Apple’s Cost-Sharing Arrangement: Frankenstein’s Monster

I. Introduction

On April 13 IRS Commissioner Charles Rettig testified in a Senate Finance Committee hearing that the annual tax gap between the amount owed and the amount paid by all taxpayers may be larger than $1 trillion, and that much of it can be attributed to high-income individuals and corporations.¹ That projection would be more than twice the most recent IRS tax gap estimate, which, for corporations, is “only” $44 billion per year (gross).² However, academic estimates of corporate offshore profit shifting alone exceed $100 billion. Which brings us to Apple Inc.

Because the IRS tax gap estimate is based largely on a small number of examination outcomes and appears to exclude corporate offshore tax avoidance, the amount of any offshore tax underpayments by Apple or other large companies included in the estimate is likely zero.³ This is despite a conclusion by the European Commission, based on substantial evidence provided by Apple, that the company underpaid as much as $41 billion of U.S. taxes between 2003 and 2014.⁴ This report finds that the true U.S. tax underpayments could be as much as $84 billion or more. It further finds that those underpayments were likely facilitated by the IRS’s failure to audit Apple’s apparent noncompliance with the 2008 temporary cost-sharing transition rules on January 5, 2009, and by inaccurate or possibly false documentation of the group’s cost-sharing arrangement (CSA) as in effect from January 5, 2009.

Apple is now the most valuable company on the planet, with a market capitalization exceeding $2 trillion. Since the company launched the iPhone under CEO Steve Jobs in June 2007, it has been a revenue and profit juggernaut. Consider that between June 2007 and 2015, annual iPhone sales increased an astounding 16,600 percent, from 1.4 million to more than 230 million.⁵ Apple’s revenue increased by 1,350 percent, and its pretax income by 2,330 percent, from the year before the launch of the iPhone through 2019. Beginning with AT&T in the United States, Apple began selling iPhones in the hundreds of millions to the world’s largest wireless carriers, which purchased them upfront to provide to customers, often with multiyear wireless services contracts. The iPhone spawned the App Store and its walled garden, generating 30 percent commissions on all purchases conducted through the store or through apps.⁶ The rest is history. Today Apple’s iOS and Alphabet’s Android mobile operating systems are a duopoly that have about a 99 percent market share globally for wireless operating systems.⁷

Around 2010 something interesting happened in Apple’s profit reporting. In the prior year Apple’s U.S. and foreign revenue and pretax income reflected only minor differences, with U.S. revenue slightly exceeding foreign revenue, but with foreign profits having just overtaken U.S. profits that year (see Table 1, derived from Apple’s Form 10-K reports). Apple’s foreign revenue was about 92 percent of U.S. revenue, and pretax income was about 120 percent of the U.S. figure. However, in 2010 Apple’s foreign revenue overtook the U.S. revenue by about 28 percent, and foreign pretax profits exploded — exceeding U.S. profits by 135 percent. Clearly something changed in 2010 that altered (that is, increased) the trajectory of profit shifting per marginal dollar of U.S. and foreign sales. Soon thereafter, Apple’s tax arrangements began to attract attention in academic and practitioner circles.⁸ And in 2013 the U.S. Senate and the European Commission each launched investigations into Apple’s transfer pricing.⁹

This coincided with greater scrutiny by the U.S. and foreign tax news media, predominantly focused on apparent indications of noncompliance with U.S. international tax and transfer pricing laws by Apple, and a possible breakdown in U.S. transfer pricing enforcement.¹⁰ Condemnation of Apple’s profit shifting and violation of at least the spirit of U.S. tax laws was almost universal, with one notable exception: the U.S. Treasury Department, as part of its challenge to the European Commission’s state aid investigations, defended Apple’s transfer pricing and international tax planning.

To date, there has been no definitive forensic transfer pricing analysis of the evidence disclosed in the Senate and European Commission investigations — or provided by Apple in filings with those bodies, the SEC, courts, and other sources — regarding the company’s compliance with U.S. transfer pricing laws. We undertake that study in this report and find that Apple could be subject to a U.S. tax adjustment under reg. section 1.482-7(i)(6) that could exceed by several multiples the amount of taxes at issue in the European Commission state aid case (now under appeal) and Apple’s reserves for uncertain tax positions. That result is consistent with prior research that, if applied to Apple, would likely find that much of the company’s stateless income is taxable in the United States because it is effectively connected to a U.S. trade or business, and that some of the rest of it is subpart F income under the foreign base sales company income branch rule.

Alternatively, the stateless income could be taxable in the United States under the economic substance doctrine.¹¹ Apple itself has effectively admitted as much in filings with the General Court of the European Union (GCEU), indicating that the company’s stateless income reported in Ireland was in fact commercialization income (exploitation or services income) attributable to U.S. sources.¹² Moreover, the European Commission disclosed that Apple had informed it that the functions listed in the company’s revised CSA contract of June 25, 2009, as being performed by Apple’s Irish affiliates, Apple Sales International (ASI) and Apple Operations Europe (AOE), were actually performed by Apple Inc. This sort of knowingly false documentation in an intercompany agreement might even suggest to some observers that the agreement itself is a sham.

This report describes those findings and their regulatory and factual support. It also provides additional background that may help explain how corporate taxpayers have been able to exploit the U.S. cost-sharing regulations to shift hundreds of billions or even trillions in U.S. profits offshore to foreign shell and holding companies in complex corporate tax shelters — apparently with little or no IRS detection or enforcement.

II. A Transfer Pricing View of Apple

A. Financial Perspective

Table 1 shows Apple’s U.S. and foreign revenue, U.S. and foreign pretax income, and estimates of its stateless income — all since 1992 (the first year after Apple’s 1991 tax ruling with Ireland). These figures are based on an extrapolation of several years of actual results disclosed by the Senate in 2013. The figures show that Apple apparently reported more pretax income in what were effectively offshore tax haven shell companies than it reported in the United States, which is consistent with greater foreign revenue. The location of revenue (and the product mixes and prices on which it is based) is, of course, the result of uncontrolled arm’s-length transactions with unrelated customers. However, the distribution of profits between Apple U.S. group members and its foreign group members, including tax haven shell companies (for reasons discussed later), is almost exclusively the result of an internal transfer pricing exercise. Whether Apple’s profit distribution is reasonable typically cannot be determined based on book income reporting alone; one would need information on the company’s internal activities. However, the information on Apple’s internal functions that is available from reliable public sources is sufficient to allow us to draw some meaningful insights on at least a preliminary or indicative basis.

Table 1. Estimates of Apple Inc. Stateless Income^a
($ million)

Year	U.S. Revenues (a)	Foreign Revenues (b)	U.S. Pretax Income (c)	Foreign Pretax Income (d)	Estimated Stateless Pretax Income (e)
1992	$3,885	$3,202	$244	$611	$306
1993	$4,388	$3,589	($276)	$416	$208
1994	$4,982	$4,206	$26	$474	$237
1995	$5,791	$5,271	$102	$572	$286
1996	$4,735	$5,098	($1,154)	($141)	($71)
1997	$3,507	$3,574	($780)	($265)	($175)
1998	$3,287	$2,654	$14	$315	$158
1999	$3,394	$2,740	$64	$612	$306
2000	$4,145	$3,838	$73	$1,019	$510
2001	$2,936	$2,427	($415)	$363	$182
2002	$3,272	$2,470	($197)	$284	$142
2003	$3,627	$2,580	($158)	$250	$165
2004	$4,893	$3,386	($1)	$384	$253
2005	$8,194	$5,737	$893	$922	$609
2006	$11,486	$7,829	$1,318	$1,500	$990
2007	$14,128	$9,878	$2,808	$2,200	$1,452
2008	$18,469	$14,010	$3,395	$3,500	$2,310
2009	$22,325	$20,580	$5,466	$6,600	$4,000
2010	$28,633	$36,592	$5,540	$13,000	$12,000
2011	$41,812	$66,437	$10,205	$24,000	$22,000
2012	$60,949	$95,559	$18,963	$36,800	$36,000
2013	$66,197	$104,713	$19,655	$30,500	$28,500
2014	$68,909	$113,886	$19,883	$33,600	$31,600
2015	$81,732	$151,983	$24,915	$47,600	$41,600
2016	$75,667	$139,972	$20,272	$41,100	$35,100
2017	$84,339	$144,895	$19,389	$44,700	$38,700
2018	$98,061	$167,534	$24,903	$48,000	$42,000
2019	$102,266	$157,908	$21,437	$44,300	$38,300
2020	$109,197	$165,318	$28,991	$38,100	$32,347
Total	$945,206	$1,447,866	$225,575	$421,316	$370,014
Proportion of pretax income			35%	65%	57%
aColumns (a)-(d) are from SEC filings, and results in column (e) for years 2009-2011 are actuals reported by the U.S. Senate as exhibits to the 2013 Senate PSI report entitled “Hearing On Offshore Profit Shifting and the U.S. Tax Code Part 2 (Apple Inc.),” May 21, 2013, and remaining years are estimates based on extrapolating the ratios established by the Senate report. An adjustment was made to reduce the stateless pretax income beginning in 2015 due to additional Irish taxes reported by Apple.

