Like many other judicial attempts to apply the arm’s-length principle to employee stock options, the Irish Tax Appeals Commission’s decision on the transfer pricing treatment of options issued by a foreign parent is seriously flawed.
In its February 21 decision (“determination” in local parlance) in case 59TACD2024, Ireland’s administrative tax appeals body faced a variation of an issue that has become a transcontinental source of tax disputes: the transfer pricing treatment of employee stock options. Neither party to the dispute, which featured an unidentified domestic subsidiary of an anonymous foreign (apparently U.S.-based) software development multinational, disputed the selection of the transactional net margin method (TNMM) as the best method, the use of the Irish subsidiary as the tested party, or the choice of the net cost-plus markup as the profit-level indicator. However, they sharply differed on the proper composition of the services cost base.
According to Ireland’s revenue commissioners, stock options that were issued by the foreign parent to the local subsidiary’s employees and reported in the subsidiary’s financial statements under Financial Reporting Standard 102 (FRS 102) should have been charged out to the U.S. parent at cost plus a 10 percent markup. The taxpayer countered that the true economic costs associated with the employee stock options were attributable to the foreign parent, and thus were properly excluded from the cost base.
The case is similar in some respects to the taxpayer’s challenge of section 482’s controlled services regulations in Abbott Laboratories v. Commissioner, No. 20227-23. In both cases, the taxpayer argued that it cannot be compelled to charge out stock option costs under the arm’s-length principle. But in Abbott Labs, just as in Altera Corp. v. Commissioner, 145 T.C. 91 (2015), rev’d, 926 F.3d 1061 (9th Cir. 2019), and Xilinx Inc. v. Commissioner, 125 T.C. No. 4 (2005), aff’d, 598 F.3d 1191 (9th Cir. 2010), before it, the issuer was also the employer of the stock option recipients. Because the U.S. parent in each of these cases was both the issuer of the stock options and the entity performing the activities to which they relate, the attribution of stock option expense between issuer and employer does not arise.
The better international comparable is probably Kontera Technologies Ltd. v. Assessing Officer Tel Aviv (CA 943/16) and Finisar Israel Ltd. v. Assessing Officer Rehovot (CA 1728/16), a single case that was consolidated. In both Kontera and Ireland’s case 59TACD2024, the issuer of the stock options was a publicly traded U.S. parent and the recipients were employees of a local cost-plus service provider. In Kontera, Israel’s Supreme Court held that the foreign parent’s stock option grants had to be included with other employee compensation costs and charged out at cost plus a markup.
The stakes in the dispute are high. Ireland is a major host to the subsidiaries of U.S. multinationals, which rely on employee stock options to a far greater extent than their international counterparts. If the stock option costs incurred by a U.S. parent are attributable to a controlled Irish cost-plus service provider, then the service provider’s taxable income in Ireland would be increased by the amount of the markup. It would also be increased by the stock option expense itself: Unlike in the United States, Ireland offers no tax deduction for employee stock options.
The decision, written by Tax Appeals Commissioner Claire Millrine, is commendable for its detailed examination of what can be a head-spinning issue under a legislative scheme that didn’t exist before 2010. Transfer pricing litigation is still in its infancy in Ireland. Ideally, a tax tribunal applying a relatively new transfer pricing scheme would get a few warm-up cases in which the parties haggle about individual comparables before having to tackle cross-border employee stock options. Despite the lack of experience, Millrine’s decision compares favorably to many of those written by U.S. Tax Court judges who have dealt with section 482 cases for years.
Nevertheless, the decision’s faulty analysis and incorrect result mean that the Irish judiciary (including quasi-administrative appellate forums) is now officially 0-1 in transfer pricing cases.
Someone Else’s Cost
The relevant facts in case 59TACD2024 are straightforward. The Irish subsidiary’s exclusive role within the group was to provide sales and marketing and research and development services on a cost-plus basis to its foreign parent. Because the TNMM was the selected method, the subsidiary was the tested party, and the net cost-plus markup was the profit-level indicator, the subsidiary’s services costs dictated the arm’s-length charge for its services.
