas. Well, replied taxpayer counsel, then the government should “erase” regulations’ reference to an arm’s length standard. In rebuttal, the government further argued that the term “arm’s length standard” is a “term of art” and that Treasury’s interpretation is entitled to deference.

The second sentence of Section 482, added in 1986, allows the government to adjust related parties’ inclusions from “transfer” of intangibles so that they are “commensurate with income.” As the government pointed out, the legislative history clearly explains that unrelated party data points – i.e., comparability analysis – are not sufficient to allow clear reflection of income in these situations involving intangibles. This is strong evidence of statutory authority for the government to write regulations that depart from comparability analysis. Taxpayer counsel suggested that a QCSA might not qualify as a “transfer” under this sentence of the statute, so that perhaps it was not statutory authority at all. But government counsel disagreed, arguing that the word “transfer” was broad enough to encompass QCSAs and noting that this issue was apparently briefed, and ignored, in Xilinx.

The Tax Court cited State Farm, which requires reasonable explanation of policy changes, in its decision to set aside the Treasury’s regulations. Other reasonable explanation cases include Fox Television and Encino Motorcars, both of which came up during oral argument. O’Malley asked government counsel why the regulations requiring cost sharing were not a change; the government replied that the policy of requiring stock options costs to be included in pools had existed since 1997, years before the regulations were promulgated. Later, taxpayer counsel pushed the State Farm argument, insisting that some of the government’s arguments in litigation were “not what they said” in the preamble. But the panel did not pursue the specifics of the preamble’s language. And taxpayer counsel’s assertion that Chevron should be the “last step” of the analysis of regulatory validity was met with silence by the court.

Stay tuned for our analysis of the Ninth Circuit’s Altera decision – we’ll blog it here in due course.

 

 

 

 

 

 

Ninth Circuit Hears Altera Tomorrow

We welcome back guest bloggers Professor Susan C. Morse from University of Texas School of Law and my colleague Senior Lecturer on Law Stephen E. Shay from Harvard.  Professors Morse and Shay, building on an earlier post as well as their amicus brief, explain that the Tax Court went too far in striking down Treasury regulations requiring the sharing of stock-based compensation costs in Altera.  The underlying issue as well as the procedural issue make this a case to watch. We have previously blogged about Altera here and here.   Keith

On Wednesday of this week, October 11, the Ninth Circuit will hear argument in Altera, a case about transfer pricing and administrative law. Politically, Altera is a case about big multinational technology companies and under-resourced government regulators. Technically, it is about the transfer of intellectual property rights from U.S. affiliates of a multinational firm (a “U.S. group”) to one or more non-U.S. offshore subsidiaries under a qualified cost sharing arrangement (QCSA).

read more...

Firms from Google and Apple to Altera, a semiconductor company owned by Intel, use the QCSA “cost sharing” strategy to support the attribution of intellectual property for tax purposes to low-tax offshore subsidiaries and thereby justify allocation of substantial taxable income to those subsidiaries. The smaller the amount of U.S. group costs included in the pool, the more tax revenue the U.S. loses with respect to the cost-shared IP. Billions of dollars are at stake. Two amicus briefs prepared pro bono by academics and former tax practitioners support the government and multiple amicus briefs on behalf of interested business groups support the taxpayer in this important litigation.

Altera challenged a final Treasury regulation that requires multinationals who enter into QCSAs with offshore affiliates to include the cost of stock options granted to employees who develop the IP (among other expenses) in the pool of costs to be shared. Under cost sharing, if net costs are borne by the U.S. group the non-U.S. affiliates must reimburse the U.S. group for that amount. Prior regulations did not specifically address the issue of stock option cost allocation in a QCSA. In a prior case, Xilinx, the Tax Court and Ninth Circuit held that the government could not make offshore affiliates pay a share of stock option expense under these earlier regulations.

The revised final regulation requires taxpayers to include stock option costs in the pool of expenses for determining cost sharing payments. They provide that this is required under the arm’s length standard and, consistent with the directive of Section 482 of the Internal Revenue Code, is necessary clearly to reflect the income of the U.S. group.

