Tax Notes logo

Coca-Cola: A Decisive IRS Transfer Pricing Victory, at Last

Posted on Dec. 14, 2020
[Editor's Note:

This article originally appeared in the December 14, 2020, issue of Tax Notes Federal.

Gianluca Mazzoni
Gianluca Mazzoni
Reuven S. Avi-Yonah
Reuven S. Avi-Yonah

Reuven S. Avi-Yonah ( is the Irwin I. Cohn Professor of Law at the University of Michigan Law School in Ann Arbor. Gianluca Mazzoni ( received his SJD in international tax in 2020 from the University of Michigan Law School.

In this article, Avi-Yonah and Mazzoni examine the Tax Court’s decision in Coca-Cola, the first decisive IRS victory in a major transfer pricing case since 1979.

Copyright 2020 Reuven S. Avi-Yonah and Gianluca Mazzoni.
All rights reserved.

Coca-Cola1 is the first decisive IRS victory in a major transfer pricing case since 1979. If not reversed on appeal, the outcome will mark an important shift in U.S. transfer pricing litigation and perhaps indicate that the IRS could win other major pending cases, such as the one against Facebook.


In 1979 the IRS won a decisive victory in a 20-year transfer pricing struggle with DuPont, involving DuPont’s attempt to use a Swiss base company to shift profits from the United States and high-tax European countries to Switzerland.2 The IRS won the case by discovering a “smoking gun” memorandum that laid out the taxpayer’s tax avoidance strategy, concluding that the worst-case scenario would lead to payment of the tax plus interest at a rate below what the taxpayer could earn on its funds in the interval.

Between 1979 and 1994, the IRS consistently lost every major transfer pricing case it litigated, including those against U.S. Steel Corp., Bausch & Lomb Inc., HCA Healthcare, Eli Lilly and Co., G.D. Searle LLC, Ciba-Geigy AG, Sundstrand Corp., and Merck & Co. Inc.3 After the new transfer pricing regulations were issued in 1994, there was a hiatus in transfer pricing litigation. When cases resumed, the IRS continued losing, including against DHL Corp. (1998), UPS (1999), Compaq (1999), Xilinx Inc. (2005), Veritas Software Corp. (2009), Medtronic Inc. (2016), and Inc. (2017).

In the last few years, there have been signs that the IRS litigation effort is improving. In 2018 Medtronic was reversed on appeal and remanded to the Tax Court, and in Altera the IRS, having lost decisively in the Tax Court, obtained a reversal on appeal in 2019, a decision that became final when the Supreme Court denied certiorari this year.

Even so, the prospects of a decisive IRS victory in a transfer pricing case remained doubtful. It is unclear what will happen to Medtronic on remand, and while Altera was a major victory, it involved a narrow issue (whether the cost of stock options must be included in the pool of costs shared under a cost-sharing agreement). But then came Coca-Cola, which is unequivocally an IRS victory in a major case (it involved a $3.3 billion deficiency).

The Case

Coca-Cola is a U.S. company with its headquarters and principal place of business in Atlanta. It is the legal owner of all the intellectual property necessary to manufacture, distribute, and sell some of the best-known nonalcoholic, ready-to-drink beverages in the world. Coca-Cola’s IP includes trademarks, product names, logos, patents, secret formulas, and proprietary manufacturing processes. Its international structure includes several foreign manufacturing affiliates (supply points), local service companies (ServCos), and independent bottlers. Coca-Cola licenses the IP to the supply points to produce beverage concentrate.4 License agreements can be terminated by Coca-Cola at will without compensation, do not grant any form of territorial exclusivity to the supply points, and do not give the supply points any ownership interest in the IP. Supply points sell concentrate to unrelated bottlers, which produce finished beverages for sale to distributors and retailers worldwide.

The figure represents how Coca-Cola’s operations were conducted during the relevant tax years (2007-2009).