In November 2017 Apple reported that in response to changes in Irish tax rules in 2015 it had moved the residency of its cash holding company into the U.K. crown dependency of Jersey (a tax haven in the Channel Islands). Apple also reported that it had restructured so that all its Irish operations were conducted through Irish resident entities, and that its Irish taxes increased an average of $500 million per year beginning after 2014.¹³ Note that before it phased out the Double Irish tax avoidance structure in 2020, Ireland established a capital allowance program for depreciation of acquired intangible assets at a rate of 100 percent for intellectual property assets purchased by an Irish entity, including deductibility of interest payments on intragroup loans used to purchase the IP. This made it particularly tax-efficient for an Apple Irish resident company to purchase the IP held by ASI at the time of this restructuring. The role of the Jersey affiliates in this new arrangement would appear to be to hold cash and provide loans to the Irish resident company to fund the purchase of the CSA-related IP, thereby eliminating taxes on most if not all of the IP income that might flow to the Irish resident entity. This would largely replicate the tax benefits of their pre-2015 structure, with the profits now ending up untaxed in Jersey instead of Ireland.¹⁴

Table 1 shows that Apple’s foreign subsidiaries are estimated to have recorded around $370 billion of pretax stateless income since 1992 — more profits than were recorded in the United States (noting that Apple obtained its first Irish tax ruling in 1991, which contributed to those results). This is important because subjecting those profits to a combined federal and state tax rate of around 40 percent would equate to approximately $150 billion of additional taxes. These tax benefits give Apple a substantial advantage over domestic competitors (which include domestic companies selling apps in Apple’s App Store that compete with Apple offerings). One could say that around a third of Apple’s earnings per share is attributable to tax planning. As demonstrated below, transfer pricing appears to be a primary means of generating that additional net income.

Tables 2 and 3 are based on information that Apple provided in its SEC filing for fiscal 2011 (covering its 2009-2011 tax years), including the consolidated income statement and information in note 8, “Segment Information and Geographic Data,” and note 5, “Income Taxes” in that filing. Importantly, in note 8 Apple breaks out the retail segment from the geographic segments but adds the following footnote: “The Americas asset figures do not include fixed assets held in the U.S. Such fixed assets are not allocated specifically to the Americas segment and are included in the corporate and Retail assets figures below.”

Because the corporate and retail assets are held in the United States, Table 2 adds the Americas and retail segment figures together to capture all Apple U.S. results in the Americas segment. Apple notes that its retail assets include what are often described as flagship venues in expensive locations, whose high costs can produce perpetual losses but are often considered a type of corporate marketing expense that benefits the group’s brand. In our experience, loss-making flagship retail operations are often reimbursed by the corporate parent, which bears the risks of what is a corporate-level flagship retail and marketing strategy. Apple in July 2011 had 326 retail stores (all based on an original design by Jobs¹⁵) and disclosed in SEC filings that it consolidated those assets into a retail segment that appears to be a U.S.-based segment that manages and controls these geographically dispersed flagship and other retail operations.¹⁶

Table 2. Marginal Rates of Return by Segment Reported by Apple Inc. (2009-2011)^a
($ million)

	2011	2010	2009	Three-Year Change
Americas and Retail:
Net Sales	52,442	34,296	25,637	105%
Operating Income	16,882	9,954	8,335	103%
Operating Margin	32%	29%	33%	0%
% Change in Income Minus % Change in Sales				-2.0%
Europe:
Net Sales	27,778	18,692	11,810	135%
Operating Income	11,528	7,524	4,296	168%
Operating Margin	42%	40%	36%	14%
% Change in Income Minus % Change in Sales				33%
Japan:
Net Sales	5,437	3,981	2,279	139%
Operating Income	2,481	1,846	961	158%
Operating Margin	46%	46%	42%	8%
% Change in Income Minus % Change in Sales				20%
Asia-Pacific:
Net Sales	22,592	8,256	3,179	611%
Operating Income	9,587	3,647	1,100	772%
Operating Margin	42%	44%	35%	23%
% Change in Income Minus % Change in Sales				161%
	2011	2010	2009	Sum
Sum of Non-Americas Op. Profit	$23,596	$13,017	$6,357	$42,970
Foreign Pretax Income (Note 5)	$24,000	$13,000	$6,600	$43,600
Difference (Non- Operating Income)	$404	($17)	$243	$630
Difference as % of Non-Americas Op. Profit				1.5%
aApple Inc. FY2011 Form 10-K, Note 8 (Segment Information and Geographic Data).

Two observations are immediately apparent. First, the Americas segment (which includes both North and South America but is mostly the U.S. parent), now including the retail segment, had a decline in marginal profit (that is, the change in profit per change in sales) over the three years, and zero growth in operating margin, despite a triple-digit increase in sales over the period.¹⁷ This is the opposite of the foreign segments in Europe, Asia-Pacific, and Japan, which show double- or triple-digit growth in these same marginal profitability metrics.

This comparison is important in a transfer pricing context because it indicates that margins are growing with sales outside the Americas but not within the Americas territory. Barring any substantial differences in product mixes, this would be unusual because Apple products are mostly manufactured in the same locations at the same costs. Given that a growth in sales requires growth in operating expenses, this could be consistent with bearing an increasing share of those marginal costs in the Americas segment, holding down profit margins despite sales growth. A second observation from Table 2 is that the operating profit of the non-Americas segments over the three years as reported in note 8 is less than the foreign pretax income reported in note 5, which appears to mean that the non-Americas segments are (1) not bearing much, if any, of the worldwide corporate and operating expenses and (2) earning the majority of Apple’s nonoperating income.

Apple operates an almost completely centralized supply chain, in which almost all critical internal supply chain activities are performed, managed, or controlled by U.S. group members. (Production of iPhones and other products is outsourced to third parties, mostly in China, in accordance with a strategy engineered by Apple’s current CEO, Tim Cook — who was hired by Jobs in March 1998 to do exactly that.¹⁸) U.S. group members also had the preponderance of the company’s fixed assets and employees during that period.¹⁹ Apple U.S. is the primary generator of Apple’s worldwide profits, and it performs almost all the IP development and IP exploitation activities, with the majority of employees, executives, and assets, yet it reports only a fraction of Apple’s pretax income.

Table 3 provides an additional reconciliation of these corporate allocations, showing they are borne almost exclusively in the Americas segment (and, in particular, the United States). This appears to be one factor in the disproportionate marginal profitability between the Americas and other segments.

Table 3. Reconciliation of Estimated Corporate Cost Allocations

Calc	Measurement (2009 Through 2011 per Apple FY2011 Form 10-K)	Value ($ mil)
(a)	Americas + Retail Segment Op. Profit	$35,171
(b)	Total Non-Operating Income From Income Statement	$896
(c)	Non-Americas Segments Non-Op. Income From Table 2	$630
(d) = b - c	Americas Non-Operating Income	$266
(e)	Apple US Pre-Tax Income From Income Statement	$21,211
(f) = a - e + d	Plug for Est. Allocations of Corp. Expenses to Americas + Retail Segment	$14,226
(g)	Corp. Expenses + Corp. Depreciation From Note 8	$14,431
(h) = f/g	Est. Corp. Expense and Depr. Allocations to Americas + Retail Segment: % Total	99%

During roughly the same period, Apple reported the following results in its schedule of UTPs in its SEC filings:

Table 4. Apple UTP Reporting in SEC Filings
($ million)

Year	Additions to UTP for Prior Years	Additions to UTP for Current Year	Reductions in UTP (assumed booked into earnings)	Tax Settlements
2008	$27	$85	$81	$-
2009	$847	$151	$24	$-
2010	$61	$240	$227	$102
2011	$49	$425	$42	$-
2012	$340	$467	$117	$3
2013	$745	$626	$127	$592
2014	$1,295	$882	$284	$574
2015	$2,056	$1,278	$358	$109
2016	$1,121	$1,578	$257	$1,618
2017	$333	$1,880	$991	$539
2018	$2,431	$1,824	$2,212	$756
2019	$5,845	$1,697	$765	$852
2020	$454	$1,347	$860	$85
Totals	$15,604	$12,480	$6,345	$5,230

As disclosed in SEC filings, the IRS has now examined and closed Apple’s tax years through 2015. According to those UTP figures, Apple reported a total of $5.2 billion in tax settlements from all countries and states since 2007, which is very small compared with Apple’s estimated stateless income booked into its Irish CSA participants since its 1991 Irish tax ruling, as reported in Table 1. Moreover, Apple has added $28 billion to UTP and booked $6.3 billion of UTP into earnings between 2008 and 2020.

