INSIGHT: A Second Bite at the APA: Altera’s Rehearing and the Potential Invalidity of Cost-Sharing Regulations—Part Two

II. CASE LAW HISTORY

The question of whether stock-based compensation must be shared in qualified cost-sharing arrangements (QCSAs) is not new. In fact, Altera marks the third time federal courts will have grappled with this issue. In response to prior rulings, Treasury has tweaked the applicable regulations many times over the years. Notwithstanding the subtle differences in the language of the regulations themselves, a review of these predecessor cases yields recurrent themes that are central to the Altera analysis.

A. Seagate

Seagate Technology, Inc. v. Commissioner is a 2000 U.S. Tax Court case that examined stock-based compensation in the context of the “original” transfer pricing regulations promulgated in 1968. The taxpayer, a multinational company that, not surprisingly, wanted to avoid treating such compensation as a sharable cost, moved for summary judgment on the question of whether such costs were in fact costs within the meaning of the regulations. The Tax Court denied the motion but it set for trial the open question of whether sharing of stock-based compensation costs was a “circumstance comparable to those which would have been adopted by unrelated parties.” While Seagate was ultimately resolved by way of a settlement and therefore does not provide a decision on the merits of this question, the case set the stage for use of the “comparable transaction” argument by taxpayers in future cases.

B. Xilinx

Xilinx v. Commissioner involved facts strikingly similar to those in Altera. Xilinx Inc. is a U.S. multinational corporation that produced programmable logic devices (PLDs) and had employed a cost-sharing arrangement with its Irish subsidiary for the future joint development of their technology. When Xilinx, Inc. opted not to share stock-based compensation with the Irish subsidiary in its cost-sharing arrangement, resulting in Xilinx Inc. enjoying U.S. deductions and R&D credits totaling more than $200 million, the Internal Revenue Service assessed a deficiency based on a then-applicable (but since amended) transfer pricing rule, analogous to the stock-award regulation at issue in Altera, which required that all costs, including employee stock options (ESOs) and other stock-based compensation, be shared under a QCSA.

The procedural posture of Xilinx was also strikingly similar to that of Altera. The Tax Court invalidated the then-applicable regulations. Although the U.S. Court of Appeals for the Ninth Circuit initially reversed the Tax Court in a 2-1 decision, the Ninth Circuit withdrew its first opinion for reconsideration and then issued its final opinion affirming the Tax Court and striking the regulations—a mirror image of the Altera procedural landscape up to this point but for the pending final decision.

In defense of its decision not to treat ESOs as a cost, Xilinx, Inc. presented evidence of third-party transactions between uncontrolled parties in which such costs indisputably went unshared. The IRS did not challenge the alleged comparability of these transactions; rather, it maintained that the applicable Treasury Regulation 1.482-7(d)(1) required that ESOs be shared in QCSAs even if unrelated parties would not share them.

The regulation at issue in Xilinx, providing that “all costs” must be shared, appeared as Treas. Reg. 1.482-7(d)(1) at the time of the Xilinx Tax Court decision, and then, due to renumbering, as Treas. Reg. 1.482-7A(d)(1) at the time of the Xilinx Ninth Circuit decision. The regulation at issue in Xilinx, however, was substantively changed as a result of its amendment in 2003, and as a result, Altera discusses a similar provision in the stock-award regulation (i.e., Treas. Reg. 1.482-7A(d)(2)). The stock-award regulation added to the “all costs” language a specific reference to stock-based compensation, removing any doubt as to the applicability of the “all costs” language to such payments. Effectively, the regulation at issue in Xilinx and the stock-award regulation in Altera set forth two analogous, albeit non-identical, rules.

The Ninth Circuit in Xilinx analyzed the issue as a conflict between the statute and its regulations. Not only was there a potential conflict between the arm’s-length provision in Treas. Reg. 1.482-1 and the QCSA provision in Treas. Reg. 1.482-7, but the general rules governing cost-sharing arrangements were considered simultaneously, under which each controlled participant’s share of costs must be in conformity with that party’s share of reasonably anticipated benefits. The court also considered relevant provisions of the income tax treaty between the U.S. and Ireland, noting that they explicitly incorporated the arm’s-length principle for purposes of transfer pricing as between two entities eligible for benefits thereunder.

The Ninth Circuit found that, in light of the accepted fact that unrelated parties do not share stock-based compensation in their cost-sharing arrangements, the general arm’s-length standard of Treas. Reg. 1.482-1(b)(1) conflicted with the “all costs” language of Treas. Reg. 1.482-7A(d)(1). According to Judge John Noonan, “the plain language [of the all-costs rule] does not permit any exceptions, even for costs that unrelated parties would not share. Each provision’s plain language mandates a different result.” To address this conflict, the court sought to (1) apply a canon of statutory construction that would mandate that the specific statement would control the general statement and then to (2) resolve the ambiguity based on the dominant purpose of the regulations. As part of step 1, the court determined that the “all costs” rule of Treas. Reg. 1.482-7A(d)(1) was part of the regulations specifically addressing cost-sharing arrangements and that this was therefore the rule that governed. Nonetheless, the court pointed out that canons of construction are not “mandatory” rules and can be overcome by other circumstances manifesting legislative intent. In this case, the court determined that the manifest legislative intent was parity between controlled and uncontrolled taxpayers. Specifically, “if Xilinx [could not] deduct all its stock option costs, Xilinx does not have tax parity with an independent taxpayer.”

Therefore, the court determined that, in light of the all-encompassing nature of Treas. Reg. 1.482-1(b)(1), to read the more specific “all costs” rule as controlling over the arm’s-length principle would afford it an unreasonable scope—one that would contravene the legislature’s intent when promulgating Section 482. In other words, to apply the “all costs” rule in lieu of the “arm’s-length” rule would effectively turn a mere canon of construction into “something like a statute.”