Coca-Cola's Operations During 2007-2009

During the relevant years, the supply points compensated Coca-Cola for the use of its IP under a formula that had been agreed on between Coca-Cola and the IRS when settling a tax audit in 1996 for the 1987-1995 tax years. The formula permitted the supply points to retain profit equal to 10 percent of their gross sales, with the remaining profit split 50-50 with Coca-Cola. However, Coca-Cola’s closing agreement with the IRS did not address what transfer pricing method would have been used after 1995, and Coca-Cola continued to use the 10-50-50 formula to report income from its foreign supply points.

For the relevant tax years, the IRS claimed that the 10-50-50 formula was not arm’s length because it overcompensated the supply points and undercompensated Coca-Cola for use of the IP. Invoking section 482, the IRS reallocated income to Coca-Cola from its foreign supply points using a comparable profits method, treating the independent Coca-Cola bottlers as comparable parties. To implement its bottler CPM, the IRS determined the average return on operating assets (ROA) for a group of independent Coca-Cola bottlers it deemed comparable. It applied that average ROA to the operating assets of each supply point, generating a deemed arm’s-length operating profit, then reallocated to Coca-Cola all income received by each supply point in excess of that benchmark.

Coca-Cola challenged the section 482 reallocations as arbitrary and capricious. First, it contended that the IRS acted arbitrarily by abandoning the 10-50-50 formula, having acquiesced in the use of that formula during five prior audit cycles spanning a decade. Second, it said the IRS erred in using the bottler CPM to reallocate income. Coca-Cola relied on a provision in the closing agreement that it said should have some prospective operation because it granted Coca-Cola penalty protection for future years:

For taxable years after 1995, to the extent the Taxpayer applies the [10-50-50] method to determine the amount of its reported Product Royalty income with respect to existing or any future Supply Points, the Taxpayer shall be considered to have met the reasonable cause and good faith exception of sections 6664(c) and 6662(e)(3)(D) and shall not be subject to the accuracy-related penalty under section 6662 with respect to the portion of any underpayment that is attributable to an adjustment of such Product Royalty.

The Tax Court refused that argument, finding the provision hurt, rather than helped, Coca-Cola.5 It said the agreement recognized that the IRS might make transfer pricing adjustments after 1995 because it gave Coca-Cola penalty protection for the portion of any underpayment “attributable to an adjustment of such Product Royalty.”

According to the court:

The short and (we think) the complete answer to petitioner’s argument is that the closing agreement says nothing whatever about the transfer pricing methodology that was to apply for years after 1995. Parties to a closing agreement may (and sometimes do) bind themselves to particular tax treatments for specified future years. For its part, petitioner may have desired the certainty that would arise from indefinite future application of the 10-50-50 method. But there is no evidence in the document that the IRS shared that desire or agreed to implement it.

In arguing that the CPM is inferior to other methods for pricing transfers of intangible property, Coca-Cola relied on a statement in the preamble to the 1994 final transfer pricing regulations (T.D. 8552) referring to the CPM as “a method of last resort.”6 The court rejected that argument, saying the best method rule must be applied based on the availability of adequate data. It referred to Example 4 of reg. section 1.482-5(e), which treats the CPM as the best method for determining an arm’s-length royalty for the transfer of intangibles to a foreign affiliate that performs routine manufacturing functions as the supply points did. It said in this case, “the circumstances that caused Treasury to refer to the CPM as a ‘method of last resort’” did not exist.

The court also held that none of the three alternative methods — the comparable uncontrolled transaction, residual profit-split (RPSM), and asset management methods — was the best method for the Coca-Cola case.

Regarding the first alternative method, the court said Coca-Cola did not identify any pricing data for transactions with unrelated parties that involve the transfer of the same intangible, so the CUT method could not be used. Sanjay Unni, one of Coca-Cola’s expert witnesses, derived his support for the CUT method from master franchising transactions that fast food chains like McDonald’s and Domino’s Pizza entered into with regional franchisees abroad. The court rejected Unni’s analysis for several reasons. First, the supply points were not responsible for Coca-Cola foreign businesses, including managing and overseeing the franchise bottlers, and were not responsible for consumer marketing activities and expenditures to exploit and develop Coca-Cola’s intangibles. Second, supply points could not be plausibly analogized to master franchisees. The master franchisee agreements had long terms (10 to 50 years) and conferred territorial exclusivity. In contrast, supply points had short-term agreements7 that Coca-Cola could (and often did) terminate at will and that did not confer any territorial exclusivity. Third, supply points manufactured beverage concentrate and did not play any role in managing the franchise, selecting subfranchisees, or overseeing bottlers in any geographic territory. All those activities were performed by Coca-Cola and the ServCos. Finally, bottlers could not be analogized to subfranchisees of the supply points because their agreements invariably ran with Coca-Cola, not with the supply points. Bottlers received direction and marketing assistance from the ServCos, not from the supply points. Thus, bottlers could not be compared with owners of restaurants that serve consumers.