It seems probable that Apple’s UTP reserves relate to various country- and issue-specific items and not to the general risk of its entire profit-shifting structure. This is because the UTP balance of approximately $16.5 billion as of the September 26, 2020, balance sheet date is small compared with what the U.S. tax would be on the cumulative $370 billion of estimated stateless income if all that income had been currently subject to U.S. or other countries’ taxes. With Apple maintaining during the European Commission investigation that its stateless income is attributable to value created in the United States, there is good reason to investigate what might be behind the apparent U.S.-source income shifted offshore and the potential that much or all of that income should have been reported in the United States on a current basis.

B. Functional Perspective

Apple is a highly centralized company, with (according to sources listed below) almost complete control over U.S. and foreign operations by U.S. executives and personnel, including generating foreign revenue and managing foreign production conducted by unrelated contract manufacturers. Beginning with sales generation and focusing on the iPhone (Apple’s primary source of profits almost since the product’s launch in 2007), consider that U.S. executives negotiated, approved, and executed virtually all the sales contracts for iPhones to the world’s largest carriers. That is indicated in the following published information, which includes information from Apple Inc. and Apple executives, and information provided in governmental investigations, court filings, and news media reports:

• “When the iPhone came out, Cook spearheaded negotiations with wireless carriers around the world.”²⁰

• “Mobile operator O2 is preparing to unveil Apple’s much anticipated iPhone in the UK tomorrow. . . . The UK’s largest mobile operator came from behind at the last minute to seal an agreement with Steve Jobs, Apple’s chief executive, to market the iPhone in the UK. . . . The deal . . . will return to Apple as much as 40 percent of any revenues it makes from customers’ use of the device.”²¹

• “The three visits in 13 months that Cook undertook after becoming CEO are part of the reason Apple was able to ink a deal in December with China Mobile. That ended six years of negotiations, according to The Wall Street Journal, which has a recap of the most recent meeting between Cook and China Mobile Chairman Xi Guohua. Those negotiations were started in 2008 by Jobs.”²²

• Regarding getting the iPhone into Japan: “According to [Masayoshi] Son, Jobs said, ‘We have not talked to anybody, but you came to see me as the first guy. I’ll give it to you.’ The negotiation continued, as Son asked that Jobs put in writing that Apple would give him exclusivity for the Japanese market.”²³

• “One day after announcing the iPhone would go on sale in the UK on November 9th . . . Apple announced the device will hit German stores on the same day, but through T-Mobile. . . . Deutsche Telekom AG’s [T-Mobile’s parent company] CEO Rene Obermann and Apple CEO Steve Jobs made the announcement in Berlin.”²⁴

• “The negotiations and the signing of contracts with customers, such as telecommunications operators . . . were signed on behalf of the Apple Group by Apple Inc.”²⁵

• “As of September 24, 2011, the Company distributed iPhone in 105 countries through 228 carriers.”²⁶

In addition to negotiating, approving, and executing foreign sales, U.S. executives directly managed foreign supply chain operations, as described in the following anecdotal information from various public sources, including published governmental investigations:

• “Mr. Cook, who joined the company in 1998, is the architect of Apple’s China business. As head of operations, he assumed responsibility for a company saddled with extra inventory and reliant on its own U.S. manufacturing plants. . . . Around 2000, he met Foxconn founder Terry Gou, who would become the dominant contract manufacturer in Asia. Foxconn struck deals to be one of the few manufacturers of iPods, which made its debut in 2001, and an early maker of the iPhone, which launched in 2007. . . . In 2001, Apple officially entered China with a Shanghai-based trading company.”²⁷

• “Contracts with third-party original equipment manufacturers . . . for the manufacture of a large proportion of the products sold by ASI, were negotiated and signed by the parent company, Apple Inc.”²⁸

• “The framework agreements with the manufacturers of Apple-branded products (or OEMs) had been concluded centrally in respect of the Apple Group as a whole by Apple Inc. and ASI in the United States.”²⁹

• “It is also apparent from that evidence that the . . . organisation of distribution on the European markets . . . had been managed at the Apple-Group level by, inter alia, the Executive Team . . . in Cupertino.”³⁰

• “For such a sprawling organization, Apple also is a headquarters-centric company. Sure, there are sales offices and retail stores throughout the world. And Apple has established China as its base of manufacturing. But its entire management team is based in Cupertino. . . . Apple people board airplanes at the drop of a hat, but . . . [m]eetings generally happen in Cupertino. Moreover, there’s a sense that only people in Cupertino are truly to be trusted.”³¹

• “The Apple approach to management and talent development is top-down. It begins with an all-knowing CEO aided by a powerful executive team — the ‘ET,’ as it is known throughout the company. ‘The purpose of the executive team is to coordinate things and set the tone for the company,’ Jobs once said. This ten-member group, including the CEO, comprises the heads of product marketing, hardware and software engineering, operations, retail stores, Internet services, and design, all of whom directly have a hand in Apple’s products. They’re joined by the heads of finance and legal.”³²

• “When a product is ready to leave the lab, two key people will take control: an engineering program manager, or EPM, and a global supply manager, or GSM. . . . The global supply manager, working on the operations group that Tim Cook built, figures out how to get the materials to make it. They do everything from sourcing to procurement to overseeing production. . . . EPMs and GSMs at Apple are based in Cupertino, but they spend much of their time in China, where Apple contracts with Chinese manufacturers to build its computers and mobile devices.”³³

• “Early in his tenure, Cook remarked at a meeting with his team that a certain situation in Asia was a real problem and that one of his executives ought to be in China dealing with it. The meeting continued for another half hour or so, when Cook stopped abruptly, looked up at one of his executives, and asked in all seriousness: ‘Why are you still here?’ The executive stood up, drove to the airport without a change of clothes, and flew to China.”³⁴

Finally, Apple’s stateless pretax income was generated by U.S. activities, according to the following information, some of which was provided by Apple executives and in sworn testimony or court filings:

• “All the functions that drive Apple’s profits are directed by Apple Inc. executives in the US and performed largely in the US.”³⁵

• “No decisions concerning the exploitation of Apple’s IP . . . are made in Ireland. No employee of the Irish branches has responsibility for . . . any decision associated with the right to use and exploit Apple’s IP . . . [D]ecisions concerning the exploitation of Apple’s IP . . . are made in the US by Apple Inc.”³⁶

• “Neither ASI nor AOE is able to monitor business risk in the absence of employees.”³⁷

• “Based on the facts presented to the Commission, it appears that during the period the contested tax rulings were in force the head offices of ASI and AOE existed on paper only.”³⁸

• “The applicants’ [ASI and AOE’s] profit-driving activities, in particular the development and commercialisation of intellectual property (‘Apple IP’), were controlled and managed in the United States. The profits from those activities were attributable to the United States, not Ireland.”³⁹

These examples point to a highly centralized U.S. management and control of foreign operations, including execution of foreign sales agreements with major carriers, and the management and conduct of foreign production through personnel who identify sources of raw materials and components and contract manufacturers, negotiate terms with them and execute contracts, manage and determine inventory levels, set order quantities and timing, oversee quality control, manage and control logistics and distribution, etc. Together with the Apple group’s predominant share of employees, assets, and expenses in the United States, this would indicate that the foreign segment profit results shown in Table 1 are largely the product of an internal transfer pricing exercise that underallocates U.S. commercialization expenses to foreign affiliates while overvaluing the IP rights in the hands of the Irish affiliates.

The super-profitability of these foreign affiliates stands in stark contrast to the central and almost exclusive role of Apple U.S. in generating these foreign profits while earning what appear to be much lower profit levels. This mismatch seems extreme and at odds with the small tax settlements of only $5.2 billion shown in Table 4, especially considering that the European Commission’s investigation and Apple’s later submissions to the GCEU established that very few U.S. commercialization and head office expenses had been charged by Apple U.S. to its Irish affiliates.⁴⁰

In contrast to all this evidence, Treasury has strongly defended Apple’s position in Ireland’s dispute with the European Commission. Treasury wrote a white paper in August 2016 that it submitted to the GCEU in support of Apple’s defense against the state aid investigation. The white paper stated: “Under Irish law, an Irish non-resident company, just like any other non-resident company, would have been subject to tax only on the income attributable to activities in Ireland.”⁴¹

Of course, Treasury excluded from this argument any discussion of where exactly the activities that created these profits took place, perhaps because if they took place in the United States, the income would have been immediately taxable there under U.S. transfer pricing laws as active services income — although clearly, this income was not taxed in the United States. Apple, however, answered this question with its own submission to the European court only months later, in December 2016, when it noted that ASI and AOE’s profit-driving activities — in particular, the development and commercialization of Apple IP — “were controlled and managed in the United States. The profits from those activities were attributable to the United States, not Ireland.”⁴²

Apple appears to be stating that the stateless income booked into Ireland was indeed the direct result of ASI and AOE’s businesses being conducted on their behalf by Apple U.S. personnel and U.S.-located ASI and AOE directors (who were also senior personnel of Apple Inc., according to the Senate investigation), although there appeared to be at least one Irish director at each company.⁴³ Importantly for tax law interpretation, with ASI and AOE being disregarded-entity subsidiaries of Apple Operations International (AOI), the controlled foreign corporation, the U.S.-located ASI and AOE directors are properly characterized for tax purposes as AOI employees (or at least agents) who have management responsibility for AOI divisions. This means that AOI has employees (or agents) physically within the United States conducting regular business activities.