The Ninth Circuit in Xilinx made clear that the canons of construction are mere “guides designed to help judges determine the Legislature’s intent” and “can be ‘overcome’ by ‘other circumstances’ manifesting that intent.” Judge Raymond Fisher added in a concurring opinion: “I am troubled by the complex, theoretical nature of many of the Commissioner’s arguments trying to reconcile the two regulations. Not only does this make it difficult for the court to navigate the regulatory framework, it shows that taxpayers have not been given clear, fair notice of how the regulations will affect them.”

C. Developments Post-Xilinx

Subsequent to Xilinx, the IRS issued proposed regulations that required that all costs be included in the QCSA pool, including, specifically, stock-based compensation. In July 2002, Treasury issued a notice of proposed rulemaking and a notice of a public hearing (NPRM) in regard to proposed amendments to the 1995 cost-sharing regulations. (67 Fed. Reg. 48,997 (July 29, 2002).) The preamble to the NPRM states that the proposed amendments to the 1995 cost-sharing regulations sought to clarify “that stock-based compensation must be considered in determining operating expenses under Treas. Reg. 1.482-7(d)(1).” In response to the NPRM, more than 12 parties submitted comments and four persons testified at the public hearing on Nov. 20, 2002. The comments were overwhelmingly to the effect that unrelated parties did not require a party to pay or reimburse the other party for amounts attributable to stock-based compensation.

While the Treasury responded to these comments, it did so in a way that was more or less devoid of any reference to empirical facts. Having completed the notice and comment procedure, on Aug. 25, 2003, Treasury issued the stock-award regulation. (Treas. Reg. 1.482-7A(d)(2).) In the preamble to the final stock-award regulation, Treasury claimed that the proffered examples of comparable transactions did “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.” (68 Fed. Reg. at 51,172–73 (Aug. 26, 2003).) Notwithstanding this point, Treasury claimed the stock-award regulation was intended to be wholly consistent with the arm’s-length standard, stating that “the [stock-award regulation] ha[s] as [its] focus reaching results consistent with what parties at arm’s length generally would do if they entered into cost sharing [sic] arrangements for the development of high-profit intangibles.”

The preamble did not explain in much detail why the examples offered were inapposite. One wonders whether it was really necessary for the court to deviate from the arm’s-length standard by discussing commensurate-with-income and whether the court could have focused more on the comparability of the examples being offered. The preamble mentioned the commensurate-with-income standard only once, stating that “[t]he 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.’”

Interestingly, one of the commenters had presented a real-life example of a situation involving a cost-sharing arrangement between Amylin Pharmaceuticals, Inc. and Hoechst Marion Roussel, Inc., showing that in such case the pool of shared costs did not include stock-based compensation. However, having apparently concluded that this example was less than persuasive, Treasury opted to provide its own hypothetical example of a transaction involving high-value pharmaceutical intangibles, providing that two parties enter into a QCSA to develop patentable pharmaceutical products, with one party agreeing to commit employees to the venture. The example concludes that this party “would not agree to commit employees to an arrangement that is based on the sharing of costs in order to obtain the benefit of independent exploitation rights unless the other party agrees to reimburse its share of the compensation costs of the employees.” It ultimately goes on to state that “Treasury and the IRS believe that if a significant element of that compensation consists of stock-based compensation, the party committing employees to the arrangement generally would not agree to do so on terms that ignore the stock-based compensation.”

Unlike the regulation at issue in Xilinx, which provided simply that “all costs” must be shared, the stock-award regulation explicitly required parties to a QCSA to share stock compensation costs. Further, the stock-award regulation provided two methods for measuring the value of stock-based compensation—a default rule and an elective rule. Under the default method, “the costs attributable to stock-based compensation generally are included as intangible development costs upon the exercise of the option and measured by the spread between the option strike price and the price of the underlying stock.” (68 Fed. Reg. at 51,174.)

Under the elective method, “the costs attributable to stock options are taken into account in certain cases in accordance with the ‘fair value’ of the option, as reported for financial accounting purposes either as a charge against income or in footnoted disclosures.” The elective method is available only with respect to options on stock that are publicly traded on an established U.S. securities market and is issued by a company whose financial statements are prepared in accordance with U.S. generally accepted accounting principles for the taxable year. While the preamble to the final regulations responded to comments, it did so in a way that was conclusory and not fact-based. The preamble also stated that “it has been determined that section 553(b) of the APA does not apply to these regulations.”

It is this regulation that applied to the years at issue in Altera. The preamble’s statement that Section 553(b) of the Administrative Procedure Act (APA) did not apply sets up the argument that becomes the centerpiece of Altera.

D. The Tax Court Decision in Altera

1. Background

Altera U.S. develops, manufactures, markets, and sells PLDs (programming tools) and related hardware, software, and predefined design building blocks for use in programming the PLDs. Altera U.S. entered into a master technology license agreement and a technology research and development cost-sharing agreement with a Cayman Islands subsidiary, Altera International, both concurrent agreements becoming effective May 23, 1997. While the original cost-sharing agreement treated ESOs or other stock-based compensation as part of the cost pool, the agreement was amended in 2005 to exclude such costs, in response to Xilinx. Altera U.S. did not include ESOs or other stock-based compensation in the cost pool under the cost-sharing agreement. The IRS issued a notice of deficiency with respect to tax years 2004 through 2007, asserting that those stock-based compensation costs should be included in the pool of shared costs to be borne partly by Altera International and, therefore, that Altera U.S.’s income should be increased by approximately $80 million in the aggregate.