The court also rejected the second alternative method for several reasons, including three in particular. First, the supply points did not perform the economic functions that created valuable IP, such as implementing consumer advertising and engaging in franchise leadership with bottlers — the ServCos did. Second, the supply points were neither the legal owners of IP under the IP laws of the relevant jurisdictions nor holders of rights constituting IP under contractual terms (such as the terms of a license) or other legal provisions. Coca-Cola owned the IP involved in the transactions at issue. Third, Coca-Cola’s application of the RPSM was solely based on relative consumer advertising expenditure in the relevant tax years without factoring in the relative value of nonroutine IP contributed by Coca-Cola and the supply points. The court stated:

Wholly apart from past advertising expenses, however, [Coca-Cola] obviously brought to the table many other valuable intangibles — its brands, trademarks, tradenames, patents, logos, secret formulas, and proprietary manufacturing processes. Consumer advertising is worth little unless the seller has a product that people wish to buy.

The Tax Court thus found Coca-Cola’s proposed RPSM a wholly unreliable method. It also said that even without the above flaws, Coca-Cola’s proposed RPSM results would have had a low level of reliability because the value of nonroutine intangibles was estimated by the capitalized cost of developing the IP less an appropriate amount of amortization based on the useful life of each IP rather than being measured by external market benchmarks that reflect the IP’s fair market value.8 It referred to reg. section 1.482-6(c)(3)(ii)(C)(3), which says the soundness of an RPSM depends on the “reliability of the data used and the assumptions made in valuing the IP contributed by the participants. . . . In particular, if capitalized costs of development are used to estimate the value of IP, the reliability of the results is reduced relative to the reliability of other methods that do not require such an estimate,” and reg. section 1.482-6(c)(3)(ii)(D), which states that, “To the extent the allocation of profits in the second step is not based on external market benchmarks, the reliability of the analysis will be decreased.”

According to the court, Coca-Cola’s capitalization of intangible development costs yielded unreliable results for at least two reasons. First, that required assumptions regarding the useful life of the IP. The court said, “There is no consensus among economists that ordinary advertising costs can be properly be capitalized as an intangible asset, much less about what the useful life of such an asset would be.” Second, those costs might not be related to the IP’s market value. The court noted that the intangible described by one of Coca-Cola’s experts was a capitalization of the historical costs of advertising Coca-Cola products, saying:

No unrelated party would pay a supply point a meaningful sum for this supposed asset, because the asset could not be usefully deployed by an unrelated party. In any event, this asset could not be deployed by an unrelated party without violating petitioner’s trademarks.

Finally, the court rejected Coca-Cola’s proposed unspecified, or asset management, method, which was based on the two-tiered fee structure typically used to compensate asset managers in the financial services sector. It said that method did not remotely resemble any of the specified methods for valuing intangibles under the section 482 regulations and compensated Coca-Cola only for asset management services involving governance, sharing of best practices, and high-level strategy, thus ignoring the relevant intangibles (for example, brands, trademarks, and secret formulas). The court noted that asset managers typically do not supply those kinds of intangibles to the portfolio companies they manage.

In conclusion, the court determined that the CPM was the best method because only Coca-Cola owned the valuable IP and the supply points had no substantive contractual rights to it.9 It also found that the unrelated bottlers were appropriate comparable parties and that the ROAs were computed using reliable data, assumptions, and adjustments. It therefore decided in favor of the IRS.