Taxpayers challenged the final regulation and won in Tax Court in a reviewed decision that was unanimous among the judges that participated. The Court held that the regulations departed from the historic understanding of “arm’s length standard” which required the use of data about unrelated party transactions. The Tax Court proceeded to conclude, under a review based on State Farm (US 1983), that the regulatory change was arbitrary and capricious under § 706(2)(A) of the Administrative Procedure Act.

The misconception in the Tax Court’s decision is fundamental. One reason is that the historic understanding of “arm’s length standard” does not require the starting point of data about unrelated party transactions. Sometimes an application of the arm’s length standard uses unrelated party data. For example, if a taxpayer sells a commodity to related affiliates and unrelated firms, the unrelated firm price is the right starting point for the related affiliate price, because it is sufficiently comparable. But in other cases, unrelated party transactions are not comparable enough to serve as good starting points.

The arm’s length standard has always been a counterfactual inquiry. It has always asked how a related party transaction would be treated if, contrary to fact, the same transaction (including the actual relationships presented in fact) were conducted by unrelated parties (i..e, as though the relationship did not exist). This does not mean insisting that the reasoning begin with an unrelated party transaction if that transaction has sufficiently different facts and is not comparable.

Several transfer pricing methods, including the comparable and residual profit split methods, do not require use of unrelated party prices as starting points.   Moreover, large chunks of the 482 rules prove that the arm’s length standard is not a brittle instruction to use whatever unrelated party information is available. The 482 regs include many pages of comparability adjustments which at every turn show that a starting unrelated party price, even if available, often needs a lot of work before it can be considered a comparable.

Altera and other multinational tech companies want to avoid paying for the stock option cost component of technology by arguing that unrelated firms that share technology do not require payment for stock option costs. They say that the arm’s length standard requires a starting unrelated party data point, and further that any departure from the unrelated party data point requirement is a significant regulatory change.

One reason that Altera should lose in the Ninth Circuit is because the arm’s length standard does not, and never has, required a starting unrelated party data point in all cases. Government briefs include this argument. They show that uncontrolled joint development agreements were not relevant to the question of whether to include stock option costs in QCSAs because clear reflection of income for high-profit intangibles cannot succeed if it relies on uncontrolled party data.  One amicus brief points out that Section 482’s reference to pricing “commensurate with income” only makes sense if the arm’s length standard embraces transfer pricing that is not bound to unrelated party pricing.

Another amicus brief (ours, with coauthors) explains that unrelated party data points cannot be starting points for an arm’s length analysis if the unrelated information is wholly incomparable to the related party situation. This is the case for the evidence that Altera points to, which consists of technology sharing deals among unrelated parties that do not mention stock option costs. This evidence is not relevant for QCSAs because it is not comparable.

The facts of Example 2 in our brief illustrate the lack of comparability between unrelated party joint ventures and related party technology transfer agreements:

Assume that Company C and Company D are unrelated and want to share the R&D costs and benefits for a new innovation on a 50/50 basis.

Company C pays cash compensation of 80 and grants stock options with an expected cost of 20 for its R&D employees. Company D pays cash compensation of 20 and grants stock options with an expected cost of 80 for its R&D employees. There are two possible ways of looking at the R&D costs in this deal:

Option 1: If stock option expenses are included, the pool of expenses is 200, and each company pays 100. No transfer between C and D is required to achieve a 50/50 split of expenses.

Option 2: If stock option expenses are not included, the pool of expenses is 100: 80 contributed by Company C and 20 contributed by Company D. D would transfer 30 to C to achieve a 50/50 split of expenses.

The correct answer is Option 1. Any rational economic actor would estimate and incorporate the stock option expense cost. Note that Company C and Company D do not need to mention stock option costs in order to consider and incorporate them into their transaction. The lack of a specific mention of stock options in the unrelated party deal document does not mean that stock option costs are priced at zero or intentionally disregarded.

The arm’s length standard has always recognized the absence of comparable third-party transactions in some areas of transfer pricing, including the large-scale licensing of IP among related parties. Thus the revised regulation at issue in Altera does not revolutionize the meaning of arm’s length. Instead it stays true to the meaning of clear reflection of income.