It seems probable that by the time Apple made this statement, it had already reached an agreement with the IRS on the years to which the statement applied (that is, through 2014). Nevertheless, this statement is at odds with what Apple told the Senate in May 2013, when it stated under oath that the stateless income reported in Ireland was foreign-source income, qualifying for deferral from U.S. taxation:

International operations accounted for 61 percent of Apple’s revenue last year. . . . These foreign earnings are taxed in the jurisdiction where they are earned (“foreign, post-tax income”). . . . Apple carefully manages this foreign, post-tax income to support its foreign operations.⁴⁴

While the concept of stateless income has generally been taken to mean that the income in question is not taxable in any state under the laws of those states, Apple is conveying the exact opposite. In other words, Apple is representing to authorities in each “state” (in this case, each state being represented by the Senate and the European Commission) that the income in question was in fact earned in and taxable under the laws of the other state. This is a much different argument than claiming that the income is not taxable in any state.

Note that under U.S. income sourcing laws, IP income is sourced where the IP was used. So if the income in question was license fees earned by the Irish affiliate for IP licensed to third parties in its territories, it would be foreign-source income in the hands of ASI and AOE⁴⁵ for U.S. purposes. It would also be foreign-source income when Irish affiliates use the IP to conduct manufacturing (whether directly or through unrelated contract manufacturers) and sales, which is contractually what has occurred. Apple’s filing with the GCEU did not address IP income because it appropriately claimed, based on the contractual form, that the income reported in ASI and AOE was exploitation income related to the IP rights they hold under the group’s CSAs. Apple’s filing addressed the commercialization (or IP exploitation) income that arose from the activities that took place in the United States that generated the totality of the income in question, except to the extent of the routine functions actually conducted in Ireland for which Ireland had issued its private rulings. (The European Commission investigation and the later GCEU opinion focused on just the two disregarded entity subsidiaries — AOE and ASI — in contrast to the U.S. treatment of those subsidiaries as mere divisions or branches of AOI. From a U.S. perspective, some of the totality of the income would be attributable to AOI’s other check-the-box subsidiaries in Europe and the rest of world.)

Although the commission’s decision and the GCEU’s opinion do not provide details, it is likely that ASI and AOE earn income both from selling tangible and intangible personal property and from performing cloud transactions that would be classified as services.⁴⁶ Under U.S. laws, various source rules apply to these sales and services income in the hands of ASI and AOE. Given the extent of sales and production activities in the United States for both tangible and intangible property sales, as well as the management and operation within the United States of the various Apple internet-based platforms through which services are provided to customers worldwide, a considerable portion of ASI and AOE’s gross income should be U.S.-source.⁴⁷

Consistent with this, an Irish tax journal looked at Apple’s submission with more scrutiny than was apparent from U.S. tax authorities:

For the first time, Apple has expressly attributed the profits of its Irish subsidiaries to the United States . . . the commercialization of the IP happening in the US . . . the US is driving sales, negotiating contracts, etc., for the rest of the world. Apple must have a high degree of comfort that the IRS is okay with this argument.⁴⁸

The European Commission concluded that the stateless income allocated to the nonexistent head offices of ASI and AOE under the tax rulings was effectively characterized in Ireland as payments for services that were performed by Apple Inc., but fees for those services were never remitted to Apple Inc., a fact that was supported by the analyses presented earlier and by Apple’s submissions to the GCEU. The commission described this situation as “fictitious remuneration” that was in effect taken as a deduction against Irish taxable profits (through the division of profits between the nonexistent head offices, and Irish branches allowed for each company under the rulings) but never paid to Apple Inc. Based on the commission’s findings, these “expenses” appear to provide a value for virtually all of Apple Inc.’s commercialization and headquarters services that generated AOI’s stateless income. In other words, the only payments to Apple Inc. by ASI and AOE were payments under the CSA and some intercompany marketing expenses. There was little or no payment by AOI or its disregarded entity subsidiaries to Apple Inc. for the day-to-day actions of its personnel that exploited ASI’s and AOE’s IP on their behalf. Referring to earlier proceedings, the European Commission stated:

The Commission then explained, in response to claims made by Apple in its observations on the Opening Decision, that functions performed by Apple Inc. employees are outside the scope of the assessment of the contested tax rulings: “fictitious remuneration for services provided for free by the group employees to the supposed benefit of ASI or AOE cannot reduce the profits to be allocated between the ASI and AOE head offices and their respective branches.”

. . .

The terms of those agreements . . . were not in line with a level of contribution that would have been agreed between independent companies negotiating at arm’s length . . . because Apple Inc. employees performed activities for the benefit of ASI and AOE beyond the remunerated contributions covered by the CSA.⁴⁹

It is unclear whether such a violation of U.S. transfer pricing laws — if flagrant enough — could also be seen as a violation of the 2008 cost-sharing regulations under reg. section 1.482-7(a)(3)(iii), which state: “Controlled transactions between controlled participants that are not PCTs or CSTs [platform contribution transactions or cost-sharing transactions] . . . require arm’s length consideration under the rules of sections 1.482-1 through 1.482-6, and 1.482-9.”

There is some debate about why Treasury included this regulation. It seems superfluous in that it simply restates the existing U.S. compliance requirements for transfer pricing in general. The consensus view that emerged after discussions with the reviewers of this report is that two objectives are apparently connected to the addition of this separate provision within reg. section 1.482-7T, addressing this requirement within the cost-sharing regulations:

1. If one scrutinizes the calculations for a periodic trigger under reg. section 1.482-7(i)(6), this trigger can be generated by violations of reg. section 1.482-1 through -6 and -9 that cause the foreign CSA participant division profit to exceed the actually experienced return ratio (AERR) limitations, and Treasury is alerting taxpayers to the fact that a violation of these regulations could trigger a violation of reg. section 1.482-7(i)(6); and

2. those violations could give the IRS more power to invalidate CSAs or prevent their grandfathering under the 2008 regulations, because taxpayers were explicitly required to substantially comply with this very regulation, according to reg. section 1.482-7(m)(1).

That transition rule reads in part: “An arrangement in existence on January 5, 2009, will be considered a CSA . . . only if the activities of the controlled participants substantially comply with the provisions of this section . . . by July 6, 2009.”

If a significant portion of Apple’s stateless income was represented by uncharged or unallocated U.S. commercialization and head office services expenses in violation of reg. section 1.482-9, -4, or other sections, as Apple has represented to the GCEU, that would appear to run afoul of reg. section 1.482-7(a)(3)(iii) — which is one of “the provisions of this section” for which such noncompliance can result in a failure to comply with reg. section 1.482-7(m)(1) and in a loss of the attendant compliance benefits.

Despite the European Commission’s identification of what appeared to be violations of the U.S. transfer pricing regulations by Apple Inc., a Treasury lawyer (presumably speaking for the IRS) claimed that the U.S. government could not perform an analysis of Apple’s transfer pricing to determine if Apple broke any U.S. laws.⁵⁰ It is not clear why that would be the case, given that Apple told the Senate in 2013 that it was regularly examined by the IRS.⁵¹ Even then-Treasury Secretary Jacob Lew spoke up around the same time, saying: “The European Commission’s finding that Apple received illegal state aid from Ireland to the tune of €13 billion represents an attempt to tax income that belongs to the U.S.”⁵²

Treasury appears to be maintaining that (1) Apple’s transfer pricing was compliant with U.S. laws; (2) Apple’s stateless income recorded within AOI and its disregarded entity subsidiaries was properly deferred under the U.S. deferral tax regime; (3) Treasury agreed with Ireland that the Irish income was properly stateless because it was taxable in neither the United States nor Ireland; and (4) the income belonged to the United States by virtue of the U.S. right to tax the income when repatriated as a dividend (or a deemed dividend). These arguments are the foundations of Treasury’s challenge to the European Commission’s state aid investigations into Apple and other U.S. companies. We must take from Treasury’s comments that it and the IRS had no intention of ever examining the commission’s allegations of U.S. tax law violations by Apple Inc., and indeed the IRS closed its exams of the tax years in question with only de minimis adjustments, if any, as mentioned around the time of these various statements.⁵³

As noted earlier, the commission’s state aid investigation revealed several potentially serious violations of U.S. laws that could be problematic for Apple if the IRS and Treasury were to reverse course and examine this evidence under applicable regulations. Facts coming to light since 2016 suggest that the IRS may simply have never examined Apple’s transition to the 2008 CSA regulations, the functions borne by its CSA participants, and whether those functions complied with regulations or were accurately represented in relevant documents.

These types of risks prompted Congress in 1986 to mandate the commensurate-with-income standard, which is embodied in the periodic adjustment regulations,⁵⁴ so that these mistakes could be corrected in a practical manner in future years. However, there is no evidence that the IRS has ever enforced the commensurate-with-income rules in any prior cases or in connection with the 2008 CSA regulations. And Treasury and the IRS’s enforcement approaches to CSAs have failed to reduce outbound profit shifting.