Although the Irish subsidiary’s employees were otherwise paid directly by their employer, they were also eligible to receive options (subject to vesting and other conditions) to purchase the publicly traded foreign parent’s shares. As one of the taxpayer’s witnesses testified, these employee stock options accounted for “a significant portion of the overall remuneration” of the subsidiary’s employees. The subsidiary reported these costs as employee compensation on its financial statements in accordance with FRS 102, which had the effect of consistently generating an accounting loss.
However, the subsidiary had a minimal role in determining its employees’ stock option awards. Those decisions were principally made by the parent company’s compensation committee, which attempts to balance the cost to shareholders of stock dilution against the need to attract and retain skilled employees. The parent company was also responsible for all plan administration functions, including the maintenance of an online brokerage platform that employees could use to cash in their options.
Stock options granted to employees of the Irish subsidiary weren’t tethered to the subsidiary. Employees of group entities often moved across different locations and employers within the group’s geographically fluid reporting structure, and they carried their stock options with them whenever they did.
Because the options were issued directly by the parent to the subsidiary’s employees, the subsidiary never held them or took on the market volatility risk associated with doing so. It also didn’t compensate the parent in any way for issuing stock options to its employees, and their value was expressly excluded from the cost base under the parties’ intercompany services contract.
According to Millrine’s decision, these features of the arrangement demonstrated that the stock options were economically attributable to the parent and excludable from the Irish subsidiary’s services costs:
The Commissioner considers that all of the evidence including the expert evidence concludes that the risk and the cost of issuing the [stock-based awards, SBAs] lies with the parent company, not with the Appellant. The Commissioner has set out the parent company’s role in issuing the SBAs in detail . . . and that it is the sole decision maker in relation to an award of the SBAs across the organisation, the administration of the infrastructure to manage awards of the SBAs and being a party to the SBAs agreement with eligible employees.
However, the analysis underlying this assessment is flawed. Much of Millrine’s decision seeks to discredit the revenue commissioners’ reliance on accounting standards, which required that the subsidiary report the stock option expense as its own in its financial statements. The decision repeatedly suggests that the accounting treatment has no relevance whatsoever to the question, referring to the stock options as mere “notional costs” of the subsidiary and describing their fictitious nature as “glaringly obvious.” Millrine was right to emphasize that accounting rules cannot dictate transfer pricing cost characterization, but the factors that she considered important were even less relevant.
Factors that figured prominently in Millrine’s decision included the foreign parent’s decisionmaking authority over employee stock option awards, its administration of the stock option plan, and its initial assumption of risk. But the decisionmaking power wielded by any ultimate parent company over its subsidiaries typically isn’t relevant under the arm’s-length principle, which is based on the counterfactual assumption of independence. Issuing employee stock options was distinct from the parent’s administration of a stock option plan, which could be accounted for by a separate service charge. The parent’s initial assumption of the costs and risks just begs the question of whether they should have been transferred to the subsidiary as part of the arrangement.
Other considerations that led Millrine astray included the exclusion of employee stock options under the parties’ intercompany services agreement and the group’s geographically flexible reporting lines and responsibilities. But associated enterprises cannot contractually opt out of their obligation to transact on arm’s-length terms, and arbitrary cost exclusions are of no use in determining what an arm’s-length result would have been. As for the group’s geographically fluid operations, which would have the same implications for base salaries as for stock options, it should have been accounted for by identifying the beneficiaries and charging them an arm’s-length service fee. Leaving the stock option costs with the parent on this basis implies that it was the beneficiary of its subsidiaries’ cross-territorial activities. This is clearly contrary to established transfer pricing principles.
But more than anything else, the notion endorsed by Millrine that a comparably situated parent company would agree to eat a sizeable chunk of its subsidiary’s employee compensation costs solely for its own benefit simply does not compute within the separate-entity framework required by the arm’s-length principle. Whether the stock option expenses should remain with the entity that initially bore them or be transferred to other group entities through an intercompany charge, ultimately hinges on the beneficiary. This is the principle underlying the benefits test described in Chapter VII, section B.1.1 of the OECD transfer pricing guidelines and the arm’s-length principle in general. As common-sense dictates and the witnesses’ testimony in the case confirmed, the subsidiary benefited from the parent’s contribution to its employees’ compensation packages.