The most surprising aspect of Coca-Cola might be the IRS’s successful use of the bottlers as comparables under the CPM. On the face of it, the supply points that manufacture the concentrate and the bottlers that mix it with water and bottle the resulting soft drinks seem to have little in common because they are at different levels of the production chain. However, the key distinction is that while the supply points are controlled (as they must be, because the Coca-Cola formula is not patent-protected and its value would be lost if disclosed to unrelated parties), the bottlers are independent. Thus, if one accepts that the entire value of the drinks lies in the IP owned by the U.S. parent, one could argue that the supply points, like the bottlers, perform a mechanical function and should not be rewarded more than contract manufacturers. As the Tax Court explained:

We agree with the Commissioner’s conclusion that independent Coca-Cola bottlers serve as appropriate comparable parties for purposes of a CPM/ROA analysis. The bottlers are comparable to the supply points because they operated in the same industry, faced similar economic risks, had similar (but more favorable) contractual and economic relationships with petitioner, employed in the same manner many of the same intangible assets (petitioner’s brand names, trademarks, and logos), and ultimately shared the same income stream from sales of petitioner’s beverages.

According to the court, the key point was that “the manufacturing activity in both cases was routine, consisting largely of mixing ingredients according to detailed protocols supplied by petitioner.” Quality control (which was important in Medtronic) was similar in both cases.

The Tax Court acknowledged that in some ways the bottlers were different from the supply points. The bottlers had long-term contracts and territorial exclusivity, while the supply points could be (and sometimes were) terminated at will and had no territorial rights. But the court concluded that those differences should have led to a higher return on assets for the bottlers, while in fact they had much lower ROAs, reflecting the allocation of profits from IP to the supply points. It said those profits should have been allocated to the parent.

Marketing Intangibles

The most important difference between the IRS and Coca-Cola’s positions involved marketing intangibles. The taxpayer argued that the supply points’ higher returns were justified because they bore most of the costs of marketing the products, and that without marketing the IP was valueless. That was similar to the position the IRS took in Glaxo, which involved a drug manufactured in the United Kingdom and marketed in the United States, and resulted in the payment of a $3.4 billion settlement (the largest transfer pricing adjustment in history).10

The court rejected that argument because the supply points did not in fact conduct any marketing — they were just allocated marketing costs. All the actual marketing was done by the parent and ServCos, which were stipulated to be compensated at arm’s length:

Petitioner contends that the supply points bore “marketing risk” because they funded consumer advertising in foreign markets. But the supply points had no operational responsibility for consumer marketing; they thus bore no risk in the sense of “mission failure.” Rather, petitioner simply charged certain ServCos marketing expenses to the supply points’ books, and it made these charges roughly concurrently with the supply points’ receipt of vastly larger amounts of income from the bottlers. Petitioner controlled how much revenue each supply point received (by shifting concentrate production among them) and how much expense each supply point was charged (by way of [direct marketing expenses] and “fees and commissions” allocated to it). Since the flow of revenue and marketing expenses to the supply points was controlled by [the parent], and since the revenue invariably exceeded the marketing expenses by a very wide margin, we do not see how the supply points bore “marketing risks” in any realistic sense. Risk is not something that can be assigned after the fact.

That point is likely to be the main focus of an appeal because the IRS previously recognized the importance of the marketing function. However, it might be difficult to overcome the stipulation that the ServCos were compensated at arm’s length for the actual marketing functions they performed.


In 1993 the IRS introduced advance pricing agreements as a way to manage transfer pricing disputes. APAs have now been adopted all over the world and are a good way to reduce transfer pricing litigation. The problem, however, is that APAs are still used only by a minority of U.S.-based multinationals because the litigation record has so favored taxpayers in large transfer pricing cases that entering into an APA felt like leaving money on the table.

If Coca-Cola is upheld on appeal, that situation could change. Perhaps more large U.S. multinationals would enter into transfer pricing APAs, as well as binding arbitration under treaties. It may also presage more IRS victories in cases like Facebook,11 especially because the IRS won in the Tax Court, where most large transfer pricing cases begin. The IRS can now apply its knowledge to litigate other cases.

As has been shown, judges tend to take the government’s revenue needs into account in deciding large tax cases.12 With the budget deficit exploding as a result of the COVID-19 pandemic and more urgent needs on the horizon, Coca-Cola may be a harbinger of more taxpayer defeats.