The situation at Apple highlights the many unintended consequences of the reg. section 1.482-7 cost-sharing regulations — a beast of a regulation whose flawed design produces exactly the opposite of its intended effects in a way that perfectly personifies a Frankenstein regulation.

III. Apple and the Frankenstein Reg

A. A Monster Tax Shelter

Consider the following two comments in 2016 concerning the European Commission’s state aid case against Apple Inc. and the Irish government. Importantly, the text describes Apple’s two cost-sharing participants, ASI and AOE, as they existed at least up to 2014 (the final year covered by the commission’s state aid investigation). Lee A. Sheppard noted:

AOE and ASI were cash boxes . . . lacking substance. . . . Bare financial capacity was regarded as good enough to be deemed to bear risk under U.S. transfer pricing rules at the time (Veritas, 133 T.C. 297). . . . AOE and ASI were hollow corporations, and all management and decision-making for them took place in Cupertino. . . . The commission found no evidence of separate remuneration beyond the cost-sharing payments.⁵⁵

Reuven S. Avi-Yonah and Gianluca Mazzoni commented that “95 percent of Apple’s research and development was conducted in the United States, making Apple’s arrangement with ASI closer to a cost reimbursement where ASI simply received a contribution from Apple Inc. and paid it back.”⁵⁶

It is almost beyond belief that these two companies (or more precisely, AOI, which is the CFC taxpayer as a result of the check-the-box elections for AOE, ASI, and other AOI direct and indirect subsidiaries that conduct operations) made only cost-sharing payments (and some marketing payments, according to the European Commission’s state aid decision) and apparently paid nearly zero remuneration to Apple U.S. for managing and executing Apple’s global supply chain. Because of this, Apple U.S. was left to bear significant operating expenses that supported AOI’s revenue stream, but without any remuneration from AOI or its disregarded entity subsidiaries. This could represent a substantial enforcement failure that helps explain why Apple Inc. may now have a periodic adjustment. Note that even if the IRS addressed any unallocated commercialization or head office services, the existence of a periodic trigger would indicate additional undercompensated or uncompensated PCTs and routine or nonroutine contributions by Apple Inc. that have not been detected by the IRS on examination.

The fact that AOI, long after reg. section 1.482-7T became effective in 2009, paid for only a portion of the total exploitation expenses that generated its territory profits (that is, primarily exploitation expenses that were direct costs within AOI’s disregarded entity subsidiaries and not the expenses expended by U.S. group members) raises the question whether Apple’s cost sharing was in compliance with some aspects of the 2008 cost-sharing regulations. In particular, if Apple did not comply with the transition rules of reg. section 1.482-7T(m)(1), its CSA could be subject to a periodic adjustment under reg. section 1.482-7(i)(6). Further, even if the IRS were to make a significant section 482 adjustment for the services Apple Inc. has conducted for AOI, the periodic adjustment rules of reg. section 1.482-7(i)(6) would still apply, as discussed later.

The 2008 temporary cost-sharing regulations were perhaps the most important upgrade to these rules since their origin in the 1968 transfer pricing regulations.⁵⁷ However, that substantial upgrade was one in a long list of modifications that tried, unsuccessfully, to repair what many believe was a disastrous regulation for tax enforcement. Perhaps more than any other, this regulation is a classic example of a Frankenstein regulation.

In Mary Shelley’s 1818 Gothic novel Frankenstein, the scientist Victor Frankenstein sought to overcome, and in some ways reverse, the laws of nature and create something he believed would be magical and wonderous — something that would establish him as one of the great scientists of all time. However, his undertaking, which technically succeeded spectacularly, instead created a monster that he rejected, causing unforeseen and disastrous consequences he could not control. What started out as a vain endeavor wrought from hubris and overconfidence became a curse he would greatly regret:

What had been the study and desire of the wisest men since the creation of the world, was now within my grasp. . . . The materials present within my command hardly appeared adequate to so arduous an undertaking; but I doubted not that I should ultimately succeed.⁵⁸

. . .

A flash of lightning illuminated the object, and discovered its shape plainly to me . . . more hideous than belongs to humanity, instantly informed me that it was the wretch, the filthy daemon to whom I had given life.⁵⁹

Perhaps similarly, the reg. section 1.482-7 qualified cost-sharing regulations first emerged from a laboratory deep inside Treasury around 1968 and immediately unleashed hideously unenforceable tax-shelter-like outcomes that defied — actually reversed — the laws of economics and the arm’s-length principle. This was a regulatorily authorized transaction that no corporate taxpayer in its right mind would even think of doing with an unrelated counterparty. Treasury obviously envisioned a transaction it would create out of thin air (because it had no counterpart in the real world) that would accomplish something that it must have thought would improve tax administration. Instead, the CSA transaction authorized by Treasury achieved exactly the opposite of what transfer pricing laws are supposed to do — prevent uncontrolled profit shifting.

One tax expert put it bluntly: ”Cost-sharing has been an expensive mistake. It enables multinationals to shift the profits from intangibles developed in the U.S. offshore without incurring any serious risk of losing the R&D deductions.”⁶⁰ Another comment hints that Treasury may have believed that cost sharing would produce a net benefit to the United States by shifting research and development deductions offshore: “Why would the IRS regulations permit this? Because if the research is unsuccessful, then the taxpayer risks losing its ability to deduct the costs sent offshore (the deduction is shifted to the CFC, which does not have U.S.-source income).”⁶¹

There are at least five immediate problems with these lines of thought:

1. It was not basic theoretical or exploratory research that was conducted within the typical CSA; rather, it was applied research focused on specific products that were inserted into a CSA only if there was some expectation that further R&D activities would not result in failure. Accordingly, there were too often spectacular successes in which the profits shifted offshore were exponential multiples of the forgone U.S. R&D deductions.

2. These new R&D activities, especially in high tech, led to an ever-growing share of successful R&D in a virtuous circle of new and successful versions, upgrades, and extensions (such as the iPhone, search algorithms, or social networking websites).

3. The system was easily gamed because much if not most R&D was already focused on successful products. Initial research on a project is often conducted solely by the U.S. parent and moved into a CSA with a foreign group member only when there is some belief in the ultimate success of the project. The recorded buy-in price of the platform contribution at the time the project is moved into the CSA is then set at an artificially low amount. Although a common refrain posits that 90 percent of all R&D is unsuccessful, or something to that effect, in these arrangements, over time the majority if not all of the CSA-related R&D spending by the participants is successful — often spectacularly.

4. To the extent that R&D activities are not the primary value driver in an enterprise, the regulation produces a massive unsupported-by-the-facts tax giveaway (for example, the significantly effective supply chain structure developed, managed, and conducted daily from Cupertino by Apple Inc. personnel, and the capabilities of its management executives such as Jobs and Cook, which are widely credited for Apple’s success and high profitability). If taking a follow-the-money approach to the importance of commercial functions to profitability, R&D might often be way down the list — behind key executive functions, supply chain, possibly marketing, and other functions — where executives are more highly paid and the corporate expenditures are much greater. The choice to write a regulation focused only on the sharing of R&D expenses that has resulted in asymmetric profit-shifting externalities seems ill-informed at best.

5. Successful R&D produces asymmetric returns that are often multiples of the limited R&D deductions moved offshore.

The IRS’s response to the disastrous out-of-control profit shifting produced by these regulations was to try to vastly increase CSA buy-in payments beginning around 2007 in ways that might recoup the future profits into the United States. The IRS tried to extend the scope of the buy-in payment and considered unspecified methods to be the best method for IP buy-ins.⁶² This response was costly and time-consuming, such that the IRS could try to apply it only to a relatively few taxpayer groups. Moreover, the Tax Court and the Ninth Circuit were unsympathetic to the strategy.⁶³ In any event, the regulatory monster was already long out of the bag and loose on the rampage; the tax losses could not be staunched. A complete rethink was needed if Treasury was not going to simply repeal these regulations. The 2008 temporary regulation was that rethink.

This expensive mistake arising from, in effect, betting on the failure of U.S. R&D efforts has been one large factor behind the estimated $100 billion or more in annual tax underpayments from offshore profit shifting, as shown in Figure 1 — estimates that are not included in the IRS tax gap estimate.

According to a former IRS international examiner, the 2008 temporary cost-sharing regulation was so complex and distortive that it was difficult to even read and impossible to enforce. The level of complexity and workload it placed on examiners (some of whom may manage as many as 50 examinations at once) was so great that it was beyond any resources the IRS has ever had. Even the examples in the 2008 regulations were of no use in practice because most implemented CSAs looked nothing like the ones described. For instance, the 2008 cost-sharing regulations contain at least 40 separate examples of various applications of the regulations. In every example, the foreign CSA participant performs as an independent taxpayer its own physical use or exploitation of the CSA-covered IP through that participant’s own production, distribution, or other applicable functions. In many cases, the foreign CSA participant even brings valuable self-developed intangibles that are independent of those developed by the U.S. participant to the arrangement. Despite all these examples, the most common application of the CSA regulations was exactly the opposite — where the U.S. parent performs both the IP development and much or all of the IP exploitation. The CSA mechanism has been used as simply one component of profit-shifting arrangements that siphon profit from mid- and high-tax countries, where real operations take place, to tax haven shell companies⁶⁴ — such as Apple’s arrangement, but also those of Google, Facebook, and others. If ever a tax rule personified a Frankenstein regulation, this is it.