Because employee stock options are ultimately just another way to pay employees, a good way to assess the decision is to ask whether the result would have been different if the foreign parent had instead paid discretionary cash bonuses to the subsidiary’s employees. If the parent had the same level of control and administrative responsibility, the group maintained the same geographically fluid structure, and the parties’ services contract contained an parallel exclusion for bonus payments, would Millrine have reached the same conclusion?
It seems highly doubtful that Millrine would have accepted the obviously illogical premise that an enterprise dealing at arm’s length would dole out cash to other enterprises’ employees for its own benefit. Employee stock options may differ from cash payments in important respects, but none of those differences bear on the question at issue in case 59TACD2024.
Missed Opportunity
These critiques of Millrine’s reasoning aren’t novel or ground-breaking. They follow from the analysis contained in the OECD secretariat’s 2004 report “Employee Stock Option Plans: Impact on Transfer Pricing.” The 2004 report partially compensates for the canonical OECD transfer pricing guidelines’ vague, equivocal, and notably curt guidance on employee stock options.
The OECD report is a fair, if not completely authoritative, guide to the proper treatment of employee stock options under article 9 and the OECD transfer pricing guidelines. The report acknowledges that it didn’t have the backing of all OECD members, but it evidently had enough support for the secretariat to publish it and advertise its “considerable input” from and “detailed discussions” with Working Party 6 delegates. Regardless of its lack of formal legal standing, it likely represents the most reliable interpretive authority on the application of the arm’s-length principle to employee stock options. It could have informed Millrine’s analysis if the parties had brought it to her attention, and it’s unfortunate that they did not.
Reviewing the 2004 OECD report would have revealed that the case bears a strong resemblance to the report’s TOPCO example. In the example, the options relate to shares in a publicly traded ultimate parent (TOPCO), the recipients are employees of a foreign subsidiary (SUBCO), and the parent has sole responsibility for administering the stock option plan and determining awards. But unlike in Millrine’s decision, these factors don’t factor prominently in the TOPCO example’s analysis.
As explained in paragraph 41 of the report, the arm’s-length principle requires a two-step analysis when the stock option recipients’ employer (like the Irish subsidiary) is an intragroup service provider:
Where the employer acts as a service provider to other members of THE GROUP with respect to the activities performed by some employees who benefit from stock options issued by TOPCO, the proper treatment, for transfer pricing purposes, would be to recognise (i) a transaction between TOPCO and the employer with respect to the provision of the stock option plan and (ii) a provision of services by the employer to other members of THE GROUP. This is consistent with the premise in this study that stock options are remuneration for employment services performed by the employees benefiting from the plan. The charge by TOPCO to the employer with respect to stock options granted to these individuals would become part of the cost of rendering services to other members of THE GROUP that might be charged separately. [Emphasis added.]
Although the stock option issuer in case 59TACD2024 is also the service recipient and payer of the cost-plus services charge, the crisscrossing exchange of value shouldn’t distract from the required analysis. In accordance with the approach endorsed in the OECD report, a subsidiary like the one in the case should first compensate its parent at arm’s length for issuing stock options to its employees. The OECD report makes clear that quantifying the value of stock options can be a technically challenging exercise, but it was necessary to comply with the arm’s-length principle.
Observing the first step would have added the arm’s-length transfer price for the options to the subsidiary’s costs, thus shifting the stock option expense from the parent to the subsidiary.
In the second step of the OECD report’s approach, the subsidiary’s stock option expense should have been treated as a services cost when applying the TNMM. As confirmed in paragraphs 167 through 178 of the OECD report, employee stock options must be included in costs when applying the TNMM or cost-plus method. Non-arm’s-length pricing terms contained in intercompany contracts, including the stock option exclusion in the services agreement between the Irish subsidiary and its foreign parent, cannot justify an unreliable application of the TNMM.
