The Coca-Cola Company and Subsidiaries v. Commissioner

155 T.C. No. 10 UNITED STATES TAX COURT THE COCA-COLA COMPANY & SUBSIDIARIES, Petitioner v. COMMISSIONER OF INTERNAL REVENUE, Respondent Docket No. 31183-15. Filed November 18, 2020. P, a U.S. corporation, was the legal owner of the intellectual property (IP) necessary to manufacture, distribute, and sell some of the best-known beverage brands in the world. This IP included trade- marks, product names, logos, patents, secret formulas, and proprietary manufacturing processes. P licensed foreign manufacturing affiliates, called “supply points,” to use this IP to produce concentrate that they sold to unrelated bottlers, who produced finished beverages for sale to distributors and retailers throughout the world. P’s contracts with its supply points gave them limited rights to use the IP in performing their manufacturing and distribution functions but gave the supply points no ownership interest in that IP. During 2007-2009 the supply points compensated P for use of its IP under a formulary apportionment method to which P and R had agreed in 1996 when settling P’s tax liabilities for 1987-1995. Under that method the supply points were permitted to satisfy their royalty obligations by paying actual royalties or by remitting dividends. Dur- ing 2007-2009 the supply points remitted to P dividends of about $1.8 billion in satisfaction of their royalty obligations. The 1996 agree- -2- ment did not address the transfer pricing methodology to be used for years after 1995. Upon examination of P’s 2007-2009 returns R determined that P’s methodology did not reflect arm’s-length norms because it over- compensated the supply points and undercompensated P for the use of its IP. R reallocated income between P and the supply points employ- ing a comparable profits method (CPM) that used P’s unrelated bot- tlers as comparable parties. See sec. 1.482-5, Income Tax Regs. These adjustments increased P’s aggregate taxable income for 2007- 2009 by more than $9 billion. 1. Held: R did not abuse his discretion under I.R.C. sec. 482 by reallocating income to P by employing a CPM that used the supply points as the tested parties and the bottlers as the uncontrolled compa- rables. 2. Held, further, R did not err by recomputing P’s I.R.C. sec. 987 losses after the CPM changed the income allocable to P’s Mexi- can supply point, a branch of P. 3. Held, further, P made a timely election to employ dividend offset treatment with respect to dividends paid by the supply points during 2007-2009 in satisfaction of their royalty obligations. R’s reallocations to P must accordingly be reduced by the amounts of those dividends. John B. Magee, Kevin L. Kenworthy, Sanford W. Stark, Saul Mezei, Steven R. Dixon, Carl Terrell Ussing, Lisandra Ortiz, Lamia R. Matta, Michael D. Kummer, Hans D. Gerling-Ritters, and John F. Craig III, for petitioner. -3- Jill A. Frisch, Anne O’Brien Hintermeister, Julie Ann P. Gasper, Heather L. Lampert, Curt M. Rubin, Lisa M. Goldberg, and Huong T. Bailie, for respondent. CONTENTS FINDINGS OF FACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 I. International Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 A. Supply Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 B. Service Companies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15 C. Bottlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 II. The Coca-Cola System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..18 A. Integrated Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 B. Functions Performed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1. Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 a. R&D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 b. Quality Assurance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 c. Concentrate Production . . . . . . . . . . . . . . . . . . . . . . . . . 24 d. Beverage Production and Bottling. . . . . . . . . . . . . . . . . 25 e. Supply Chain Management . . . . . . . . . . . . . . . . . . . . . . 26 2. Marketing/Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 a. Consumer Marketing. . . . . . . . . . . . . . . . . . . . . . . . . . . 31 b. Trade Marketing and Distribution . . . . . . . . . . . . . . . . 37 III. Contractual Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 A. Supply Point Agreements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 1. Rights and Obligations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 a. Production and Sale of Concentrate . . . . . . . . . . . . . . . 43 b. Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2. Term Length and Exclusivity . . . . . . . . . . . . . . . . . . . . . . . . . 46 3. Remuneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 B. Service Company Agreements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 1. Standard Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 -4- 2. Other Provisions . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3. Invoicing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 C. Bottler Agreements . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 1. Rights and Obligations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 a. Production and Sale of Finished Beverages . . . . . . . . . 57 b. Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2. Term Length and Exclusivity . . . . . . . . . . . . . . . . . . . . . . . . . 59 3. Remuneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 IV. Assets and Income . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 A. Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 1. HQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 2. Supply Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 B. Income and Expenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 1. HQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 2. Supply Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 C. Brazilian Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .76 V. Tax Reporting and IRS Examination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 OPINION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 I. Burden of Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 II. Standard of Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . 86 III. Threshold Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 A. The 1996 Closing Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 B. Relevant Parties and Transactions . . . . . . . . . . . . . . . . . . . . . . . . . . 98 C. The “Best Method Rule” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .102 IV. Respondent’s Bottler CPM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109 A. Reasonableness of CPM Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . 115 B. Selection of Bottlers as Comparable Parties. . . . . . . . . . . . . . . . . . .120 C. Data, Assumptions, and Comparability Adjustments . . . . . . . . . . . 133 1. Selection of Bottlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 2. Computational Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . 137 -5- a. Operating Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 b. Operating Profit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 3. Implementation of CPM/ROA . . . . . . . . . . . . . . . . . . . . . . . .143 V. “Split Invoicing”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 VI. Petitioner’s Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .150 A. Supposed “Marketing Intangibles” . . . . . . . . . . . . . . . . . . . . . . . . . 150 1. Legal Ownership. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154 2. Economic Substance . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . 159 a. Setting Aside Contract Terms .. . . . . . . . . . . . . . . . . . 160 b. Consistency With Economic Substance . . . . . . . . . . . 167 B. Supposed “Long-Term Licenses”. . . . . . . . . . . . . . . . . . . . . . . . . . . 172 C. Royalties Payable by Brazilian Supply Point . . . . . . . . . . . . . . . . . 175 1. Ownership of Brazilian Trademarks . . . . . . . . . . . . . . . . . . . 175 2. Brazilian “Blocked Income”. . . . . . . . . . . . . . . . . . . . . . . . . . 1.84 D. Bottlers’ Ownership of Intangibles . . . . . . . . . . . . . . . . . . . . . . . . . 186 E. Proposed Alternative Transfer Pricing Methodologies . . . . . . . . . . 191 1. Proposed CUT Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 2. Proposed “Residual Profit Split Method” . . . . . . . . . . . . . . 197 3. Proposed “Unspecified Method” . . . . . . . . . . . . . . . . . . . . . . 2.06 VII. Collateral Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 A. Recomputation of Section 987 Loss . . . . . . . . . . . . . . . . . . . . . . . . 209 B. Dividend Offset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 APPENDIX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 LAUBER, Judge: The Coca-Cola Co. (TCCC) is the ultimate parent of a group of entities (Company) that do business in more than 200 countries through- out the world. TCCC and its domestic subsidiaries (petitioner) joined in filing consolidated Federal income tax returns for 2007, 2008, and 2009. Upon exami- -6- nation of those returns, the Internal Revenue Service (IRS or respondent) made adjustments that increased petitioner’s aggregate taxable income by more than $9 billion, resulting in tax deficiencies as follows: Year Deficiency 2007 $1,114,116,873 2008 1,069,425,951 2009 1,121,220,625 By amendment to answer, respondent determined additional deficiencies attribut- able to the use of “split invoicing” by certain of petitioner’s foreign affiliates. See infra pp. 64-66. The additional deficiencies are as follows: Increase in Year deficiency 2007 $28,124,719 2008 43,314,595 2009 63,465,860 These deficiencies result from transfer pricing adjustments under section 482 by which the IRS reallocated substantial amounts of income to petitioner, chiefly from its foreign manufacturing affiliates.1 These affiliates had plants in 1 Unless otherwise indicated, all statutory references are to the Internal Rev- enue Code (Code) in effect at the relevant times, and all Rule references are to the Tax Court Rules of Practice and Procedure. We round most monetary amounts to the nearest dollar. Dollar amounts appearing in tables occasionally do not sum exactly because of rounding. -7- Brazil, Chile, Costa Rica, Egypt, Ireland, Mexico, and Swaziland.2 The plants produced “concentrate”--syrups, flavorings, powder, and other ingredients--used in the production of petitioner’s branded soft drinks (including Coca-Cola, Fanta, and Sprite) and other nonalcoholic, ready-to-drink beverages. These affiliates sold and distributed concentrate to hundreds of Coca-Cola bottlers in Europe, Africa, Asia, Latin America, and Australasia. The bottlers, most of which were independent of petitioner, ranged from small family-owned businesses to large multinational companies. The bottlers used this concentrate to produce finished beverages that they marketed (directly or through distributors) to millions of retail establishments throughout the world (excluding the United States and Canada). Because the foreign manufacturing affiliates supplied concentrate to bottlers, these affiliates are often called “supply points,” and we will generally refer to them as such. To enable the supply points to manufacture and sell concentrate, petitioner licensed them to use petitioner’s intangible property, including trademarks, brand names, logos, patents, secret formulas, and proprietary manufacturing processes. This intangible property is extremely valuable: Coca-Cola is the best known 2 Swaziland has since changed its name to the Kingdom of Eswatini. We refer to it as Swaziland in this Opinion to match the parties’ terminology. -8- brand in the world, recognized by more of the planet’s 7.7 billion inhabitants than any other English word but “OK.” The gist of respondent’s position is that the supply points paid insufficient compensation to petitioner for the rights to use petitioner’s intangible property. The Irish and Brazilian supply points account for roughly 85% of the disputed income adjustments.3 For 2007-2009 petitioner reported income from its foreign supply points us- ing the “10-50-50 method,” as it had done for the previous 11 years. This was a formulary apportionment method to which petitioner and the IRS had agreed in a closing agreement executed in 1996, which resolved petitioner’s tax liabilities for 1987-1995. This method permitted the supply points to retain profit equal to 10% 3 All of the supply points except the Mexican supply point were controlled foreign corporations (CFCs). See sec. 957(a). The Mexican supply point operated as a branch, and its income was reported on petitioner’s U.S. consolidated return. As applied to the Mexican supply point, therefore, the transfer pricing adjustment did not increase petitioner’s gross income. Rather, the IRS sought to reduce peti- tioner’s foreign tax credits on the theory that the Mexican branch had reported in- sufficient royalty expenses for use of petitioner’s intangible property, thus artifici- ally inflating the branch’s income and the Mexican corporate tax paid thereon. Respondent contended that the Mexican taxes were to that extent noncompulsory payments ineligible for the foreign tax credit. See sec. 901; sec. 1.901-2(a)(2)(i), Income Tax Regs. We resolved that issue in petitioner’s favor on summary judg- ment. See Coca-Cola Co. & Subs. v. Commissioner, 149 T.C. 446 (2017). The tax liabilities attributable to the Mexican supply point for 2007-2009 have thus been resolved, with the exception of a foreign currency adjustment under section 987. See infra pp. 209-218. But the operations of the Mexican supply point are relevant to the overall transfer pricing analysis and were the subject of extensive testimony at trial. -9- of their gross sales, with the remaining profit being split 50%-50% with petitioner. The closing agreement did not address what transfer pricing methodology would be used for years after 1995. But petitioner continued to employ the 10-50-50 method, from 1996 onwards, to report income from its foreign supply points un- less an advance pricing agreement or competent authority proceeding dictated otherwise. Because the closing