1 Coca-Cola Co. v. Commissioner, 155 T.C. No. 10 (2020).

2 E.I. DuPont de Nemours & Co. v. Commissioner, 608 F.2d 445 (Ct. Cl. 1979).

3 Reuven S. Avi-Yonah, “The Rise and Fall of Arm’s Length: A Study in the Evolution of U.S. International Taxation,” University of Michigan, Law & Economics Olin Working Paper No. 07-017 and Public Law Working Paper No. 92 (Sept. 27, 2007).

4 The case involved income adjustments from supply points in Brazil, Chile, Costa Rica, Egypt, Ireland, Mexico, and Swaziland. The Irish and Brazilian supply points accounted for roughly 85 percent of the disputed income adjustments. All except the Mexican supply point, which operated as a branch so its income was reported on Coca-Cola’s U.S. consolidated return, were controlled foreign corporations. Therefore, for the Mexican supply point, the IRS sought to reduce Coca-Cola’s foreign tax credits on the ground that the branch had reported insufficient royalty expenses for the use of Coca-Cola’s intangible property, thus artificially inflating its income and the Mexican corporate tax paid. The IRS contended that the Mexican taxes were to that extent noncompulsory payments ineligible for the FTC. The Tax Court concluded in an earlier decision that the taxes were creditable because the taxpayer met both prongs of the compulsory test. See Coca-Cola Co. v. Commissioner, 149 T.C. 446 (2017).

5 For comparison, see Eaton v. Commissioner, T.C. Memo. 2017-147, which involved an IRS departure from past closing agreements (and advance pricing agreements). Under Eaton, the IRS cannot cancel an APA based on: (i) a perceived material factual omission or misrepresentation during APA negotiations if the taxpayer disclosed all information it reasonably believed relevant and responded thoroughly to all questions asked; or (ii) the taxpayer’s immaterial and inadvertent APA compliance errors that it promptly reported and attempted to correct in good faith. For comment, see Caplin & Drysdale Chtd., “Different Viewpoint Not a Misrepresentation: Tax Court Holds IRS Abused Its Discretion in Cancelling Eaton’s APAs” (Aug. 24, 2017).

6 In the original 1993 proposed transfer pricing regulations, the CPM was designed as a super-method to check the application of the other methods. However, that led to vehement protests from trading partners, who claimed the CPM was not a valid application of the arm’s-length standard. Thus, the CPM was denigrated in the final regulations. See Avi-Yonah, supra note 3.

7 Aside from the Brazilian agreement, which ran indefinitely but could be terminated by Coca-Cola’s unilateral action or either party’s breach of contract, and the Costa Rican agreement, which had an initial two-month term, all supply point agreements had an initial 12-month term and were renewed automatically for one-year periods absent prior notice from Coca-Cola or the supply point. Under the agreements with Ireland, Mexico, and Swaziland, during any 12-month term, either party could terminate for any reason on giving 30 or 60 days’ notice.

8 See reg. section 1.482-6T(c)(3)(i)(B)(2).

9 See Ryan Finley, “Tax Court’s Coca-Cola Ruling: Early Sign of a New Approach?” Tax Notes Federal, Dec. 7, 2020, p. 1651.

10 In GlaxoSmithKline Holdings (Americas) Inc. v. Commissioner, No. 5750-04 (T.C. 2006), in which the original asserted deficiency was $10 billion, the key issue was marketing intangibles used in the global pharmaceutical business. The IRS made virtually the opposite argument — that the U.S. subsidiary of the foreign parent was the economic owner of the marketing intangibles. The case settled transfer pricing disputes covering tax years going back as far as 1989, and the parties reached a separate agreement to resolve transfer pricing issues covering tax years 2001 through 2005.

11 See Finley, supra note 9, and Finley, “Facebook Rejects Aggregate IP Valuation in Transfer Pricing Case,” Tax Notes Federal, Feb. 10, 2020, p. 991.

12 Nancy Staudt, The Judicial Power of the Purse: How Courts Fund National Defense in Times of Crisis (2011).


Copy RID