One reason this regulation is so calamitous is that it actually reverses the arm’s-length standard. Take, for example, a paper company with no employees that owns no valuable intangible or tangible assets, makes no operational contributions, and neither manages nor controls any exploitation activities (meaning the existence of other affiliates under its control with sufficient substance to exploit the IP). At arm’s length, that company should have no bargaining power or competitive advantage that would enable it to force the much larger and more capable parent company to pay over the majority of its profits for a simple cash contribution for a fraction of its operating expenses (that is, solely a portion of its R&D costs, which may be a small portion of total operating expenses). The regulation also treats the R&D function as paramount, ignoring in particular the exploitation function that often represents the bulk of other operating expenses after the execution of the CSA. Yes, the current regulations do require payment for platform and other contributions, but in practice, this requirement has in no way created real arm’s-length results. Rather, the CSA regulations provide a tool that has been aggressively used to implement profit-shifting structures that often violate other regulations and statutes, such as effectively connected income and economic substance.

In Apple’s case, between 2009 and 2011 for instance, its total operating expenses (including R&D) were $22.8 billion, but R&D development was only $5.5 billion of that, or about 24 percent. It is an open question why Treasury decided to apply a formula apportionment method (that is, apportionment based on reasonably anticipated benefits) to only that 24 percent of operating expenses and then attribute to that particular subset of costs all revenue and profits from foreign territories that immediately shifted to the foreign CSA affiliate. Apportioning all operating expenses may have been better than the ultimate result in which, in some cases, such as Apple, the R&D allocations were effectively the only expenses apportioned against the shifted foreign revenue.

As to Treasury’s selection of the R&D function as being more important to profits than any other corporate function (because whomever pays for the R&D earns the rights to all the profits, regardless of whether they originated from this R&D), one only need look at the development of the iPhone. This was not Apple’s first attempt to build a smartphone. In fact, as early as 2004, Apple began creating a smartphone in a partnership with Motorola (a maker of wireless handsets) and Cingular (a provider of wireless services), with Apple developing the software. The result was the Motorola Rokr, and it failed in the market. The success of the iPhone in 2007 compared with the failure of the Rokr in 2005 cannot be attributed to any change in R&D activities, but instead to executive decisions by Apple’s CEO.

Jobs realized that to develop the iPhone that he envisioned and to achieve its full potential, he would need to wrest control of the operating system from the wireless carriers. This was because before the iPhone, handset makers had to design their devices to work within the limitations of each carrier’s network, which reduced innovation of the devices. Jobs understood that the device he envisioned would reverse this arrangement and make the handset and its apps more important than the network, which would become a commodity that had to conform to the iPhone’s requirements and operating system. And he was prepared to transform Apple into a wireless carrier to release the company from these design constraints.⁶⁵ These were management decisions — not R&D decisions. It seems clear that Jobs was likely the single most important factor in the creation of the iPhone and Apple’s subsequent business success. He was a management executive who oversaw the design, R&D, production, and sales strategies of the iPhone, and together with COO Cook built a world-leading supply chain to get it to market with a level of speed, quality, and profitability that was truly astounding. Research firm Gartner ranked Apple’s supply chain the best in the world from 2010 to 2013 — a period encompassing 700 percent growth in shipments of the iPhone from 20.7 million to 150.2 million.⁶⁶ Jobs and Cook were management executives. R&D is important, but it is always only a piece of the puzzle.

By assuming that the value of people like Jobs or Cook to profits is approximately zero (or perhaps no more than the cost of a portion of their compensation as assumed in reg. section 1.482-9), and that the functions of the R&D department are worth the entire profits of the enterprise by fiat of a CSA election, the cost-sharing regulations override the arm’s-length principle and are fundamentally flawed to the point of being catastrophic to tax administration. Taxpayers know this and exploit these regulations to transfer windfall profits to tax haven shell companies for bearing only a sliver of the company’s U.S. expenses through a cash contribution the company could frankly obtain from a bank at single-digit interest rates.

Other particularly important functions of companies (for example, creation and day-to-day management of the supply chain; commercialization, including sales activities; and executive functions) were historically (before the 2008 CSA regulations) not explicitly within the scope of the CSA regulations but were covered by other transfer pricing regulations, such as reg. section 1.482-9 and -4. This report shows how that loophole was exploited by Apple Inc. and AOI, capitalizing on the fact that these activities might rarely be examined in an audit of the CSA activity because of these flawed perceptions and examination guidance (such as the coordinated issue paper mentioned earlier). Would Apple ever pay Google, for instance, or two guys working out of a van in Dublin, the majority of its worldwide income for setting up a shell company and running a loan through it? Because that is exactly how many CSAs work.⁶⁷

Note that in a conventional joint development agreement among unrelated parties, each party brings to the table nonroutine, high-value assets that, when combined, create synergies. The arrangement is almost always structured as a profit split, generally aligned with the bargaining power or the relative value of the contributions by each party. An excellent example is a 2001 joint development and commercialization agreement involving Bayer AG and CuraGen Corp. CuraGen was responsible for identifying several gene and protein targets for which Bayer then developed small-molecule compounds that could bind to those targets. Both parties also engaged in commercialization efforts, with profits split in proportion to the costs represented by these widely divergent but nonroutine contributions.⁶⁸ Note that:

• the respective costs were for completely different activities, but included all activities, including commercialization;

• the different activities by each party were each nonroutine and not easily replicated by the other party or by competitors; and

• each party made valuable IP development and exploitation contributions to the arrangement.

This differs significantly from a common intercompany CSA under reg. section 1.482-7, in which only one party might perform all the IP development and a significant portion of the exploitation activities. Highly important contributions to profits (in particular, management and operation of the supply chain and group internet-based platforms) are effectively ignored or given less than cursory recognition by many taxpayers despite those important contributions being subject to other relevant section 482 regulations.

In the case of Apple, it appears that AOI paid almost nothing for the various routine and nonroutine commercialization functions performed within the United States. Another example is Caterpillar’s Swiss strategy, described in connection with a report from the Senate Homeland Security and Governmental Affairs Permanent Subcommittee on Investigations (PSI). Although the profit-shifting structure did not involve a CSA, it was disclosed that an intercompany service agreement valued the U.S. management, supply chain, inventory management, and other major functions as being worth only cost plus 5 percent.⁶⁹ Too often in CSA arrangements, but also in non-CSA contexts (such as IP licensing arrangements), the foreign participants in these transactions can earn the majority of profits from the arrangement while undertaking few of the operational or financial contributions and related expenses that in fact generate the foreign affiliate’s profits.⁷⁰

Real-world taxpayers and their accountants could not fail to see this mistake, yet over the past 30 years or so, Treasury has found it impossible to repair these regulations to produce arm’s-length outcomes. It should be no surprise that academics and practitioners have been calling for the repeal of these regulations for years.⁷¹ A highly esteemed tax policy expert stated the following about these arrangements in 2011:

The most remarkable aspect of the entire structure . . . is the ready acceptance by countries of the fantastic notions that (i) a wholly-owned subsidiary has a mind of its own with which to negotiate “arm’s-length” contractual terms with its parent, (ii) capital provided to the subsidiary by the parent somehow becomes the property of an independent actor (the subsidiary) with which it can take business risks that for tax purposes are not simply assimilated into those borne by the parent (as both provider of the capital and ultimate economic owner of the assets acquired therewith), and (iii) a multinational enterprise that exists as a global platform to exploit a core set of intangible assets best is analogized to wholly independent actors taking on limited and straightforward roles in a vertical chain of production or a horizontal array of distribution of a product. The second and third of these notions in particular transcend the question of transfer pricing.⁷²

At its core, cost sharing is nothing more than a formulary apportionment method applied to split IP development expenses in isolation. However, this comes with a massive externality in which the choice of this method results in the wholesale transfer of the foreign territory revenue and profits instantly to the foreign shell company — even if the IP development function is not the most important contributor to profits.⁷³

Unlike the monster created by Dr. Frankenstein, which eventually gets onto a ship and sails into oblivion, these cost-sharing regulations shamble along and won’t die or leave. Nevertheless, Treasury has continued to work on them, and in 2008 made two important changes that were just ingenious enough that they could — if enforced — reverse at least two of the most disastrous aspects of the regulations. The first was the addition of the reg. section 1.482-7T(m) transition rules, which were clearly designed to end the no-substance CSAs with shell companies. The second was the addition of the reg. section 1.482-7T(i)(6) periodic adjustment rules to implement a CSA-specific commensurate-with-income rule, which simply limited the amount of profits that could be transferred offshore through a CSA. Perhaps the monster could be tamed?