agreement specified the compensation due petitioner for use of its intangible property, the amounts due petitioner under the 10-50-50 method were in the nature of royalties. However, the closing agreement permitted the foreign supply points to satisfy their royalty obligations by paying actual royal- ties or by repatriating funds to petitioner in other ways, e.g., by paying dividends. During 2007-2009 more than $1.8 billion of the income petitioner received from its foreign supply points pursuant to the 10-50-50 method took the form of divi- dends rather than royalties. Petitioner claimed “deemed paid” foreign tax credits (FTCs) under section 902 with respect to these dividends, as the closing agreement had permitted for 1987-1995. Upon examination of petitioner’s 2007-2009 returns the IRS concluded that the 10-50-50 method did not reflect arm’s-length pricing because it overcompen- sated the supply points and undercompensated petitioner for the use of its intan- - 10 - gible property. Invoking section 482, the IRS reallocated income to petitioner us- ing a comparable profits method (CPM), treating independent Coca-Cola bottlers as comparable parties. The IRS regarded these bottlers as comparable to the sup- ply points because they operated in the same industry, faced similar economic risks, had similar contractual relationships with petitioner, employed many of the same intangible assets (petitioner’s brand names, trademarks, and logos), and ulti- mately shared the same income stream from sales of petitioner’s beverages. To implement its bottler CPM, the IRS determined the average return on operating assets (ROA) for a group of independent Coca-Cola bottlers that it deemed comparable. It applied that average ROA to the operating assets of each supply point, generating a deemed arm’s-length operating profit. The IRS then reallocated to petitioner all income received by each supply point in excess of that benchmark. This methodology produced very substantial reallocations from the Irish and Brazilian supply points and somewhat smaller reallocations from the Costa Rican, Chilean, and Swazi supply points. The IRS methodology generated a reverse allocation of income from petitioner to the Egyptian supply point, which for historical reasons had endured many years of economic underperformance. Petitioner challenges respondent’s section 482 reallocations as arbitrary and capricious. It contends that the IRS acted arbitrarily by abandoning the 10-50-50 - 11 - method, having acquiesced in the use of that method during five prior audit cycles spanning a decade. In any event, petitioner argues that the IRS erred in employing the bottler CPM to reallocate income. Petitioner contends that independent Coca-Cola bottlers are not comparable to the supply points because the latter own immensely valuable intangible assets that do not appear on their balance sheets or in any written contract. These assets, which petitioner calls “marketing intangibles” or “IP associated with trademarks,” allegedly were created when the supply points financed consumer advertising in foreign markets. Petitioner urges that the bottlers by comparison are “marketing- light” businesses that operate at a different level of the market. Petitioner urges that the supply points owned (in substance if not in form) local rights to petitioner’s valuable brands and should thus enjoy supranormal returns as “master franchisees” or long-term licensees. To implement that theory petitioner offers, as alternatives to respondent’s bottler CPM, a comparable uncontrolled transaction (CUT) model and a residual profit split method (RPSM) as the best methods for determining the supply points’ true economic income. Alternatively, if a bottler ROA is applied to the supply points, petitioner contends that each supply point’s asset base should be increased to reflect the value of its supposed “marketing intangibles.” - 12 - If we sustain respondent’s position in whole or part, petitioner urges that the transfer pricing adjustments should be reduced to reflect dividends paid by the supply points, to the extent those amounts were repatriated to satisfy the supply points’ royalty obligations. Although petitioner elected “dividend offset” treat- ment on timely filed returns for 2007-2009, it did not include in those returns ex- planatory statements as directed by Rev. Proc. 99-32, 1999-2 C.B. 296. Respon- dent contends that petitioner’s failure to include these statements is fatal to its claim to dividend offsets. Petitioner urges that it substantially complied with the revenue procedure’s requirements and that substantial compliance was sufficient.4 FINDINGS OF FACT I. International Structure In 1886 TCCC produced the first Coca-Cola beverage, which it sold initial- ly at soda fountains. In 1899 it transferred to third parties, for $1, the exclusive rights to bottle and distribute finished Coca-Cola beverages throughout the United States. This created the “Coca-Cola System,” comprising the Company and its 4 Petitioner concedes that allowing dividend offsets would cause the divi- dends to lose their character as such, necessitating forfeiture of the deemed-paid FTCs petitioner had claimed with respect to those dividends. Respondent has amended his answer to allege that FTCs of $40,717,804 for 2007, $65,941,179 for 2008, and $49,977,463 for 2009 should be disallowed in the event we permit peti- tioner to offset, against a reallocation of royalty income, the dividends paid by the supply points in satisfaction of their royalty obligation. - 13 - (largely independent) bottlers. At all relevant times petitioner has had its head- quarters (HQ) and principal place of business in Atlanta, Georgia. Petitioner expanded internationally in the early 1900s, arriving in Europe and Latin America during the 1920s. As a vehicle for this growth petitioner estab- lished in 1930 the Coca-Cola Export Corp. (Export), a wholly-owned domestic subsidiary of TCCC. Export expanded aggressively, creating branches in 27 for- eign countries by 1975. By 2008, 74% of the Company’s sales were made outside the United States. A. Supply Points Petitioner engaged in significant restructuring as its international market matured. During World War II it had built numerous plants in Europe and Asia to supply Coca-Cola to U.S. soldiers. After the war petitioner sold the bottling facilities to private-sector companies. As bottlers were divested to third parties, Export began contributing its concentrate plants and other branch assets to foreign subsidiaries. Export’s contributions to these subsidiaries generally consisted of tangible operating assets, associated goodwill, and similar items. The subsidiaries acquired via these transactions no meaningful intangible property in the form of trademarks, tradenames, copyrights, franchises, licenses, or bottler agreements. - 14 - Export initially established affiliates in virtually every country to manufac- ture and supply concentrate to local bottlers. Before 1988, for example, Export had a fully integrated concentrate plant in every Western European country. Over time the Company gradually consolidated