Cost-Sharing Agreements and the Arm’s-Length Standard

By Brett Wilkinson, Ph.D., and Tracy J. Noga, Ph.D., CPA



  • In 2005, the Tax Court held that under the Regs. Sec. 1.482-1 arm’s-length requirement, Xilinx, Inc., was not required to share the costs associated with employee stock options it issued to its U.S. employees with an Irish subsidiary with which it had a cost-sharing agreement.
  • In 2009, on appeal, the Ninth Circuit reversed the Tax Court and held that under Regs. Sec. 1.482-7, Xilinx was required to share the costs with the subsidiary, thus reducing Xilinx’s deductible expenses and increasing its tax liability.
  • Faced with considerable criticism of its holding in the case, in 2010 the Ninth Circuit withdrew its first opinion. In its second opinion, it affirmed the Tax Court’s decision that under the arm’s-length requirement, Xilinx did not have to share the stock option costs with its subsidiary.

In 2009, the Ninth Circuit Court of Appeals ignited a firestorm with its Xilinx decision. 1 At issue in the case was whether related parties in a cost-sharing agreement should be required to share the costs of employee stock options. Although the court agreed that unrelated parties would not share such costs, it concluded that the more specific cost-sharing provision in the regulations (which required parties to share “all costs”) trumped the general arm’s-length requirement. That is, although the court accepted that sharing the stock option costs would be inconsistent with the arm’s-length standard, there were instances in which the law imposed a requirement that went beyond arm’s length. Not surprisingly, the position that the arm’s-length standard could be overridden was highly controversial. 2

Following the public outcry over its initial decision, the court withdrew the decision in early 2010. On March 22, 2010, the Ninth Circuit issued a new decision in which it reversed its prior holding in Xilinx. 3 The new decision affirms the Tax Court’s position that unrelated parties to cost-sharing agreements should not be forced to include the cost of stock options in cost-sharing agreements because unrelated parties do not share such costs. 4

Despite this reversal, there is still uncertainty around this issue for several reasons. First, the court was divided in its decision, and it is not clear that this is the final word on stock options and cost-sharing agreements. In fact, the comments of the dissenting judge suggest some possible grounds for appeal. In his dissenting opinion, Judge Reinhardt argued that although there may be an error in the law, the courts are given only the task of interpreting the law and not correcting perceived inequities or errors. 5 The latter role belongs to congressional lawmakers and not to the courts. All three judges appeared to acknowledge inconsistency in the law but resolved this in different ways. It would therefore seem that there is a high likelihood of further legal review to resolve this inconsistency.

Second, if the current decision stands, it calls into question the validity of the revised cost-sharing regulations that explicitly require stock option costs to be shared. Several commentators raised this concern after the initial Tax Court decision 6 and questioned whether Treasury can require a result that has been shown to be inconsistent with arm’s-length behavior. If anything, the Ninth Circuit’s affirmation likely exacerbates this concern. Noting the conflict in the law, Judge Fisher of the Ninth Circuit stated in a footnote to his concurring opinion that “[i]t is an open question whether these flaws have been addressed in the new regulations Treasury issued after the tax years at issue in this case.” 7 In effect, a question remains as to whether Treasury will issue new or revised regulations and whether the regulations as they exist would withstand a challenge.

Tax Court

The Xilinx controversy centers on the cost-sharing agreement between Xilinx Incorporated (Xilinx) and its Irish subsidiary, Xilinx Ireland (XI).The parties agreed to split both direct and indirect costs involved in the research and development process in a ratio that reflected the reasonably anticipated benefits that each party would derive from the resulting intangible property. With respect to the employees engaged in the active research process, the parties shared a variety of salary and payroll expenses but chose not to share any costs related to the stock options made available to Xilinx’s U.S. employees. In effect, this permitted the entire cost of the options to be deducted in the United States, despite the fact that the resultant technology was shared between Xilinx and XI.

In support of its position, Xilinx contended that unrelated parties that engage in cost-sharing agreements do not share the cost of stock options. The Tax Court considered two possible approaches to valuing options: the spread between the exercise price and the market price on the exercise date, and the grant date valuation method. With respect to the former, Xilinx’s argument centered on the fact that unrelated parties would not be willing to share a cost that, as the court noted, “could rise and fall in line with the vicissitudes and vagaries of the market.” 8 Furthermore, Xilinx noted that any such sharing would introduce inappropriate incentives between unrelated parties, because the party sharing in the cost would benefit from lowering the subsequent stock price of the party that issued the options and thus the proportion of the cost in which it must share.

With regard to the second valuation methodology, Xilinx provided “uncontradicted evidence [that] established that in determining cost allocations unrelated parties would not include any cost related to the issuance of ESOs.” 9 Xilinx further asserted that there was no cash outlay associated with the options that were issued and that any cost was in fact borne by existing shareholders due to the dilution of their ownership interests.