It follows that the Irish revenue commissioners were right about the correct result, even if their singular reliance on FRS 102 was misplaced. It was Millrine who reached the wrong conclusion, and her failure to follow anything that resembles the OECD report’s approach explains why. The first step would have made clear that the employee stock option costs had been economically borne by the subsidiary, and their character as intercompany services costs would have been easier to recognize. Once the stock options had been identified as services costs incurred by the subsidiary, the case for excluding them when applying the TNMM would have evaporated.
Notional Nonsense
A wrongly decided case anywhere is never a good thing. And the faulty logic in case 59TACD2024 could have the unhelpful effect of encouraging or validating flawed arguments in other jurisdictions. However, to the likely dismay of many U.S. multinationals and tax advisers, Millrine did not accept the foundation for what might be called the Xilinx-Altera theory: the view that the arm’s-length standard cannot require that controlled parties share stock-based compensation costs because uncontrolled parties do not share such costs at arm’s length. Millrine also tacitly rejected the premise that employee stock options aren’t even costs in the first place.
Although she wasn’t required to decide the question because the subsidiary abandoned the argument in litigation, Millrine never questioned the premise that stock-based compensation costs are real economic costs. In fact, her decision tacitly accepts that they are. Millrine noted in the decision that she “considers that the relevant question for transfer pricing purposes is whether the SBAs issued by the parent company created an economic cost for the Appellant.” In other words, the stock option costs may have been “notional” for the Irish subsidiary, but they were real for the foreign parent.
Even if the decision botched the attribution issue, it’s gratifying to see a foreign tax law judge set aside a flawed but common argument among U.S. taxpayers. Courts, sometimes intentionally and sometimes unwittingly, have allowed the theory to fester. For example, the Ninth Circuit majority’s Altera opinion was right on many points, but its observation that the economic costs of granting stock-based compensation “are debatable” is at best incomplete. Although the amount of the cost and the party that bore it may be in question, whether it represents a cost in some amount for someone shouldn’t be.
A direct cash outflow isn’t necessary for there to be a cost. If employee stock options were just notional expenses — accounting entries with no economic costs underlying them — then there would be no reason for companies not to grant every employee and random passerby $900 trillion worth of stock options. This practice is no more common in arm’s-length dealings than is the sharing of stock-based compensation. There must be some economic forces at work that induce employers to restrict their stock option grants.
There are. One of them is concern about the cost of stock dilution, which was something the taxpayer’s group in case 59TACD2024 evidently took very seriously. As one of its witnesses testified, the foreign parent closely and continuously monitored the stock dilution cost of granting employee stock options and strove to balance it against the need to attract employees.
There’s a more fundamental reason as well: Stock options have undeniable economic value, which is why employers can better entice employees by offering them and why the Black-Scholes model and more advanced options pricing methods have gained broad acceptance. As repeatedly stated in the OECD’s 2004 report, giving away something that has monetary value to an unrelated party, as opposed to retaining or selling it, is a cost. If arm’s-length-market investors would be willing to pay more than $0 for a company’s stock options on the open market, giving the options to employees instead is an economic cost.
The testimony of one of the taxpayer’s witnesses corroborates this:
The witness stated that . . . some employees would treat the SBAs as additional bonus pay, using it as part of their regular monthly income, selling the SBAs every month to pay their bills, while others would let the SBAs build up and increase over time, until a lump sum of cash is required.
In other words, employees could push a button and convert their stock options into cash. If something can immediately generate income for the recipient and is readily convertible into a cash lump sum, then relinquishing it was a cost.
As to the Xilinx-Altera theory’s core tenet regarding the significance of transactional evidence, Millrine’s decision rejected it by quietly ignoring it. Evidently, neither she nor the taxpayer considered it worth examining or even acknowledging. It was enough to establish that employee stock options are a form of employee compensation cost, leaving only the only question of who bore it. Although there may be something to be said for de-platforming a bad arguments, the Xilinx-Altera theory retains too many adherents in the United States to be ignored.