its concentrate manufacturing into larger plants that supplied concentrate to bottlers in diverse national markets. The Irish supply point, which reported average annual gross revenues of $6.89 billion dur- ing 2007-2009, ultimately sold concentrate to bottlers in more than 90 countries, some as distant as New Zealand and Papua New Guinea. Export owned (directly or indirectly) the seven supply points involved here. The Mexican supply point was a branch of Export and its income was reported on petitioner’s U.S. consolidated return. The Brazilian supply point5 and the Chilean supply point6 were CFCs wholly-owned by Export. The Costa Rican, Egyptian, 5 The Brazilian supply point, Coca-Cola Indústrias Ltda. (CCIL), was the parent of Recofarma Indústria do Amazonas Ltda. (Recofarma), which operated the Brazilian manufacturing facilities. In August 2009 Recofarma acquired Coca- Cola Concentrados e Refrigerantes Ltda. (CCRL), which it thereafter operated as a flavoring plant. For U.S. tax purposes Recofarma and CCRL elected to be treated as disregarded entities of CCIL, and we will refer to CCIL and its subsidiaries col- lectively as the Brazilian supply point. 6 The Chilean supply point, Coca-Cola de Chile, S.A., formed Nuevas Beb- idas de Colombia Ltda. as a wholly owned subsidiary in March 2009, and the lat- ter elected for U.S. tax purposes to be treated as a disregarded entity. We will re- fer to these entities collectively as the Chilean supply point. - 15 - Irish, and Swazi supply points were branches or disregarded subsidiaries of Atlan- tic Industries (Atlantic), a Cayman Islands CFC wholly-owned by Export. B. Service Companies As concentrate manufacturing became consolidated into fewer and fewer supply-point affiliates, the Company’s other foreign activities were typically taken over by local service companies (ServCos). During 2007-2009 the Company ap- pears to have had at least 60 foreign ServCos, each serving one or more national markets. The ServCos were responsible for local advertising and in-country con- sumer marketing, which they carried out with assistance from third-party media companies and creative design firms. The ServCos were also responsible for liai- son with local bottlers, a function petitioner called “franchise leadership.” A few ServCos had research and development (R&D) centers, which served multiple national markets. The supply points had little or no direct ownership interest in the ServCos that served these national markets. Most of the ServCos were owned by Export, generally through a chain of subsidiary CFCs. Atlantic owned two ServCos (both Irish entities) and 48% of the Mexican ServCo. TCCC itself owned (directly or indirectly) CFCs that operated ServCos in Panama, Costa Rica, and Peru. - 16 - C. Bottlers The vast bulk of the Company’s beverages were (and are) produced and distributed by independent Coca-Cola bottlers. At the outset many bottlers were small, often family-owned, enterprises that distributed to retailers within a narrow geographic market. But bottlers were likewise transformed by consolidation, and many became large multinational companies. During 2007-2009 the Company had about 300 independent bottlers that served (directly or indirectly) about 20 million retailers. The three largest inde- pendent bottlers were Coca-Cola Enterprises (CCE), Coca-Cola FEMSA, and Coca-Cola Hellenic (Hellenic). CCE, which operated in Western Europe and North America, sold about 42 billion units of Coca-Cola beverages annually. Coca-Cola FEMSA served more than 1.5 million retailers throughout Latin America.7 Hellenic served 28 national markets in Western and Central Europe, the Balkans, Russia, and Ukraine. The bottlers produced numerous nonalcoholic ready-to-drink (NARTD) beverages, generally (but not exclusively) under petitioner’s brands. These in- cluded the Company’s iconic carbonated soft drinks (CSDs): original Coca-Cola 7 TCCC held minority equity interests in Coca-Cola FEMSA and certain oth- er bottlers. In no case did these stock holdings permit petitioner to control those bottlers’ activities or dictate their decisions. - 17 - (Coke Red), Fanta, Sprite, and variations and extensions of these brands (such as Diet Coke and Coke Zero). In more recent years, as the Company expanded its beverage portfolio, the bottlers produced an increasing array of noncarbonated drinks (non-CSDs), including juices, teas, bottled waters, energy drinks, and coffee-flavored beverages. The bottlers produced most of these beverages using concentrate manufac- tured by the supply points. As appropriate to the particular drink, the bottlers mix- ed the concentrate with purified water, carbon dioxide, sweeteners, and/or flavor- ings; injected the finished beverages into bottles and cans of various serving sizes; packaged and warehoused these items pending distribution; and delivered the bev- erages to retail establishments that included supermarkets, small retail stores, bars, and restaurants. In certain European markets bottlers relied on intermediate dis- tributors to deliver the beverages to those retail customers. Although independent bottlers were crucial for the Coca-Cola System, peti- tioner occasionally acquired bottlers and brought them temporarily “in house.” This occurred (for example) when a bottler encountered financial difficulty or had to be divested in a merger. In 2006 TCCC grouped these controlled bottlers into a single management unit--the Bottling Investments Group (BIG), colloquially known as the “bottler hospital”--and supervised their activities directly from - 18 - Atlanta. Generally, petitioner’s objective was to divest ownership of these con- trolled bottlers as soon as they had recovered their footing operationally and finan- cially. At any point in time, however, controlled bottlers could account for 10% or more of the Company’s unit volume in foreign markets. II. The Coca-Cola System The Company and its authorized bottlers coordinated their functions in or- der to manufacture, market, and distribute--every day of the year--about 1.6 billion servings of NARTD beverages. This daily coordination created a shared identity and synergistic relationship between the Company and its bottlers. Each regarded itself as an integrated component of the Coca-Cola System. A. Integrated Management The Company used a flexible management structure that permitted local ad- aptation and encouraged close coordination with bottlers. By 2007 the Company had adopted a governance model called “Freedom within a Framework.” Through its HQ function in Atlanta, TCCC set detailed guidelines for brand identity, visual identity of products, quality assurance, business goals, and marketing strategies. But it permitted local units to adapt these rules (within limits) to the cultural, religious, linguistic, and culinary traditions of their particular foreign markets. - 19 - During 2007 TCCC delegated authority to regional operating groups (OGs) for the following territories: North America, Latin America, the European Union (EU), Eurasia, Africa, and the Pacific. (Eurasia and Africa were merged in 2008.) Each geographical OG supervised