Under the technology license agreement, Altera U.S. licensed to Altera International the right to use and exploit, everywhere except the U.S. and Canada, all of Altera U.S.’s intangible property relating to PLDs and its programming tools that existed before the R&D cost-sharing agreement (pre-cost-sharing intangible property). In exchange for the rights granted under the technology license agreement, Altera International paid royalties to Altera U.S. in each year from 1997 through 2003. As of Dec. 31, 2003, Altera International owned a fully paid-up license to use the pre-cost-sharing intangible property in its territory. Under the R&D cost-sharing agreement (in effect from May 23, 1997, through 2007), Altera U.S. and Altera International agreed to pool their respective resources to conduct research and development using the pre-cost-sharing intangible property. Profits of Altera International, from the exploitation of the technology derived from the R&D, would then be retained by Altera International.

2. APA Analysis

The taxpayer alleged that the stock-award regulation was invalid because, in its promulgation, Treasury had not complied with the procedure required by the APA.

Pursuant to Section 553 of the APA, in promulgating regulations, an agency must (1) publish a notice of proposed rulemaking in the Federal Register; (2) provide “interested persons an opportunity to participate in the rule making through submission of written data, views, or arguments with or without opportunity for oral presentation”; and (3) “[a]fter consideration of the relevant matter presented, the agency shall incorporate in the rules adopted a concise general statement of their basis and purpose.” Notably, these requirements generally apply only to legislative, and not interpretive rules, unless such requirements are explicitly called for by statute. (5 U.S.C. Section 553.) Pursuant to Section 706(2)(A) of the APA, a court must hold unlawful and set aside agency action, findings, and conclusions that the court finds to be arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.

Citing State Farm, the Tax Court noted that a court’s review under this “standard is narrow and a court is not to substitute its judgment for that of the agency.” (Altera Corp. v. Commissioner, 145 T.C. 91, 112 (2015) (hereinafter “Altera Tax Court”).) To engage in reasoned decisionmaking, “the agency must examine the relevant data and articulate a satisfactory explanation for its action, including a ‘rational connection between the facts found and the choice made,’” the Tax Court said.

The Tax Court also reviewed the Chevron rule, which requires deference to an agency that exercises its rule-making function. The standard set out in Chevron, U.S.A., Inc. v. Nat. Res. Defense Council, Inc., a 1984 U.S. Supreme Court decision, has become the foundational standard regarding regulatory authority and the standard of deference to which agencies are entitled. There are two steps in the Chevron analysis. Under Chevron step one, applying the ordinary tools of statutory construction, a court must determine “whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” (Mayo Found. for Med. Educ. and Research v. United States.)

In considering the standards of agency action set forth in both Chevron and State Farm, it bears noting that the facts involved in each case were, perhaps in contrast to the typical Treasury Regulation, largely empirical and data-driven. Chevron, for example, involved an intensely factual evaluation of environmental law in the examination of an agency’s establishment of the specific parameters of a program required by states in order to come into compliance with the 1977 Clean Air Act. The examination necessarily involved collecting sufficient data from which it would be possible to conclude whether the agency’s policy actually reduced ambient emissions. Similarly, State Farm involved an equally fact-intensive examination of the National Highway Traffic Association’s rescission of a motor vehicle occupant crash protection standard. The standard had been formulated under a statute that contemplated an empirical approach, stating that the agency must consider “relevant available motor vehicle safety data” with respect to various specific different types of vehicles, and further was required to consider “the extent to which such standards will contribute to carrying out the purposes” of the Act.

Under Chevron step two, a court must defer to the agency’s authoritative interpretation of an ambiguous statute, “unless it is ‘arbitrary or capricious in substance, or manifestly contrary to the statute.’”

i. Legislative vs. Interpretive Rule

In determining the applicable standard of review under the APA, the Tax Court first addressed the question of whether the rule was a legislative rule or an interpretive rule. The distinction is important for several reasons. For one, with respect to Treasury regulations, case law has established a direct link between the categorization of the rules at issue and the level of deference to which the rules are entitled upon taxpayer challenge. In Mayo Foundation, for example, it was determined that legislative Treasury regulations are subject to Chevron deference. Conversely, regulations not subject to this level of deference would presumably be subject only to Skidmore v. Swift & Co. deference. Under the Skidmore standard, the validity of administrative action is examined carefully, rather than automatically granted a presumption of validity. The pronouncement’s validity will depend on its “thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade.” The Skidmore standard has been applied in other instances where the challenged action does not have the force of law, such as IRS revenue rulings. (See In re Worldcom, Inc.) Indeed, for a taxpayer, a challenge under the Chevron standard is far more daunting than a challenge to which the Skidmore standard applies.

A second important distinction between interpretive and legislative rules is that while legislative rules would be entitled to Chevron deference, they would, in contrast to interpretive regulations, be subject to the notice and comment requirements of APA Section 553. Under the APA, one of the ways interpretive rules are treated differently from legislative rules is that agencies may issue the former without any public input. It was necessary to make this determination because non-legislative regulations would technically not have the force of law or otherwise be binding on the taxpayer. The court concluded that the regulation was a legislative rule by examining Section 7805(a), which provides that Treasury is authorized to “prescribe all needful rules and regulations for the enforcement of” the tax code. Because it was a legislative rule and Treasury did not find for good cause that notice and comment were impracticable, unnecessary, or contrary to the public interest, the Tax Court found that the notice requirements apply to the stock-award regulation.

Interestingly, the Tax Court’s conclusions that the stock-award regulation was a legislative regulation, and that therefore notice and comment review was mandated under APA Section 553, is at odds with language in the “special analyses” segment of the preamble, stating explicitly that it had been determined that APA Section 553(b) would not apply to the regulation. While the Tax Court did not address this in its analysis, the court was clearly aware of the inherent clash between its findings and this provision, as it cited to the applicable language in the Preamble. (Altera Tax Court.) Based on the importance of its findings as to deficiencies in the notice and comment procedure, which were central to the court’s invalidation of the stock-award regulation, one can only conclude that the Tax Court must have intended to implicitly overrule Treasury’s assertion in the Preamble.