B. The 2008 CSA Transition Rules

The 2008 temporary cost-sharing regulations were issued as T.D. 9441 on December 31, 2008, and published in the Federal Register January 5, 2009. They contained the following transition rules, which remained unchanged in the final regulations (emphasis added):

Section 1.482-7T(m) Transition rule — (1) In general. An arrangement in existence on January 5, 2009, will be considered a CSA, as described under paragraph (b) of this section, if, prior to such date, it was a qualified cost sharing arrangement under the provisions of section 1.482-7 (as contained in the 26 CFR part 1 edition revised as of January 1, 1996), hereafter referred to as “former section 1.482-7”), but only if . . . the activities of the controlled participants substantially comply with, the provisions of this section . . . by July 6, 2009.

(2) Transitional modification of applicable provisions. For purposes of this paragraph (m), conformity and substantial compliance with the provisions of this section shall be determined with the following modifications: (i) CSTs and PCTs occurring prior to January 5, 2009, shall be subject to the provisions of former section 1.482-7 rather than this section. (ii) Except to the extent provided in paragraph (m)(3) of this section, PCTs that occur under a CSA that was a qualified cost sharing arrangement under the provisions of former section 1.482-7 and remained in effect on January 5, 2009, shall be subject to the periodic adjustment rules of section 1.482-4(f)(2) rather than the rules of paragraph (i)(6) of this section. [This is not applicable to Apple because it was never subject to the former reg. section 1.482-7 — see discussion below.]

For a preexisting CSA to successfully transition to the 2008 regulations and become exempt from the reg. section 1.482-7(i)(6) periodic adjustment rules in those regulations for PCTs that occurred before January 5, 2009, and afterward, the taxpayer would have had to do the following (described here as parts 1 and 2):

1. establish that the CSA before January 5, 2009, was a qualified CSA meeting all the requirements of “former” reg. section 1.482-7 (26 CFR part 1 edition revised as of January 1, 1996); and

2. amend by July 6, 2009, the existing CSA in effect on January 5, 2009, to meet the new requirements of the 2008 temporary regulations, except for specific requirements as described in reg. section 1.482-7(m)(2).

The remainder of this section focuses on what appears to be noncompliance with the reg. section 1.482-7T(m)(1) and (2) rules that subject Apple’s preexisting and post-2008 PCTs to the reg. section 1.482-7(i)(6) periodic adjustment rules.

C. What Does ‘Former’ Mean?

Note that the word “former” in the 2008 temporary regulations is not used colloquially; this is a highly defined use, designating a particular regulation in effect on a particular date (the 1995 regulations in 26 CFR part 1 existing as of January 1, 1996). If one instead mistakenly substituted the colloquial meaning of the term, “former” would have been interpreted to mean the regulation in effect just before the January 5, 2009, effective date of these December 2008 temporary regulations. This is an important point because some advisers may not have recognized this, as demonstrated by the loose language in some of their published guidance that substituted other words such as “old” in place of “former” with no definition or qualification, or that mistakenly attached the colloquial meaning to “former.” Here are two examples:

• “A CSA in existence on 5 January 2009 may be grandfathered, in part, if, prior to this date, it was a qualified CSA under section 1.482-7 (the old regulations).”⁷⁴

• “Transition for pre-existing CSAs. An arrangement in existence on January 5, 2009, will be considered a qualifying CSA, if it was a qualified CSA under the provisions of former regulation section 1.482-7.”⁷⁵

Given these loose readings of “former” by some commentators, it seems likely that some taxpayers believed their preexisting CSAs would be grandfathered automatically. It appears that the primary reason the drafters of the 2008 reg. section 1.482-7(m) transition regulations specifically chose the 1995 regulations and the fixed date of January 1, 1996, as the primary compliance requirement for transition was the participation requirements in paragraphs (c)(1)-(3). This is critical because compliance with these regulations is mandatory for taxpayers seeking to avoid exposing their past and future PCTs to a periodic adjustment under reg. section 1.482-7(i)(6). The facts in Apple’s situation show just how costly such an oversight could be.

D. Apple CSA Transition Failure

The January 1, 1996, “former” version of the CSA regulations (referenced by the 2008 temporary cost-sharing regulations in reg. section 1.482-7T(m)(1)) had been issued in final form on December 20, 1995, in T.D. 8632, and was effective for tax years beginning on or after January 1, 1996, for taxpayers without preexisting CSAs.⁷⁶ Taxpayers such as Apple with a preexisting CSA would continue to have a qualified CSA only if the CSA was a “bona fide CSA” under the prior regulation (as that regulation was in effect on April 1, 1995), and if that CSA was conformed to meet the requirements of the new regulation by December 31, 1996. This regulation had been proposed in 1992 and was issued with some changes after Treasury received numerous comments.

The new regulations promulgated in T.D. 8632 presented two particularly troubling provisions for taxpayers like Apple. First, they made the explicit point in paragraph (a)(1) that a CSA must involve parties that “agree to share the costs of development of one or more intangibles in proportion to their shares of reasonably anticipated benefits from their individual exploitation of the interests in the intangibles assigned to them under the arrangement” (emphasis added). The narrow explanation accompanying the original proposed regulation released in 1992 indicated that this was intended to restrict “membership in a cost sharing arrangement to those in a position to exploit developed intangibles by means of the manufacture of products.” Second, the new regulations included guidance, found in reg. section 1.482-7(c), for determining when a controlled entity could be a participant in a CSA.

In considering this individual exploitation requirement and the participant rules, keep in mind what was discussed in Section III.B above that taxpayers such as Apple (which, of course, wanted its preexisting CSAs to successfully transition to the 2008 regulations) had to ensure that the January 2009 facts concerning their CSAs met the requirements for CSAs as those requirements existed on January 1, 1996. Factual conditions can change significantly over a 13-year period. What might have been reasonably, or even clearly, within the requirements in 1996 may no longer be true in 2009.

The paragraph (c) participant guidance reads in part:

(c) Participant —

(1) . . . A controlled taxpayer may be a controlled participant only if it — (i) Uses or reasonably expects to use covered intangibles in the active conduct of a trade or business, under the rules of paragraphs (c)(2) . . . of this section. . . .

(2) Active conduct of a trade or business — . . . (ii) Active conduct. In general, a controlled taxpayer actively conducts a trade or business only if it carries out substantial managerial and operational activities. For purposes only of this paragraph (c)(2), activities carried out on behalf of a controlled taxpayer by another person may be attributed to the controlled taxpayer, but only if the controlled taxpayer exercises substantial managerial and operational control over those activities. [Emphasis added.]

Notably, these participant requirements were removed approximately five months later through an amendment to the cost-sharing regulations issued in T.D. 8670 on May 13, 1996. However, this amendment is irrelevant to the 2008 regulation transition rules of reg. section 1.482-7T(m)(1) because this provision cites these earlier regulations as they existed as of January 1, 1996, which, of course, ignores later amendments, such as those in T.D. 8670.

Treasury clearly sought to make it difficult if not impossible for taxpayers with low-substance/tax shelter-type CSAs to be grandfathered under the 2008 regulations. The IRS, however, appears to have never enforced these regulations.

Solely for purposes of this discussion, it is conservatively assumed that in 1996, which was before the issuance of the entity classification check-the-box rules, Apple’s CSA with AOE or its applicable predecessor (it is understood that ASI did not join the CSA until 1999) met all applicable requirements, including both the individual exploitation requirement and the reg. section 1.482-7(c) participant requirement (after amendment by T.D. 8670). As of January 2009, however, the manner in which Apple conducted its international business had significantly changed, both factually and structurally. Both ASI and AOE were now disregarded entity subsidiaries of AOI, meaning that AOI and all its disregarded entity subsidiaries must be treated as one CFC conducting business through branches in Ireland and various other countries around the world. Further, whereas in 1996 many of the Apple products distributed into Europe and other foreign markets may have been manufactured in its Cork, Ireland facility, by January 2009 the bulk of products “manufactured” by AOI and sold to customers in its defined territory were sourced by engaging Asian contract manufacturers that were arranged for, directed, and controlled by Apple Inc. personnel in the United States. It appears factually that no AOI personnel in Ireland or elsewhere conducted “substantial managerial and operational control” over these Apple Inc. personnel. (Specific evidence of some of these facts are set out in Section III.E below.)

With this changed situation by January 2009, it seems certain that AOI could not in any manner be seen to meet either the individual exploitation requirement or the participant requirement as in effect on January 1, 1996, before the T.D. 8670 amendment.

Regarding the individual exploitation requirement, although AOI and its disregarded entity subsidiaries do have thousands of employees and considerable operations, including limited manufacturing in its Irish facilities, AOI in no sense conducts its business in the independent manner that individual exploitation would require. AOI itself neither manufactures the vast majority of the products it sells nor manages and directs in any way the contract manufacturers that physically produce its products. It also in no way manages or controls Apple Inc., which, on AOI’s behalf, conducts all production activities short of physical production. That AOI does not meet this individual exploitation requirement is fully consistent with the earlier-mentioned explanation accompanying the regulation when originally proposed in 1992.