multiple business units (BUs), formerly called divisions, which typically had responsibility for one or more national markets, de- pending on their size. The OGs and BUs were not legal entities. Rather, they identified lines of managerial reporting from smaller to larger geographical terri- tories and ultimately to HQ in Atlanta. Almost all Company personnel involved in the manufacture of concentrate worked for the supply points.8 The Irish, Mexican, Costa Rican, and Swazi supply points had virtually no workers other than those engaged in producing concentrate and their support staff. Most other personnel, including those holding leadership positions in the OGs and BUs, were employed by the ServCos. During the years at issue, the ServCos employed all of the OG leadership and about 90% of the 200 officers who made up the BU leadership. The ServCo leadership teams acted as the liaison between the Company and local bottlers. These teams acted in a day-to-day advisory role to bottlers, facilitat- 8 Personnel who worked for the Mexican supply point were nominally on the payroll of the Mexican ServCo. This was apparently done to solve a Mexican labor-law problem. - 20 - ing bottlers’ access to the Company’s statistical data, consumer insights, advertis- ing plans, and marketing strategies. They shared with bottlers the responsibility for creating coordinated annual business plans that fulfilled TCCC’s global strate- gy and the needs of the local market. These annual business plans reflected detailed discussions with bottlers con- cerning beverage pricing, packaging, marketing, and distribution channels. The ServCos and the bottlers relied on Company data and guidelines for the granular- level details of these plans. But the budgets and overall strategies were reviewed and approved by TCCC and the top leadership of each bottler. B. Functions Performed The Coca-Cola System required that its participants discharge two principal functions: manufacturing and marketing/distribution. The Company and the bot- tlers jointly discharged these functions, performing complementary tasks in a syn- ergistic way. 1. Manufacturing The Coca-Cola System relied on an integrated manufacturing supply chain that employed personnel from all of the entities discussed above. TCCC, assisted by the ServCos, took principal responsibility for R&D and quality assurance. Ac- tual production was split between the supply points and the bottlers: The supply - 21 - points manufactured concentrate, and the bottlers used the concentrate to produce Coca-Cola beverages. TCCC was chiefly responsible for supply chain manage- ment regarding concentrate, and the bottlers were responsible for supply chain management regarding finished products. a. R&D Much of the system’s value rested on familiar, consistently flavored drinks delivered by well-established production processes. Perhaps for that reason, the Company’s R&D budget was smaller (as a percentage of revenues) than the R&D budgets of some of its competitors. But the Company maintained an active R&D program to explore new beverages, ingredients, sweeteners, and packaging. The annual budget for this program during 2007-2009 averaged about $200 million, roughly 1% of the Company’s worldwide revenues. The Company divided its R&D projects into two major subsets: research projects and development projects. Most research projects were undertaken by TCCC’s central R&D laboratory in Atlanta. These projects consisted of new, un- proven methods that, if successful, could be implemented across many countries and product lines. Examples included research into new sugar substitutes and environmentally friendly packaging materials. - 22 - Development projects usually focused on customizing global products and concepts for local implementation, taking account of local regulations, taste pref- erences, and other variables. These projects were undertaken primarily by the Company’s six regional R&D centers. Two of these were in the United States. As far as the record reveals, the other four--located in Belgium, Brazil, China, and Japan--were operated by ServCos. TCCC and the ServCos were responsible for virtually all of the Company’s R&D. TCCC owned and staffed the three domestic R&D centers and employed roughly 60% of the Company’s researchers. ServCos employed all other R&D personnel except for 20 employees who worked for the Brazilian supply point. The other supply points had no R&D personnel on their staffs. b. Quality Assurance TCCC personnel discharged most of the Company’s quality control func- tions. The Ingredient Quality Department, part of the HQ function in Atlanta, worked with the regional R&D centers to ensure consistent production quality by codifying recipes, creating global ingredient standards, and approving third-party suppliers of raw materials. Because TCCC was ultimately responsible for all formulations of Coca-Cola products, any reformulations of these beverages (e.g., to use new sweeteners) had to be approved by HQ. TCCC published quality assur- - 23 - ance information on a central database (Optiva in 2007 and Picasso in 2008 and 2009) that supply points and bottlers could easily access. TCCC personnel, with assistance from outside professionals, performed reg- ular quality control audits of supply points, flavoring plants, and other manufactur- ing facilities, including plants owned by bottlers. TCCC audited supply point fa- cilities every two or three years. Although the bottlers relied on the Company for quality assurance with respect to concentrate, they were responsible for quality assurance with respect to their own production processes. Bottlers engaged in extensive testing of finished products in their own on-site laboratories. The supply points, using their production personnel, engaged in day-to-day quality control, e.g., by performing in-process and product release testing. They performed this testing by following the Coca-Cola Management System, which provided an outline of the Company’s quality control expectations. None of the supply points (apart from the Brazilian supply point) had any employees specifi- cally dedicated to quality assurance.9 9 During 2007-2009 the Brazilian supply point employed (on average) about 50 workers identified by petitioner as primarily engaged in quality assurance. - 24 - c. Concentrate Production The supply points manufactured concentrate. Their manufacturing activity consisted of procuring raw materials and using TCCC’s guidelines and production technologies to mix and convert raw materials into concentrate. Their procure- ment activities were limited: Many ingredients could be obtained only through Company-owned flavor plants, and other ingredient purchases were negotiated by bulk procurement specialists employed by TCCC or the ServCos. Only three sup- ply point employees (one in Chile and two in Brazil) were specifically dedicated to procurement. After completing the manufacturing process, the supply points packaged the concentrate into kits tailored to the needs and capacities of the bot- tlers to whom they distributed. The manufacturing process entailed various forms of