It is important to note that there is a difference in terminology as between tax cases and APA matters with respect to the use of the terms “legislative regulations” and “interpretive regulations.” In the tax context, various tax code sections provide that Treasury shall prescribe regulations to carry out the purposes of the statute—signifying that Congress intended to delegate the task of developing the body of law loosely referenced in the statute.

Section 1502 is a classic example of such a provision. It provides that “[t]he Secretary [of the Treasury] shall prescribe such regulations as he may deem necessary in order that the tax liability of any affiliated group of corporations making a consolidated return and of each corporation in the group…may be returned…in such manner as clearly to reflect [] income tax liability…” For tax purposes, these are “legislative” regulations, under which Treasury has virtually unlimited authority to effectively write substantive law. (See generally Section 7805.) By contrast, in the case of tax statutes that do not expressly delegate this authority to Treasury, regulations promulgated under such a statute would be considered “interpretive” regulations for tax purposes. Under such interpretive regulations, Treasury and the IRS are confined to the structure and language of the statute itself. Importantly, under an APA analysis, regulations that are both legislative and interpretive under this nomenclature would be characterized as legislative and subject to the notice and comment requirements.

ii. How Should the Standard of Review Be Applied? Chevron Only? State Farm Only? Or Both?

Next, the Tax Court addressed how to apply the appropriate standard for review. Altera contended that the court should review the stock-award regulation under the State Farm standard. The government contended that the court should review the stock-award regulation under the Chevron standard of deference. The Tax Court concluded that the stock-award regulation must satisfy State Farm’s reasoned decisionmaking standard. In fact, the Tax Court goes even further by expressing its view that State Farm review is implicit under any Chevron step two analysis: “Ultimately, however, whether State Farm or Chevron supplies the standard of review is immaterial because Chevron step two incorporates the reasoned decisionmaking standard of State Farm.” (Altera Tax Court.)

According to the Tax Court, the validity of the regulation therefore turns on whether Treasury reasonably concluded that its rule was consistent with the arm’s-length standard, and that is necessarily an empirical determination based on evidence of arm’s-length transactions. Because the validity of the stock-award regulation turns on whether Treasury reasonably concluded that it is consistent with the arm’s-length standard, the regulation must satisfy State Farm’s reasoned decisionmaking standard to avoid being rendered arbitrary or capricious.

iii. Application of the Reasoned Decisionmaking Standard

The Tax Court then went on to determine whether Treasury applied an analysis consistent with the “reasoned decisionmaking” standard of State Farm. Altera U.S. contended that the regulation was invalid because (a) it lacked a basis in fact, (b) Treasury failed to rationally connect the choice it made with the facts it found, (c) Treasury failed to respond to significant comments, and (d) the regulation was contrary to the evidence before Treasury. The Tax Court agreed with all of the taxpayer’s contentions, finding that “Treasury’s ipse dixit conclusion, coupled with its failure to respond to contrary arguments resting on solid data, epitomizes arbitrary and capricious decisionmaking.” (Altera Tax Court.)

An ipse dixit (Latin for “he said it himself”) statement is a statement that by its nature is dogmatic and unproven, and amounts to an unsupported factual assertion that “this is simply the way things are.” In the context of this article, the term is often used to criticize Treasury’s attempts to claim that the stock-award regulation, which requires taxpayers to share stock-based compensation irrespective of what taxpayers in uncontrolled transactions do, nonetheless comports with the arm’s-length standard—amounting, according to critics, to a statement that the stock-award regulation produces an arm’s-length result merely “because Treasury says it does.”

Significantly, the Tax Court linked the failure to respond to significant comments to the notice and comment procedure standards of FCC v. Fox Television Stations, Inc. Under this standard, “[i]n providing a reasoned explanation for agency action that departs from an agency’s prior position the agency must ‘display awareness that it is changing position.” While the agency does not need to show that “the reasons for the new policy are better than the reasons for the old one,” providing sufficient notice to the taxpayer would necessarily entail a demonstration that a new policy had in fact been adopted.

In Fox, the U.S. Supreme Court considered the Federal Communication Commission’s policy of enforcing a broadly worded federal profanity statute in the context of isolated utterances of certain “four letter” expletives on Fox Television broadcasts. Specifically, the federal statute, 18 U.S.C. Section 1464, prohibited the broadcasting of “any…indecent…language,” including references to sexual or excretory activity or organs. While the original FCC policy in reference to the statute was that broadcasts featuring the use of offensive words that otherwise referenced such items were per se inactionable as long as such utterances were nonrepetitive, the agency changed its policy to pursue findings of indecency even in the case of fleeting, otherwise isolated use of such expletives. The Court upheld the stark policy change largely because “the FCC forthrightly acknowledged that its recent actions have broken new ground, taking account of inconsistent prior FCC and staff actions, and explicitly disavowing them as no longer good law.” The Court was thus persuaded by the agency’s self-aware and forthright announcement of the new direction it was taking.

iv. Harmless Error

Lastly, the Tax Court looked at the harmless error rule. Section 706 of the APA instructs reviewing courts to take due account of the rule of harmless error. This rule reflects the notion that if the agency’s mistake did not affect the outcome, and if it did not prejudice the petitioner, it would be senseless to vacate the agency action. The government contended that the harmless error rule should apply because: (1) Treasury had sufficient alternative reasons for adopting the regulation, and (2) in the years following Treasury’s adoption of the regulation, the Financial Accounting Standards Board, the International Accounting Standards Board, and the Organization for Economic Cooperation and Development (OECD) had adopted policy positions that concur with Treasury’s. The Tax Court determined that because it is not clear that Treasury would have adopted the regulation had it concluded that the regulation is inconsistent with the arm’s-length standard, the harmless error rule is inapplicable and could not support the validity of the regulation.

v. Decision

Thus, the Tax Court determined that the regulations were invalid because the requirements of the APA were not met. In the words of the Tax Court:

“Because the [stock-award regulation] lacks a basis in fact, Treasury failed to rationally connect the choice it made with the facts found, Treasury failed to respond to significant comments when it issued the [stock-award regulation], and Treasury’s conclusion that the [stock-award regulation] is consistent with the arm’s-length standard is contrary to all of the evidence before it, we conclude that the [stock-award regulation] fails to satisfy State Farm’s reasoned decisionmaking standard and therefore is invalid.” (Altera Tax Court.)