Regarding the participant requirement, a controlled taxpayer must carry out substantial managerial and operational activities. There is no question that disregarded entity subsidiaries of AOI conduct substantial operational activities. These activities include functions such as sales and marketing, customer support, retail stores, and limited manufacturing in Ireland. There will, of course, be some localized and regional management for these various functions. However, with the information provided in the European Commission’s 2016 decision, the 2013 Senate report, and the 2020 GCEU opinion — which include Apple’s representations about the conduct of its business in Ireland — it is clear that AOI in January 2009 did not conduct substantial managerial activities over critical portions of its own business, including, for example, management over component purchasing and contract manufacturers; management of the provision of cloud services and the various internet-based platforms through which AOI sells to customers and provides them with cloud services; and sales management, particularly for key customers such as cellular networks and major resellers.

It is also important to note the sentence in paragraph (c)(2) requiring that “activities carried out on behalf of a controlled taxpayer by another person may be attributed to the controlled taxpayer, but only if the controlled taxpayer exercises substantial managerial and operational control over those activities.” (Emphasis added.) Clearly, AOI exercises no such managerial or operational control over Apple Inc. and its U.S. affiliates in the activities they conduct on AOI’s behalf. The reality is that AOI’s substantial managerial activities were conducted in Cupertino and not by AOI.

It is worth saying more about the above comment that AOI disregarded entity subsidiaries conduct some localized and regional management for some functions. Of course, these AOI subsidiaries do have appropriate management personnel responsible for the localized portions of Apple’s business that they manage. However, these personnel do not have broad authority over all of AOI’s worldwide business, and they operate only under the close direction and supervision of Cupertino management, who direct and participate in the conduct of AOI’s business. The Cupertino-based management and operations go far beyond just group policy direction and the mere oversight and stewardship functions that a parent company normally conducts for its independently run subsidiaries. Further, some AOI business and operational functions (in particular, the production functions and the operation of worldwide internet-based app stores and other platforms through which AOI sells products and provides cloud services) are conducted mostly or wholly by Cupertino-based personnel on behalf of AOI. In no way may it be said that in January 2009 AOI was “exercis[ing] substantial managerial and operational control over those activities” conducted on its behalf by Apple Inc. and other U.S. group members as required by former reg. section 1.482-7(c)(2) as in effect on January 1, 1996.

E. CSA and Evidence of Transition Failure

Although Treasury later defended Apple’s transfer pricing and claimed that it was unable to examine assertions of tax violations by the European Commission, the Senate did take issue with Apple’s tax structure and transfer pricing and tried to investigate it, as did the commission. These two investigations each disclosed portions of Apple’s restated CSA, shown below in Figure 2, reconstructed from the Senate and commission disclosures.

This document was prepared by Apple and signed in June 2009 to meet the 2008 reg. section 1.482-7T(k)(1) and (m)(1) requirement that taxpayers execute written contracts that reflect the new 2008 regulation requirements by July 6, 2009. Apple would have also presumably provided the IRS with a CSA statement under reg. section 1.482-7T(k)(3) and (m)(2)(viii) no later than September 2, 2009.

Figures 8 and 9 shown below in Apple’s restated CSA contract were disclosed in the 2016 European Commission decision. The commission requested that Ireland provide information on the ASI and AOE activities represented by the functions and risks presented in figures 8 and 9, as well as concrete examples of those activities. Ireland and Apple responded to that request with a letter from Apple dated April 22, 2016.⁷⁷ The European Commission described the content of that letter as follows (with one footnote omitted and emphasis added):

Ireland and Apple indicated that the tables from the CSA of 2009 . . . do not purport to show activities actually performed by the parties to the CSA but merely summarise the activities that each party is authorised to perform under that agreement. Apple further explained that the tables . . . were added to the CSA to satisfy the requirements under temporary regulations issued by the US Treasury Department, effective 5 January 2009, that a cost sharing agreement among related parties reflect the parties’ functions and risks.⁷⁸

Apple further indicated that the . . . activities listed in the table reproduced in Figure 8 . . . were performed almost entirely by Apple Inc. employees in the US.⁷⁹

Apple further stated that ASI and AOE do not have any role in the management of Apple’s IP and that they have only a limited role, within the strict parameters set by Apple Inc. executives in the US, in commercial contract negotiations. . . . All the functions that drive Apple’s profits are directed by Apple Inc. executives in the US and performed largely in the US.⁸⁰

A plain reading of Apple’s disclosure to the European Commission would indicate that its June 25, 2009, restated CSA contract was simply false. Many of the activities Apple told the IRS that ASI and AOE factually perform were actually performed by Apple Inc.

The July 15, 2020, decision of the GCEU⁸¹ provides further evidence of Apple’s representations:

It is also apparent from that evidence that the strategies relating to new product launches and, in particular, the organisation of distribution on the European markets in the months leading up to the proposed launch date had been managed at the Apple-Group level by, inter alia, the Executive Team under the direction of the Chief Executive Officer in Cupertino.⁸²

In addition, it is apparent from the file that contracts with third-party original equipment manufacturers (“OEMs”), which were responsible for the manufacture of a large proportion of the products sold by ASI, were negotiated and signed by the parent company, Apple Inc., and ASI through their respective [U.S.-based] directors,⁸³ either directly or by power of attorney. ASI and AOE also submitted evidence regarding the negotiations and the signing of contracts with customers, such as telecommunications operators, which were responsible for a significant proportion of the retail sales of Apple-branded products, in particular mobile phones. It is apparent from that evidence that the negotiations in question were led by directors of the Apple Group and that the contracts were signed on behalf of the Apple Group by Apple Inc. and ASI through their respective [U.S. based] directors, either directly or by power of attorney. . . .⁸⁴

Apple Inc., ASI and AOE submitted, in the context of both the administrative procedure and the present action, evidence demonstrating that the framework agreements with the manufacturers of Apple-branded products (or OEMs) had been concluded centrally in respect of the Apple Group as a whole by Apple Inc. and ASI in the United States.⁸⁵

Recall from Section III.D above that the regulation in effect on January 1, 1996, required that each participant conduct “individual exploitation of the interests in the intangibles assigned to them under the arrangement.” Based on the above, Apple has emphasized in its defense to the European Commission that “all the functions that drive Apple’s profits are directed by Apple Inc. executives in the US and performed largely in the US.” This is not just top-level management direction. Rather, as the above shows, the GCEU accepted as fact Apple’s representations that two of the most critically major aspects of AOI’s business were conducted from Cupertino: (1) the negotiations and the signing of contracts with major telecommunications customers, and (2) the negotiations and signing of contracts with third-party OEMs (that is, the Asian contract manufacturers that produced iPhones and other products for Apple’s worldwide needs). This is not the picture of a subsidiary that conducts individual exploitation of its share of IP. Instead, it looks as if the Irish CSA participants are not involved in exploitation at all, and the entire structure is a sham.

Also recall that to be eligible to be grandfathered under the reg. section 1.482-7(m) transition rule, AOI must conduct substantial managerial and operational activities. Yes, AOI through its disregarded entity subsidiaries does have some thousands of employees in Ireland and in many other countries. It does perform substantial operational activities. However, it is clear from these representations to the European Commission and to the GCEU that in January 2009 AOI in no way conducted substantial managerial activities within the meaning of reg. section 1.482-7(c) as in effect on January 1, 1996, related to exploitation of CSA-covered IP.

F. Effect of Transition Failure

The discussion in the prior two sections shows that the two Apple CSAs were ineligible to be grandfathered on January 5, 2009, under the reg. section 1.482-7(m)(1) transition rule. As such, on the surface, the Apple Inc./AOI relationship is not governed by reg. section 1.482-7 and is instead to be subject to other applicable section 482 regulations.

Interestingly, Apple affixed the following label to its revised CSA (highlighting by the authors), purporting that its CSA was indeed grandfathered, though clearly it did not meet the requirements to be so⁸⁶:

Despite this “surface result” of being covered not by reg. section 1.482-7 but by other applicable section 482 regulations, it is clear that Apple has treated the Apple Inc./AOI CSAs as being valid for several years and that the IRS has not challenged their purported validity. With this being the case, Apple’s clear failure to meet the reg. section 1.482-7(m)(1) transition rule can only mean that the Apple Inc./AOI CSA must be considered a new CSA covered by the 2008 regulations with effect from January 5, 2009. As a result of this and earlier analyses, all of Apple’s PCTs, including those in prior years, are subject to the periodic adjustment regulations of reg. section 1.482-7T(i)(6). This is critically important because the profits booked by AOI may have exceeded the commensurate-with-income limitations of reg. section 1.482-7(i)(6) since 2009 by substantial margins. That result and the calculations behind it are discussed in Part 2 of this report.