extraction, filtration, mixing, blending, aging, and precision filing. In performing these activities the supply points employed TCCC’s secret formulas, confidential ingredients, and proprietary mixing specifications. All of these steps were governed by a detailed manufacturing protocol dictated by TCCC. Petitioner’s experts agreed that this manufacturing activity was a routine activity that could be benchmarked to the activities of contract manufacturers. Two of petitioner’s experts, Drs. Michael - 25 - Cragg and Sanjay Unni, applied an 8.5% markup on costs to determine an appro- priate return for the supply points’ concentrate manufacturing function.10 The vast majority of the people who worked at the supply points were en- gaged solely in concentrate production. In 2009 the Irish supply point had 599 employees, at least 588 of whom were engaged in concentrate production. The Costa Rican supply point had 60 employees, all of whom were engaged in concen- trate production. The Swazi supply point had 153 employees, 135 of whom were engaged in concentrate production. The Brazilian, Chilean, and Egyptian supply points performed other business activities, including marketing, sales, and finance. To the extent supply point employees engaged in such nonproduction activities, they generally performed functions similar to those performed by ServCo employ- ees and overseen by BU leadership. As explained infra p. 50, ServCos were com- pensated for their services on a cost-plus basis. d. Beverage Production and Bottling Bottlers performed all finished product manufacturing. Having procured concentrate from supply points, the bottlers prepared finished beverages by mixing the concentrate with purified water, carbon dioxide (for sparkling drinks), sugar or 10 An alphabetical listing of the parties’ expert witnesses, together with a brief résumé of each, appears in an appendix to this Opinion. - 26 - other sweeteners, and additional ingredients obtained from Company-approved suppliers. The Company imposed strict standards for water quality, and each bot- tling facility was equipped with an advanced water treatment system. As a rule, each class of beverage (CSDs, juices, and table waters) ran on a specialized, high- speed production line that typically could handle only one product in one package size at a time. Bottlers thus needed multiple production lines to cover all bever- ages in all forms of packaging. Bottlers printed and appended brand labels to the cans and bottles before distributing or warehousing the products. e. Supply Chain Management The Company and the bottlers each performed supply chain management over their respective shares of the production and distribution cycle. The Com- pany managed the supply chain from the sourcing of raw ingredients through the production of concentrate to the allocation of concentrate to bottlers. Bottlers managed the supply chain from that point forward. TCCC performed virtually all supply chain management for the Company during 2007-2009. Many years earlier, when concentrate production was widely dispersed on a country-by-country basis, the Company had delegated supply chain management to local BUs. But that form of supervision became inefficient as con- - 27 - centrate manufacturing was consolidated into fewer plants that sold to hundreds of bottlers worldwide. In a bid to rationalize this system and reduce production costs, the Company in the late 1990s centralized supply chain management into the Commercial Prod- uct Supply (CPS) group. During the tax years at issue CPS was a subdivision of BIG and (like it) was centrally managed by HQ in Atlanta. A Supply Point Com- mittee, including CPS managers and top officials from TCCC’s tax and treasury departments, made key recommendations about concentrate supply. CPS leadership regularly shifted and reorganized concentrate production to enhance efficiency, reduce costs, and ensure backup sources of concentrate in the event of a supply disruption. On the basis of recommendations from CPS, the Company constructed new supply points or expanded existing plants, often in countries with low tax rates and favorable tariff regimes. CPS then shifted con- centrate production away from established plants to these newer (and typically larger) facilities. CPS sometimes shifted production among supply points to re- flect its assessments of risks from political unrest and natural disasters (such as earthquakes and typhoons). The Company, which had 52 concentrate plants in the 1980s, has pursued a steady policy of consolidating concentrate production. Between 1986 and 2006 - 28 - the Company closed (or shifted substantial production away from) 15 concentrate plants on five continents. During 2007-2009 the Company closed three concen- trate plants (in Australia, Morocco, and Peru), leaving it with only 18 foreign sup- ply points as of 2010. These closures and production shifts caused the supply points that lost production to suffer a reduction in (or the total elimination of) their manufacturing income. In virtually none of these instances was the losing supply point compensated--by TCCC or by the supply point that took over its produc- tion--for this loss of economic value.11 CPS leadership often shifted production to supply points located in jurisdic- tions that offered tax or tariff incentives. The Irish supply point, which reported an income tax rate of 1.4% during the period at issue, built a state-of-the-art plant at Ballina in 1999. In 2001 the Company shifted to the Irish supply point, from the Mexican supply point, roughly 50% of the latter’s production of concentrate for Coke Red. The Irish supply point then exported that concentrate back to bot- tlers in the Mexican market. CPS directed numerous other shifts of production to the Irish supply point between 1984 and the tax years at issue. During 2007-2009 11 On three occasions between 1962 and 1994, when concentrate production was shifted from supply points owned by Export, Export received some stock in the supply point that took over its production. On no other occasion was the los- ing supply point compensated when its production was shifted elsewhere. - 29 - the Irish supply point had by far the largest production of any foreign concentrate plant, supplying bottlers in more than 90 national markets.12 On CPS’ recommendation the Company in 2008 began construction of a new concentrate plant in Singapore. CPS caused the Irish supply point to ship to Singapore 30 containers of second-hand equipment, including mixing tanks, drum fillers, conveyers, racking systems, pumps, piping, and valves. The new Singa- pore plant was completed in two years at a cost of about $60 million. The Company consolidated concentrate production in Singapore to gain economies of scale, leverage free trade agreements, and take advantage of tax and tariff incentives. To qualify for these benefits, the Singapore plant had to meet local authorities’ targets for production volume. TCCC satisfied these require- ments by shifting concentrate production to Singapore from other supply points. The Singapore supply point thereafter supplied