E. The Ninth Circuit’s Decision in Altera

On appeal, the Ninth Circuit majority revisited the two-step analysis espoused by the Tax Court. It sought to determine (1) whether the stock-award regulation complied with the State Farm standard, and (2) assuming it met the criteria of State Farm, whether the regulation passed under the standard of review provided for in Chevron.

The majority began with the State Farm standard, which, as mentioned above, seeks to determine whether the regulatory action complied with the APA by being the result of “reasoned decisionmaking,” which in turn requires a “rational connection between the facts found and the choice made.” Under this standard, the majority emphasized that the promulgation process does not have to be perfect, but rather, it will pass muster as long as the agency’s “path may reasonably be discerned.” As discussed above, this necessitated a review of the notice and comment procedure implemented at the time the regulations had been promulgated.

1. APA Analysis at the Ninth Circuit

The Ninth Circuit’s APA analysis centered on both the notice and comment requirement and the reasoned decisionmaking standard of State Farm. As to the notice and comment procedure undertaken in 2002, the parties had substantial factual disagreements. Indeed, the only aspect agreed upon was that commenters had attacked the proposed regulations by presenting evidence of cost-sharing arrangements involving unrelated parties acting at arm’s length and identifying that such parties were overwhelmingly not opting to treat stock-based compensation as costs to be shared for these purposes. The parties disagreed on (1) whether the transactions proffered by commenters were in fact comparable to the transactions underlying the Altera cost-sharing arrangement, and (2) whether Treasury’s reference to the “commensurate with income” standard in its proposed rulemaking provided sufficient notice of its intent to depart from the arm’s-length standard.

The majority ultimately concluded that (1) Treasury was reasonable to conclude that there were no comparable transactions presented by commenters at the time of the rulemaking, and (2) even if the transactions presented had been sufficiently comparable, the legislative history of Section 482 granted Treasury the authority to adopt a commensurate-with-income approach in lieu of a more traditional arm’s-length approach. Under the State Farm standard, consistent with the APA, an agency need only respond to significant comments, i.e., comments that, if adopted, would necessarily require a change in the agency’s proposed rule. Because this standard adopts an after-the-fact “outcome determinative” approach, the government’s argument was therefore that Treasury was not bound to consider evidence of comparable transactions because it had the authority to promulgate the regulations without respect to the arm’s-length standard anyway.

Under the reasoned decisionmaking analysis, the Ninth Circuit found that Treasury’s path could have been reasonably discerned by Treasury’s citations to legislative history in its NPRM and the Preamble to the stock-award regulation. Further, Treasury “set forth its understanding that it should not examine comparable transactions when they do not in fact exist, and it should instead focus on a fair and reasonable allocation of costs and income” and “relied on Congressional rejection of the primacy of the traditional arm’s length standard.” Finally, because the comments “had no bearing on ‘relevant factors’ to the rulemaking, nor any bearing on the [stock-award regulation], there was no APA violation.”

In its decision, the majority thus focused largely on the legislative history of Section 482. Looking specifically at the addition of the second sentence, the majority mentioned that the House Ways and Means Committee recommended the addition of the commensurate-with-income clause because it was attempting to assure adequate allocations to the U.S. taxable entity of income attributable to intangibles and to rectify the inconsistent results of attempting to impose an arm’s length concept in the absence of comparables.“ In other words, the majority concluded that the legislative history evidenced an intent to adopt the commensurate-with-income standard in cases where comparable transactions could not be found.

The Ninth Circuit majority sought to bolster this statement by citing to Treasury’s 1988 white paper (the “White Paper”) (Notice 88-123.), which served as an initial government reaction to the transfer pricing changes effected in the 1986 Tax Reform Act. The majority claimed that, for two main reasons, the White Paper evidenced a “dramatic shift” from the typical arm’s-length standard, as that standard had come to be understood after the promulgation of the 1968 regulations. For one, the White Paper appeared to specifically bless a standard that at least was conceptually distinct from the arm’s-length standard in that it “made clear that Treasury initially understood the commensurate with income standard to be consistent with the arm’s length standard,” the Ninth Circuit said. Second, it seemed to suggest that this standard would have to be applied frequently: “Treasury wrote that a comparability analysis must be performed where possible, but it also suggested a ‘clear and convincing evidence’ standard for comparables, suggesting that a comparability analysis would rarely suffice.”

The Ninth Circuit majority opinion also asserted that the arm’s-length standard has never been used to the exclusion of other standards, citing a pre-1986 study which found that direct comparables were located and applied in only approximately 3 percent of the IRS’s transfer pricing adjustments prior to the 1986 Tax Reform Act. The study the opinion references dates from 1981, however, long predating the 1986 statutory change and the promulgation of associated regulations.

2. Ninth Circuit Chevron Analysis

Turning to the Chevron analysis, the panel concluded that Congress had “clearly authorized” both the IRS and Treasury to determine what should and should not constitute a QCSA and that the failure to adhere to negative commentary at the time of promulgation did not render the regulations arbitrary and capricious within the Chevron and State Farm standards.