In contrast to Xilinx’s position, the IRS asserted that it was not necessary to demonstrate comparability with independent parties. The IRS alleged that although independent parties may not explicitly share the cost of stock options, these costs are implicitly factored in during the negotiation process; however, the IRS adduced no evidence to prove this position.

In reaching its decision in favor of the taxpayer, the Tax Court held that the behavior of independent parties was conclusive. It noted that it was not essential for the IRS to have actual knowledge of the behavior of uncontrolled parties, but that where the behavior of those parties can be shown to be inconsistent with the IRS’s determination, such a determination is arbitrary, capricious, or unreasonable. The fact that both Xilinx and the IRS agreed that uncontrolled parties do not share option costs, and the IRS was unable to provide evidence of implicit cost sharing, was compelling evidence that the inclusion of such costs was inconsistent with the arm’s-length standard. The court rejected the IRS’s contention that the regulations effectively supplanted the role of independent comparisons by defining arm’s length as including “all costs.” Instead, the court concluded that the intent of the regulations is that the arm’s-length standard applies in all cases, including cost-sharing agreements. Thus, the Tax Court observed that “if unrelated parties would not share the spread or the grant date value, [the IRS’s] determinations are arbitrary and capricious.” 10

Ninth Circuit’s Initial Decision

Following its defeat in the Tax Court, the IRS appealed the Xilinx case to the Ninth Circuit. Interestingly, in its now withdrawn decision, the Ninth Circuit rejected the positions of both Xilinx and the IRS and determined that the cost-sharing regulations and the more general arm’s-length requirement embodied in the regulations were irreconcilable. 11 That is, the court expressly rejected both parties’ “attempts to harmonize the two provisions” and determined instead that the more specific provision (in this case the cost-sharing regulation) dominated the more general provision (the arm’s-length requirement).

In arriving at its original decision, the Ninth Circuit noted that Regs. Sec. 1.482-1 expressly requires that “the standard to be applied in every case is that of a taxpayer dealing at arm’s length with an uncontrolled taxpayer.” 12 The court concluded that there could be no exceptions within the arm’s-length context whereby a transaction could override what would occur between independent parties and still remain arm’s length. For that reason, the court determined that the “all costs” requirement of Regs. Sec. 1.482-7, which necessitated the sharing of costs of any form regardless of whether uncontrolled parties would share such costs, was inconsistent with Regs. Sec. 1.482-1. The court rejected the IRS’s position that arm’s length was defined in the case of cost sharing as a sharing of the totality of costs. Rather, it found the two provisions to be fundamentally at odds and, in light of such irreconcilability, determined that the more specific of the two provisions (the cost-sharing regulation) controlled.

In reversing the Tax Court’s decision, the now withdrawn Ninth Circuit opinion generated much controversy, which was concerned less with the decision to require Xilinx to share the stock option costs with its Irish subsidiary and more with the basis on which the decision rested. This is because the Ninth Circuit’s position represented a dramatic shift away from the centrality of the arm’s-length standard and allowed for a more formulaic approach to transfer pricing. In response to the overwhelming concerns expressed by the members of the tax community, the Ninth Circuit withdrew its decision in January 2010, adding further to the uncertainty surrounding the stock option issue, including the validity of Regs. Sec. 1.482-7 and how the Tax Court would apply them.

Ninth Circuit’s Revised Decision

On March 22, 2010, the Ninth Circuit released a new decision, 13 which affirmed the Tax Court. In a reversal of its withdrawn decision, the court concluded that because independent parties to a cost-sharing agreement would not share stock option costs, any IRS attempt to require such sharing was arbitrary and capricious. The court again pointed out the conflict between the requirement to apply the arm’s-length standard in all cases and the all costs requirement of the cost-sharing regulations. However, rather than relying on the canon of construction that formed the basis of its original decision—namely that the more specific provision dominates the more general—the court determined to resolve the ambiguity by considering the dominant purpose of the regulations. It ruled that the purpose of the transfer pricing regulations was that of parity between controlled and uncontrolled taxpayers, and thus the behaviors of independent parties were fundamental.

Although the new decision of the Ninth Circuit appears to bring some relief to taxpayers, several interesting aspects arise from the decision. First, the court was divided 2–1 in its decision, and the dissenting judge continued to assert the importance of relying on the canon that the more specific provision governs the more general one. This seemingly has far-reaching implications for the revised regulations that essentially provide a more specific requirement to include stock options in cost-sharing agreements, but it may still be interpreted as being at odds with the general arm’s-length provision.

Second, it is interesting that Judge Fisher, who originally supported this position but retreated from it in the revised decision, seemed to attribute some of the change in his position to the IRS’s reaction to the now withdrawn decision. He stated:

I no longer share Judge Reinhardt’s confidence in that resolution [that the more specific provision controls] because the Commissioner’s response to Xilinx’s petition for rehearing declined to fully endorse its reasoning. Instead, the thrust of the Commissioner’s response was that our result was correct, even though our reasoning was not. 14

This attracted an interesting response from the dissenting judge, Judge Reinhardt, who stated that

I guess I am just not as sensitive as Judge Fisher. Simply because the Commissioner advanced an argument that we reject, but then argued that if we reject it, we should apply the rule that we held applicable in our opinion is hardly a reason for abandoning the rule that we believed to be correct. 15

These comments should raise some concern among the tax community. In essence, there appears to be some lingering uncertainty in how the court resolved the conflict that it identified. It seems likely that a further review will be needed to remove any uncertainty.