concentrate to bottlers in 16 mar- kets that had previously been served by 14 supply points in Asia and elsewhere. The bottlers were responsible for supply chain management from their receipt of concentrate through distribution of finished beverages to wholesalers 12 Although production shifts involving the Irish supply point show the hand of centralized supply chain management, it is not always obvious what agenda CPS was pursuing. For example, the Irish supply point was the primary supplier of the French market during 1985-1990. In 1990 that market was given to a French supply point, only to be given back to the Irish supply point in 1999. - 30 - and retailers. TCCC identified approved suppliers for most raw materials, as for concentrate. But bottlers had responsibility for securing those materials, which included aluminum, steel, plastic, and carbon dioxide. Each bottler generally had a geographic territory within which it was the exclusive supplier of Company products. This exclusivity allowed the bottlers to cultivate an intimate understanding of the thousands of local retailers and whole- salers, anticipate their needs, and build bottling and storage capacity to match. 2. Marketing/Distribution To stimulate demand for its beverages the Coca-Cola System relied in part on consumers’ past consumption experiences. But the Company and the bottlers also conducted aggressive advertising and marketing campaigns to keep their pro- ducts fresh and at the top of consumers’ minds. During the tax years at issue the System expended billions of dollars annually for marketing, split about evenly be- tween the Company and its bottlers. TCCC and its bottlers implemented an infor- mal “true up” strategy to ensure that marketing expenses were split roughly 50-50 between them. In the NARTD business, where purchases are often impulse driven, two types of marketing are needed to stimulate new demand: consumer marketing and trade marketing. Consumer marketing, coupled with past consumption experi- - 31 - ences, creates in the minds of consumers favorable associations with the product. Trade marketing, which includes efficient distribution and product placement in stores, makes the product readily available to consumers, reinforces their favorable associations with the product, and stimulates purchase at the point of sale. a. Consumer Marketing The Company took principal responsibility for consumer marketing, that is, advertising and other messages directed toward the individuals who were the final consumers of its products. The Company aimed to create demand by maintaining and exploiting its brands. The Company’s most important brand was Coca-Cola, including Coke Red, Diet Coke, Coke Zero, and their lines and extensions (collec- tively Trademark Coke). Trademark Coke products accounted for more than 50% of the Company’s profits. The Company’s core brands consisted of Trademark Coke, Fanta, Sprite, and their lines and extensions. These core brands accounted for about 85% of total net revenue and 86% of total profits for the seven supply points at issue. Consumer marketing began with TCCC, which created a uniform system for all global branding. With a few exceptions (mainly in Canada) TCCC was the reg- istered legal owner of all worldwide trademarks related to Trademark Coke, Fanta, Sprite, and their lines and extensions. For Trademark Coke products these trade- - 32 - marks covered the “Spencerian script,” the dynamic ribbon, the red-and-white col- or palette, and the contour bottle shape. TCCC sustained and perpetuated each global brand by maintaining rigorous standards for its core visual design elements and messaging. These standards provided detailed guidance that ensured a consis- tent look and feel for all global marketing. TCCC maintained for each global brand a “brand vision and architecture” that articulated what the brand aspired to stand for in consumers’ minds. The brand vision included a visual identity system (VIS), a brand strategy, core design principles, and a detailed marketing strategy. TCCC specified requirements con- cerning the use of existing designs (e.g., the Coke logo and the Spencerian Script) as well as instructions for the creation of new materials. TCCC uploaded all per- missible designs and model photographs to an online database called the “Design Machine.” It provided instructions concerning appropriate advertising copy (e.g., how to write an ad “in the Brand Voice”) and imaging (e.g., how photographs should display condensation and ice). Major deviations from these standards required explicit review and approval by TCCC. Global marketing campaigns were designed by TCCC in Atlanta, with input from ServCo personnel in various markets. A global campaign package typically included a brand representation accompanied by suggested visual images and ad- - 33 - vertising messages. Each campaign had a “core creative idea” or “underlying con- ceptual structure” that expressed what the brand stood for in the marketplace. Some global campaigns, incorporating TV ads and memorable tag lines, were launched to “refresh” Coke Red and other global brands. These included the “Coke Side of Life” campaign, launched in 2006, and the “Open Happiness” cam- paign, launched in 2009. The “Coke Side of Life” campaign ran in 200 national markets that together represented 85% of worldwide Coke Red volume. Other global campaigns centered on the Company’s sponsorship of major sporting events, including the Olympics and the World Cup. TCCC negotiated the financial terms of these sponsorships and set parameters for recommended slo- gans, graphics, and visual images. TCCC then created a package of promotional and advertising material that could be used on a global scale in association with these events. One witness estimated that this toolkit gave local marketers “70% to 80% of the solution” but allowed them space to customize the campaign to their local audience. TCCC made these global campaign materials available to its marketing per- sonnel around the world. The local marketers, nearly all of whom were employed - 34 - by ServCos,13 made the initial decision (in conjunction with bottlers) whether to “activate” a particular campaign in their marketplace. Assuming an affirmative answer to that question, they worked to customize the global campaign to meet lo- cal conditions. A global ad would be customized (for example) by hiring local ac- tors who spoke the local language, substituting songs and music that would be popular in that country, and avoiding themes and images that might offend local cultural and religious sensitivities. Marketing personnel in the ServCos generally took responsibility for marketing material that promoted local brands (such as Kuat, a Brazilian beverage derived from an Amazon fruit). Although TCCC generated material for most global campaigns, ServCos often played a leading role in regional marketing efforts. Under the “charter model,” a BU with a special interest in a particular subject or event often devel- oped platform material, including TV ads and point-of-sale promotions, that would eventually be shared with other BUs. A campaign