3. The Ninth Circuit Distinguishes Xilinx

Importantly, in its decision, the Ninth Circuit chose to distinguish, rather than overrule, Xilinx. The court created an interesting distinction between the two cases, stating that unlike Altera, Xilinx did not involve a debate of executive authority but rather of regulatory interpretation. Moreover, the court noted that unlike Altera, which dealt with the procedural permissibility of promulgating one distinct rule, Xilinx was faced with a substantive conflict between two distinguishable rules that were in direct conflict with one another (i.e., the requirement to account for stock-based compensation costs as a specific allocable cost versus the requirement to account for stock-based compensation the way unrelated parties would in the course of arm’s-length transactions). On the other hand, the court determined that Altera focused on whether the procedural process of disregarding commentary regarding allegedly comparable transactions, when adopting the regulations, rendered the process deficient under the APA.

4. The Dissent

Judge Kathleen O’Malley, a judge on the Federal Circuit sitting on the Ninth Circuit by designation, filed a dissenting opinion. Fundamentally, the dissent agreed with the Tax Court and argued that Treasury had not adequately explained the stock-award regulation, rendering them arbitrary and capricious under State Farm. The dissent was very detailed and thoughtfully written.

The dissent begins by quoting from SEC v. Chenery Corp., stating that “a foundational principle of administrative law [is] that a court may uphold agency action only on the grounds that the agency invoked when it took the action.” (Altera, slip. op. at 46.) Known as the “Chenery doctrine,” this principle stands for the notion that courts may not uphold an agency’s action based solely on post-hoc reasoning or rationalization. This doctrine is a departure from traditional Chevron deference and has become a hallmark of State Farm review. The dissent noted that, in promulgating the rule, Treasury repeatedly insisted that it was applying the traditional arm’s-length standard and that the resulting rule was consistent with that standard. In other words, Treasury was substituting its mandatory rule for a rule that would call for review of empirical data. The dissent noted that the majority held that Treasury’s citation to the legislative history surrounding the enactment of the Tax Reform Act of 1986 “communicated its understanding that Congress had called upon it to move away from the historical arm’s-length standard.” As a result, the dissent found that the majority impermissibly shifted gears, supplying a reasoned basis for the agency’s action that the agency itself had never given.

The dissent then provided a detailed analysis leading to its primary findings—that (a) the revised regulations were invalid under State Farm; and (b) Section 482, by itself and in the absence of regulations, does not require sharing of stock-based compensation costs.

The dissent pointed out that, since the 1930s, the regulations under Section 482 have provided consistently that “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.” The dissent also observed that the 1986 commensurate-with-income amendment did not “dislodge” the arm’s-length test, but merely addressed a “recurring problem” with transfers of highly valuable intangible property for which comparable transactions were absent. To address the gap, the dissent maintained, Congress found it “appropriate to require that the payment made on a transfer of intangibles to a related foreign corporation…be commensurate with the income attributable to the intangible.”

In this vein, the dissent made the interesting, and perhaps crucial, observation that the 1986 statutory amendment may not be relevant at all to QCSAs because, by their nature, such arrangements do not evidence an actual transfer of an intangible, but rather a joint agreement to share in the future development of an intangible and thereby earn the right to share in the profits generated by use of such intangible. In other words, QCSAs do not involve a transfer of intangible property because it is only in the technology developed as a result of the QCSA that the foreign affiliate is entitled to an interest. Thus, there appears to be at least an argument that only the “platform contribution” (buy-in) aspects of QCSAs are implicated here, because only the buy-in portion of such agreements could be indisputably characterized as “transfers.”

The dissent noted that, in the White Paper, Treasury explained that “[o]nly ‘in situations in which comparables do not exist’ would the commensurate with income standard apply.” The dissent expressed its view that Treasury thus failed to meet the “reasoned explanation” requirement of State Farm, explaining:

The APA’s safeguards ensure that those regulated do not have to guess at the regulator’s reasoning; just as importantly, they afford regulated parties a meaningful opportunity to respond to that reasoning. Treasury’s notice of proposed rulemaking [with respect to the allocation of stock-based compensation] ran afoul of these safeguards by failing to put the relevant public on notice of its intention to depart from traditional arm’s length analysis.

The dissent also found that the majority improperly distinguished Xilinx from the case at bar because Xilinx should control as to the primary issues, which the dissent considered to include the question as to whether Treas. Reg. 1.482-1(b)(1) superseded Treas. Reg. 1.482-7A(d)(2). Because Xilinx controlled, it can be inferred that the most relevant conclusions to the dissent in Altera were that (1) the substantive factual determination on the question of whether unrelated parties share the cost of stock-based compensation, finding that such parties would not do so, and (2) Treasury was therefore required to at least attempt to gather empirical evidence as to uncontrolled comparable transactions before making a blanket conclusion that no such evidence was available.

Significantly, in reference to the harmless error standard under State Farm, the dissent also addressed the majority’s conclusion that Treasury was entitled to ignore comments opposing the regulation and evidence of purportedly comparable transactions because “they did not bear on the agency’s consideration of the relevant factors.” In the words of Judge O’Malley:

“I do not share the majority’s view. Treasury may well have believed that, given the fundamental characteristics of stock-based compensation in QCSAs, it could dispense with arm’s length analysis entirely. Cf. Xilinx II, 598 F.3d at 1197 (Fisher, J., concurring) (hypothesizing why unrelated companies may not share stock-based compensation costs). But the APA required Treasury to say that it was taking this position, which departed starkly from Treasury’s previous regulations.”