Tune in again after October 11 to hear how the taxpayer, the government and the judges of the Ninth Circuit approached this case at oral argument.

 

Treasury on the Right Side of the APA in Altera

We welcome back guest blogger Professor Susan C. Morse from University of Texas School of Law and first-time PT blogger Professor Stephen E. Shay from Harvard.  Professors Morse and Shay joined forces with other law professors with expertise in tax administration and international tax identified in the body of the blog to produce an amicus brief designed to persuade the 9th Circuit that the Tax Court went too far in striking down Treasury regulations requiring the sharing of stock-based compensation costs in Altera.  This post explains the arguments presented in their brief.  We have previously blogged about Altera here.  It is certainly no exaggeration to describe Altera as the most important decision of the Tax Court in 2015.  The outcome of the case at the Circuit Court level has significant importance and the amicus brief offers the Court valuable insight.  Keith

In 2013, one of us did a presentation at a Tax Executives’ Institute lunch panel in the heart of Silicon Valley.   In the presentation, she dismissed the idea that Treasury’s 2003 regulations requiring the sharing of stock-based compensation costs in cost-sharing agreements could be anything but valid.  There was an audience question, “What about Altera?”  She simply replied, “What about Mayo?”  It seemed the obvious response.  But, apparently, Mayo was not sufficient for the Tax Court.

read more...

Mayo confirmed that Treasury regulations promulgated under Administrative Procedure Act (APA) notice and comment procedures, like administrative regulations under non-tax law, receive full Chevron deference.  For several years around the time that Mayo was decided, the tax administrative law literature was largely absorbed with issues like the deference due to more informal guidance, such as notices and revenue rulings.  But the final, notice-and-comment regulations that required the inclusion of stock-based compensation in the cost base for cost-sharing agreements were outside that discussion.   Surely, deference would be due.

Yet the Tax Court in Altera invalidated Treas. Reg.  § 1.482-7(d)(2)(2003) under the APA.  The Tax Court’s decision was not based on APA § 553(c), which contemplates the notice-and-comment rulemaking process and supports deference under Mayo (as well as Chevron).  Rather, the decision was based on APA § 706(2)(A), which empowers a court to invalidate a rule that is “arbitrary” and “capricious.”  The Tax Court relied on case law including State Farm, a 1983 Supreme Court case that found an administrative action reversing prior action to be arbitrary because it was unexplained and contrary to evidence in the regulatory record.   In Encino Motorcars, a June 2016 case, the Supreme Court said the arbitrary and capricious standard required of an agency “adequate reasons for its decisions.”

The Altera Tax Court focused its arbitrary and capricious analysis on Treasury’s decision to require the sharing of stock-based compensation expense for controlled party cost-sharing agreements in the presence of evidence (submitted in the notice-and-comment process) that uncontrolled parties did not share costs in joint development agreements.  There are now three briefs that support the government’s appeal and request to the Ninth Circuit that it reverse the Tax Court decision, including the Department of Justice’s brief and two amicus briefs.

The government’s brief, filed on June 27, argues that the uncontrolled joint development agreements were not relevant because clear reflection of income for high-profit intangibles is not supposed to rely on uncontrolled party data.  The government points to “coordinating amendments” promulgated with 1.482-7(d)(2) to show coordination between the “commensurate with the income” language of § 482, and its 1986 legislative history, and the general arm’s length standard thereby supporting exclusive reference to facts internal to the transaction.  As we say in our brief, we agree with the government.  A brief principally drafted by NYU’s Clint Wallace and joined by 18 law professors argues that the “commensurate with the income” portion of the statute provides an independent basis for the validity of the regulation (whether or not the general rule is satisfied).  Our brief agrees with that position as well.

The brief that we wrote with the help and advice of our fellow amici (Dick Harvey, Leandra Lederman, Ruth Mason and Bret Wells) makes a complementary, alternative argument under the “traditional” view of the arm’s length standard.  We argue that uncontrolled joint development agreements that do not take account of stock option expense do not provide good evidence of the prices that will “clearly reflect income” in controlled transactions.  This is because they are not sufficiently comparable to be reliable evidence under the standards of I.R.C. §482.