Is Excluding Options Really Consistent with an Arm’s-Length Result?

A final concern seems to be whether Treasury in fact had a valid point that unrelated parties would implicitly factor into their agreements the costs of stock options. There seem to be inherent problems associated with determining the costs to be shared; for example, the Tax Court noted the perverse incentive it would create for one firm to seek to lower the stock price of the other. On the other hand, an OECD report on stock options 16 notes that cost-sharing agreements are largely predicated on the sharing of expected benefits in a ratio consistent with the value contributed (i.e., the costs). Thus, approaching the question from the other angle is revealing. The OECD report acknowledges this problem but also aptly notes that

Entity A would probably not enter into an arrangement in which it contributes all of the services in developing the intangible, where a significant part of the contributed compensation for such services is in the form of stock options, if Entity A received only 40% of the anticipated benefits from the arrangement (unless Entity B omitted an equally valuable element of its contribution). 17

This seems to be the crux of the issue: Even if the options represent a noncash cost to one of the parties, the labor (which is compensated by such options) delivers something of value for which the contributing party likely expects an increased share of the return from the resulting jointly owned intangible.


The Xilinx case has created turmoil in the international tax community for quite some time, and there is no clear end in sight. Although the Ninth Circuit reversed its controversial decision, uncertainty surrounds the validity of Treasury’s new regulations explicitly requiring that cost sharing include stock options. In addition, there seems to be a logical disconnect between the claim that independent parties would not share stock option costs (because such sharing is problematic) and the high likelihood that independent parties would take into account the value of their respective labor contributions in determining the ownership shares in the intangible being created.

An interesting aside is that the controversy over Xilinx adds fuel to the fire of opponents of the arm’s-length standard. Taxpayers and tax authorities alike spend vast amounts of time and resources establishing positions in transfer pricing cases. The standards are often challenging due to a lack of comparable, observable transactions, and judgment and interpretation are required. Thus, opponents argue that moving toward definitive and objective rules such as formulary apportionment may be the best path to take. 18 Not only might this reduce uncertainty, but it might arguably reduce both the compliance and administrative costs of transfer pricing. If Treasury regulations are permitted to override the arm’s-length standard by expressly requiring that stock option costs be included, this might be interpreted as yet another step toward a formula-based approach.


1 Xilinx, 567 F.3d 482 (9th Cir. 2009).

2 See, e.g., Kohl, “Clear Reflections on How the Ninth Circuit Got Xilinx Wrong,” 124 Tax Notes 259 (July 20, 2009).

3 Xilinx, 598 F.3d 1191 (9th Cir. 2010), aff’g 125 T.C. 37 (2005), acq. in result only AOD 2010-03 (7/29/10).

4 Xilinx, 125 T.C. 37 (2005).

5 Xilinx, 598 F.3d 1191, slip op. at 4616 (Reinhardt, J., dissenting).

6 See, e.g., Levy and Wrappe, Transfer Pricing: Rules, Compliance and Controversy (CCH 2007).

7 Xilinx, 598 F.3d 1191, slip op. at 4613, n. 4.

8 Xilinx, 125 T.C. 37, slip op. at 37.

9 Id. at 38.

10 Id. at 33.

11 Xilinx, 567 F.3d 482 (9th Cir. 2009).

12 Id., slip op. at 6161, quoting Regs. Sec. 1.482-1(b)(1).

13 Xilinx, 598 F.3d 1191 (9th Cir. 2010).

14 Id., slip op. at 4612, n. 3 (emphasis in original).

15 Id., slip op. at 4616, n. 2 (Reinhardt, J., dissenting).

16 Organisation for Economic Co-operation and Development, Employee Stock Option Plans: Impact on Transfer Pricing (September 3, 2004).

17 Id. at 55 (emphasis added).

18 See Avi-Yonah, Between Formulary Apportionment and the OECD Guidelines: A Proposal for Reconciliation. Working Paper (University of Michigan Law School 2009).


Brett Wilkinson is an associate professor and Roderick L. Holmes Chair of Accountancy at Baylor University in Waco, TX. Tracy Noga is an assistant professor in the Department of Accountancy at Bentley University in Waltham, MA. For more information about this article, contact Prof. Noga at

Tax Insider Articles


Business meal deductions after the TCJA

This article discusses the history of the deduction of business meal expenses and the new rules under the TCJA and the regulations and provides a framework for documenting and substantiating the deduction.


Quirks spurred by COVID-19 tax relief

This article discusses some procedural and administrative quirks that have emerged with the new tax legislative, regulatory, and procedural guidance related to COVID-19.