The dissent stated that, rather than providing notice to taxpayers that it was changing its position, Treasury simply indicated that there were no comparable transactions to be found by stating in the Preamble to the regulations that: “While the results actually realized in similar transactions under similar circumstances ordinarily provide significant evidence in determining whether a controlled transaction meets the arm’s length standard, in the case of QCSAs such data may not be available.” (68 Fed. Reg., at 51,172–73 (Aug. 26, 2003).) Judge O’Malley pointed out that requiring Treasury to “show its work” in the Preamble was not “excessive proceduralism” as the majority suggested, but rather was the essence of the reasoning that State Farm demands. (Altera, slip. op. at 62.) Moreover, Judge O’Malley said that, even if Xilinx did not control, she would hold that related parties in QCSAs need not share costs associated with stock-based compensation.

The dissenting opinion also disagrees with the majority’s conclusion that Treasury’s reading of Section 482 satisfied the second step of the Chevron test. Judge O’Malley did not believe that a QCSA evidenced a bona fide transfer and found that the 1986 legislative history reviewed extensively by the majority was therefore irrelevant, going on to state that the Commissioner’s argument that the commensurate-with-income standard applies to “intangible transactions in general, and cost-sharing arrangements in particular,” is inconsistent with the plain language of the statute. The only reasonable interpretation of Section 482, therefore, the dissent said, was that the commensurate-with-income standard does not apply to QCSAs. The dissent thus found that Treasury failed both steps of the process (i.e., the APA analysis and the analysis under Chevron) when it needed to satisfy both.

5. Analysis of Ninth Circuit Decision

While the Ninth Circuit majority presented an interesting analysis of the history of transfer pricing rules, its reliance on the legislative history associated with the 1986 addition of the second sentence of Section 482 seems misguided. As the appellee pointed out in its brief, “legislative history…is [] often murky, ambiguous, and contradictory” and has been described as “an exercise in looking over a crowd and picking out your friends.” (Altera brief at 65.) Although it is indeed common for courts to turn to legislative history to attempt to glean the legislative intent and theory behind statutes, tax statutes in particular, the inherent murkiness of the Section 482 legislative history here is compounded by the distinct possibility that the court focused on the history of the wrong rule. As the dissent pointed out, the second sentence to the statute, by its terms, applies only to a “transfer or a license” of intangible property, when it is largely understood that a cost-sharing agreement evidences no such transfer.

Whether a true transfer of intangible property has occurred carries meaningful and often adverse tax consequences for U.S. taxpayers. See generally Section 367(d), providing that if a U.S. person transfers certain intangible property to a foreign corporation in an otherwise tax-free transaction, the U.S. transferor will take into account deemed royalty income over the course of the intangible property’s useful life. Indeed, the purpose of QCSAs in many instances is to avoid such consequences or other consequences that would cause immediate gain recognition. It does not follow, therefore, that Congress would have intended a broad interpretation of the term “transfer” in the QCSA context, as such a reading would render inconsistent many established provisions dealing with intangible property in international structures.

Treasury itself confirmed this point in the White Paper, stating that “[t]he [1986] legislative history [to Section 482] envisions the use of bona fide research and development cost sharing arrangements as an appropriate method of attributing the ownership of intangibles ab initio to the user of the intangible,…thus avoiding section 482 transfer pricing issues related to the licensing or other transfer of intangibles.” (White Paper at 464.)

The Ninth Circuit’s attempt to synthesize the legislative history seems internally inconsistent. On the one hand, it suggested that any failure to consider comparable transactions under an arm’s-length standard analysis would have been harmless error. The decision suggests that the error is harmless because Treasury would have been free to completely disregard such evidence by instead adopting a commensurate-with-income approach to allocate costs under a QCSA, claiming that “despite the asserted focus on comparability, the arm’s length standard has never been used to the exclusion of other, more flexible approaches.” However, the Ninth Circuit simultaneously points out that “Congress intended the commensurate with income standard to displace a comparability analysis where comparable transactions cannot be found.” These two statements are extremely difficult to reconcile—either Treasury was required to look for comparable transactions, or it was not.

Other aspects of the Ninth Circuit’s analysis seem to suffer from similar inconsistencies. The court noted, for example, that “[t]he Tax Reform Act of 1986 reflected Congress’s view that strict adherence to the arm’s length standard prevented tax parity.” Contrary to this suggestion, the Ninth Circuit itself held in Xilinx that strict adherence to the arm’s-length standard is in fact the only way to ensure tax parity. Xilinx had gone on to hold that the parity principle not only required such strict adherence to the arm’s-length standard, but it was also the essence of the transfer pricing regulations as a whole. In Xilinx, the Ninth Circuit pointed out that if unrelated parties in comparable transactions could deduct costs related to ESOs, parity is not achieved unless the taxpayer could deduct such costs as well.

As mentioned above, the Ninth Circuit refers to a study concluding that direct comparables were located and applied in only approximately 3 percent of the IRS’s transfer pricing adjustments prior to the 1986 Tax Reform Act. The study the opinion references dates from 1981, however, long predating the 1986 statutory change and the promulgation of associated regulations. The Ninth Circuit does not explain why the extent of the cases in which direct comparables were ultimately utilized tells one anything about the statutory requirement calling for taxpayers to exercise diligence in searching for such comparables, nor does it account for market factors that would invariably reduce the probative value of such a study over time even if its findings were to be taken at face value. If such dated evidence is relevant for anything, such relevance should be confined to the regulations promulgated by Treasury in 1994. The current stock-award regulation simply cannot be justified by reliance on a study that occurred more than a decade before Treasury promulgated its first set of regulations addressing stock-based compensation.

Indeed, the premise of the study itself seems inherently questionable. Stated another way, if only 3 percent of all transactions are comparable, 97 percent of transactions are therefore incomparable. This seems at odds with the notion that businesses are not like fingerprints, and rarely generate “unicorns,” at least for long. More often, competitors develop distinguishing approaches that can be adjusted for. In the context of a supply and full value chain, the so-called first movers for whom comparables would not exist eventually experience competitors and comparables begin to be developed. Thus, belief in a perpetual lack of comparable transactions within an industry or even a segment thereof necessarily requires believing in unicorns. With this in mind, one would need to account for the effect of this trend towards competition on the study’s 1981 findings, even if one were to accept such findings at face value.