A key example in the brief is “Example 2”, which assumes unrelated parties in a joint development agreement have stock-based compensation costs disproportionate to expected benefits:

Company C and Company D are not commonly controlled and want to share the R&D costs for a new innovation on a 50/50 sharing ratio (based on expected future benefits from the innovation).   Company C and Company D will jointly own the resulting intellectual property on a 50/50 basis. Company C pays cash compensation of 80 and grants stock options with a cost of 20 for its R&D employees. Company D pays cash compensation of 20 and grants stock options with a cost of 80 for its R&D employees.

If stock option expenses are included, the pool of expenses is 200, and each company pays 100, so no cost-sharing payment is necessary. This is the correct answer. If stock option expenses are disregarded, however, the pool of expenses appears to be 100, and Company D appears to contribute only 20 to the pool of expenses. Under this (incorrect) analysis, Company D would be required to make a net payment of 30 to Company C as its share of costs. In other words, Company D and its shareholders will suffer an additional compensation burden of 30 if the stock-based compensation costs are not shared. This burden would be in addition to the 20 of cash compensation expense and the 80 of stock-based compensation expense that Company D already incurs.

In this litigation and in litigation over the sharing of stock-based compensation before such sharing was explicitly required by the cost-sharing arrangement regulations, taxpayers argued that stock-based compensation was disregarded because they were not real economic costs.  Yet the economics and accounting disciplines, in addition to the tax law, have recognized stock compensation, including stock options, as economic costs for some time.  A 1995 FASB release, for example, stated that financials would be more “representationally faithful if the estimated fair value of employee stock options was included in determining an entity’s net income, just as all other forms of compensation are included.”

Within the bounds of “traditional” arm’s length analysis, we think the most reasonable conclusion is that the uncontrolled party agreements cannot further the objective of clearly reflecting income, because they are not reliable comparables. The brief highlights that controlled parties generally have a common issuer of stock underlying stock-based compensation whereas uncontrolled parties do not, which presents incentives and risks in the uncontrolled transaction not found in the controlled transaction.

Altera is a case about administrative procedure. The issue presented is not whether the Treasury’s decision to disregard uncontrolled party joint development agreements was the only permitted interpretation, but rather whether Treasury’s decision was arbitrary and capricious under the APA § 706(2)(A) (not the IRC § 482) standard. Under the APA § 706(2)(A) arbitrary and capricious standard, the government need only show that it provided a sufficient explanation for its conclusion that these agreements could not form a basis for clearly reflecting income. This the government did, we argue, in the regulatory Preamble. For instance, it explained that “[t]he uncontrolled transactions cited by commentators do not share enough characteristics of QCSAs involving the development of high-profit intangibles to establish that parties at arm’s length would not take stock options into account in the context of an arrangement similar to a QCSA.” It also stated that “at arm’s length the parties to an arrangement … would ensure through bargaining that the arrangement reflected all relevant costs, including all costs of compensating employees.”

As tax law encounters administrative law more regularly in litigation, it turns out that things can be made more complicated than a straightforward application of Mayo.   The Ninth Circuit faces the challenge of translating the APA arbitrary and capricious legal standard to the tax setting in this case. Depending on how the case plays out, it may also have an opportunity to consider whether the Treasury’s interpretation of its own regulations deserves Auer deference.

If the Ninth Circuit were to disagree with us and also with the other arguments in favor of the government’s position, the question of remedy would arise.   Perhaps the regulations should be remanded to Treasury, as the NYU brief argues; as we say in our brief, we concur with that argument. Or, if the reg is invalid and the remand remedy is not pursued, the Ninth Circuit may have a chance to say what should happen to an elaborate statutory safe harbor – here, the -7 regulations that authorize cost sharing agreements – when one piece of the regime is invalidated. There appears to be the assumption that the rest of the taxpayer-favorable safe harbor stands even if one of the building blocks falls. Yet the safe harbor falls far short of achieving clear reflection of income in many cases without the stock-based compensation regulation. If the stock-based compensation reg is invalid, our brief observes that Treasury might reasonably conclude that the whole safe harbor ought to be withdrawn.