As Millrine emphasized in her decision, the case principally hinged on the meaning of the OECD transfer pricing guidelines. Ireland’s transfer pricing legislation, like that in many other countries, requires that domestic law be interpreted in the manner that best aligns with the OECD guidelines. Although the Xilinx-Altera theory’s most glaring faults relate to U.S.-specific legal considerations, it’s important to recognize that the theory should fail under the OECD guidelines (and jurisdictions that follow them) as well.
Unlike the 2004 OECD secretariat paper, which expressly and repeatedly rejects the Xilinx-Altera theory, the OECD guidelines take no explicit position on the proper treatment of stock options when applying the TNMM. But that doesn’t mean the guidelines are agnostic on the question. The principles underlying the guidelines’ general prohibition on nonrecognition firmly reject the Xilinx-Altera theory by implication.
The guidelines repeatedly emphasize that controlled parties’ arrangements cannot be ignored merely because arm’s-length parties don’t enter similar arrangements. As paragraph 1.142 of the guidelines, in relevant part, explains:
Importantly, the mere fact that the transaction may not be seen between independent parties does not mean that it should not be recognised. Associated enterprises may have the ability to enter into a much greater variety of arrangements than can independent enterprises, and may conclude transactions of a specific nature that are not encountered, or are only very rarely encountered, between independent parties, and may do so for sound business reasons. [Emphasis added.]
Paragraph 1.143 reinforces the point:
The key question in the analysis is whether the actual transaction possesses the commercial rationality of arrangements that would be agreed between unrelated parties under comparable economic circumstances, not whether the same transaction can be observed between independent parties. The non-recognition of a transaction that possesses the commercial rationality of an arm’s-length arrangement is not an appropriate application of the arm’s-length principle. [Emphasis added.]
The sine qua non of the Xilinx-Altera theory — that arm’s-length parties do not share stock option costs — is thus nothing more than a “mere fact” that doesn’t answer the question under the OECD guidelines. Transactions involving the sharing or reimbursement of stock-based compensation costs are clearly among the “much greater variety of arrangements” that associated enterprises can enter for “sound business reasons.”
Accepting, as Millrine and the parties did in case 59TACD2024, that stock-based compensation costs are real costs casts doubt on the Xilinx-Altera theory’s plausibility. All else being equal, an enterprise dealing at arm’s length wouldn’t incur employee compensation costs that can’t be reimbursed when reimbursable alternatives exist. But that doesn’t entirely rule out the possibility that stock-based compensation costs should be ignored in cost-plus services transactions or cost-sharing arrangements. Rejecting that premise requires further recognition that the arm’s-length principle, contrary to the assumptions of many taxpayers, commentators, and even some judges, is not blindly tethered to transactional evidence.
As repeatedly emphasized in the OECD’s 2004 report and explained in detail elsewhere, the reasons that arm’s-length parties distinguish employee stock options from other compensation costs are irrelevant when a parent company transacts with its subsidiary. Selectively ignoring stock-based compensation when, as it is for controlled transactions, it would be arbitrary to do so defies economic logic. Or to borrow the OECD’s choice of words, it would lack the “commercial rationality of an arm’s-length arrangement.”
When transactional data points to a result that would defy basic economic principles, nothing in article 9 or the OECD transfer pricing guidelines dictates that tax administrations, taxpayers, or courts must blindly accept it. In fact, paragraph 1.143 of the guidelines prioritizes commercial rationality over transactional evidence by stating that the “mere fact that the transaction may not be seen between independent parties does not mean that it does not have characteristics of an arm’s length arrangement.” In other words, an arrangement between controlled parties can be arm’s length irrespective of whether arm’s-length parties enter it as well.
If transactional data and the assumption of arm’s-length economic behavior conflict, as they do for employee stock options, transactional data can and should give way. If the Xilinx-Altera argument arises in future Irish transfer pricing litigation, Irish judges should explicitly reject it.
Read the full article here