Moreover, the 3 percent figure must be viewed in the historical context of 1981—a pre-internet time period characterized by a lack of available information and electronic (pay for) databases that could conceivably capture such information. Second, the domestic economy itself was less mature and robust in 1981, with much less competition and thus far fewer transactions to be compared. Third, the world was not yet a globalized competitive economy. The OECD, EU, and Treasury—perhaps due to a lack of resources, a lack of training, or both—appear increasingly to be resigning themselves to the notion that no comparables exist instead of doing the hard work required by the arm’s-length standard.

In the dissent, Judge O’Malley states that Treasury did not provide sufficient notice under the APA of its intent to switch policies from an arm’s-length approach characterized by the search for comparables, to what would amount to a “mandatory” commensurate-with-income approach, under which Treasury had the regulatory authority to forgo an arm’s-length analysis altogether—a point that is well made. The reasoned decisionmaking standard under State Farm would appear to require that, if Treasury intended to change its position as to the role of the arm’s-length standard in this analysis, it would have to put taxpayers on notice that it was in fact doing just that by “saying so.” Judge O’Malley makes this point eloquently in the dissent: “The APA’s safeguards ensure that those regulated do not have to guess at the regulator’s reasoning; just as importantly, they afford regulated parties a meaningful opportunity to respond to that reasoning. Treasury’s notice of proposed rulemaking ran afoul of these safeguards by failing to put the relevant public on notice of its intention to depart from traditional arm’s length analysis.”

In the opinion, the Ninth Circuit even seems to implicitly acknowledge the gravity of this departure, citing “Congressional rejection of the primacy of the arm’s length standard” and basing conclusions on its finding that “Treasury communicated its understanding that Congress had called upon it to move away from the traditional arm’s length standard.” These statements make clear, at the very least, that the Ninth Circuit majority recognized that what it was doing was either moving away from, or rejecting the primacy of, the traditional arm’s-length standard. The Ninth Circuit did not attempt to reconcile allowing Treasury to base its rulemaking on the conclusion that there were no comparables during the notice and comment period, while simultaneously determining that Treasury never needed to consider comparables in the first place.

F. Good Fortune Shipping v. Commissioner

Although not within the Altera-Xilinx line of cases, Good Fortune Shipping SA v. Commissioner, decided on July 27, 2018, by the U.S. Court of Appeals for the District of Columbia Circuit, is also relevant to the analysis and may provide a preview of where future challenges, including “Altera revisited,” are headed. In Good Fortune Shipping, a taxpayer mounted a successful challenge to the validity of Treasury regulations, causing the D.C. Circuit to overturn a prior Tax Court decision in favor of the government. The case dealt with the validity of Treasury regulations under Section 883 that would have denied the taxpayer, a Marshall Islands-based shipper, an exemption from U.S. tax for which it would have otherwise qualified, merely because its equity was held in the form of bearer shares rather than registered shares. During the notice and comment process with respect to this regulation, Treasury had stated that the regulation’s exclusion of bearer stock from the implied statutory definition of “stock” was based on the inherent difficulty in tracking ownership of bearer shares—an exercise that was necessary to determine that a taxpayer claiming the exemption was owned by stockholders having the requisite residency.

Section 883(a)(1) exempts from U.S. taxation certain international shipping income of a foreign corporation provided the foreign corporation was both (1) organized in a jurisdiction that grants an equivalent exemption for U.S. shipping corporations and (2) at least 50 percent owned by individuals who are residents of such a jurisdiction. The rule appears to be aimed at preventing foreign persons in jurisdictions that would not otherwise qualify for the exemption from trying to sidestep the applicable requirements by setting up a mere paper company in a third-party jurisdiction to which its shareholders otherwise have no connection.

The D.C. Circuit court, however, found that even though Treasury’s stated concern underlying the rule was legitimate, the regulation did not survive upon application of step two of the Chevron analysis because Treasury did not provide sufficient explanation for its per se exclusion of bearer shares. Specifically, it stated that “[e]ven if [the statute] grants the IRS significant discretion to establish how to prove ownership [of a company seeking exemption thereunder], it hardly authorizes the agency to categorically deny consideration of a recognized form of ownership based on only a single, undeveloped statement that it is ‘difficult[]’ to reliably track the location of a given owner.” Emphasizing that other forms of equity often create the same types of transparency issues as bearer shares without being categorically excluded from consideration as a valid form of ownership, the court went on to state that “[t]he IRS’s interpretation [of the statute as evidenced by the regulations] … appears to [impermissibly] rewrite [the statute] to require not only valid ownership, but ownership that is not ‘difficult’ to track.”

The rigid application of Chevron step two in this case may be a sign of a trend that could act as a warning call to Treasury: namely, the substance of the State Farm reasoned decisionmaking standard is actually lurking within Chevron step two, and may continue to be invoked in taxpayer challenges of Treasury regulations. We note that in the now “revived” Altera Tax Court case, Judge Marvel had found that Chevron step two encompassed an analysis that was the functional equivalent of the State Farm reasoned decisionmaking standard.

Peter J. Connors is a tax partner in the New York office of Orrick, Herrington & Sutcliffe LLP. Barbara S. de Marigny is a tax partner and Michael R. Rodgers is a senior tax associate in the Houston office of Orrick, Herrington & Sutcliffe LLP. The authors also extend special thanks to Professor William H. Byrnes of the Texas A&M University School of Law for his helpful thoughts, comments, and suggestions.