focused on the Christmas holi- day, for example, might be generated by the Mexican ServCo; a campaign focused on Ramadan might be generated by the Egyptian ServCo; a campaign focused on Latin American teens might be generated by the Brazilian ServCo; and a campaign 13 Four supply points employed no marketing personnel whatever. The Brazilian, Egyptian, and Chilean supply points employed an average of 44, 15, and 13 marketing-designated employees, respectively. - 35 - focused on a major soccer event might be generated by the ServCo in the host country. In such cases TCCC would appoint a charter team, handle negotiations with major stakeholders, and coordinate efforts between the charter team and other BUs desiring to use the material. Those other BUs would then adapt the charter campaign to suit their local needs. TCCC provided local marketers with various tools to help them craft local ads and improve local decision-making. The Knowledge & Insights unit (K&I) in HQ performed data analysis about consumer behavior and made these data availa- ble to bottlers and ServCos (e.g., by disseminating monthly “brand health perfor- mance” reports to local managers). Customized marketing designs and tactics were uploaded to the Design Machine. Spark City, created in 2007, was a compi- lation of various training and information portals including Marketing Xchange, CSD Portal, and “the DNA of Marketing.” These portals supplied local marketers with access to an extensive database of processes and standardized frameworks for marketing each of the Company’s global brands. TCCC provided ServCos and bottlers with market research tools to help them gauge the success of their advertising efforts. K&I created protocols and metrics for measuring changes in “brand equity,” enabling marketers to assess local consumers’ brand awareness and the effectiveness of advertising messages. - 36 - TCCC packaged these metrics into user-friendly tools such as the Marketing Variance Analysis, Beverage Brand Barometer, and Consumer Beverage Land- scape. These tools were implemented throughout the Company’s global distribu- tion network, allowing ServCos and bottlers to spot trends discernible only from a global perspective. TCCC also provided tools and frameworks for training local marketers. The Integrated Marketing Communications unit (IMC) at HQ developed the curricu- lum for training marketers around the world. IMC maintained an online learning platform--Coca-Cola University--that was used by ServCos to train marketers in the field. IMC also supervised the Company’s contracts with the Olympics, FIFA (which organizes the World Cup), and the National Basketball Association. The ServCos generally hired third-party consultants (such as Nielsen) to perform local market research and testing. They delegated to outside creative firms the production of consumer advertisements. Outsourced functions included hiring actors, selecting music, filming commercials, providing voiceovers for glo- bal marketing materials, and purchasing advertising time in local media. Outside consultants often convened focus groups to assess whether a new ad hit the de- sired spot. TCCC maintained a list of approved agencies (such as Ogilvy) with - 37 - whom it had negotiated master service agreements. Local managers generally used approved agencies but were permitted to use others if necessary. Consumer marketing budgets were set in the Company’s annual business plans. Following intense negotiations with local bottlers, each BU proposed a marketing budget on a TCCC-mandated template. That proposal was reviewed by the OG and ultimately approved by HQ in Atlanta. Local management generally pegged direct marketing expenses (DME) to grow in line with gross profit targets. b. Trade Marketing and Distribution The bottlers took principal responsibility for trade marketing, that is, com- munications and other efforts directed toward (and undertaken through) the retail establishments (supermarkets, mom-and-pop stores, bars, and restaurants) that sold the Company’s beverages to consumers. Trade marketing, often called “push marketing,” increased consumers’ awareness of the Company’s brands and stimu- lated consumer demand. It covered a wide range of activities designed to ensure that the Company’s products were always “within arm’s reach of desire.” Bottlers expended efforts to acquire and retain retail customers, sometimes by creating loyalty programs. To ensure that the Company’s products were contin- uously available to consumers, bottlers had to manage inventory and ensure timely delivery. Bottlers were responsible for securing advantageous product placement - 38 - in stores, arranging point-of-sale promotions (such as floor decals and end-of-aisle displays), and offering in-store samples of new products. Bottlers managed most trade promotions (including coupons, product discounts, and digital redemption codes), which often keyed off holidays and sporting events. Bottlers often inte- grated these retail promotions with the Company’s global sponsorship activities and consumer marketing themes. In Europe, where third-party distributors deliv- ered most beverages to retailers, bottlers sent merchandisers into stores to assure proper product placement and point-of-sale displays. Responsibility for managing relationships with retail customers was divided among TCCC, the ServCos, and the bottlers. For historical reasons, the relation- ship with McDonald’s was managed directly by the Company’s chief operating officer at HQ. TCCC’s Global Customer and Commercial Leadership group maintained relationships with the system’s top 50 other customers, including Wal- Mart, Tesco, and 7-Eleven. Management of smaller multinational accounts was generally shared between the bottlers and marketing personnel in the ServCos. The bottlers had sole responsibility for managing most customer relationships at the country level. Bottlers created marketing plans for key accounts, which aligned consumer marketing with point-of-sale marketing. Bottler field service representatives, who - 39 - lived in the residential communities where retailers were located, formed close relationships with mom-and-pop stores, enabling them to suggest marketing inno- vations that included coolers, plasma TVs, and end-of-aisle displays. None of the supply points--apart from the Brazilian and Chilean supply points--had any staff devoted to sales. Through the ServCos TCCC supplied bottlers with a variety of tools to assist them with in-store marketing. Marketing professionals at HQ designed most point-of-sale materials; by accessing the Design Machine, bottlers could secure these images and photographs, then customize them for local consumption. The “picture of success,” the apparent precursor to “Right Execution Daily” (RED), supplied an ideal image of how a particular store should look to maximize sale of the Company’s beverages. RED, which was developed by Coca-Cola FEMSA in collaboration with the Company, provided bottlers with recommended point-of- sale materials, suggested price points, inventory management tools, and metrics for measuring the quality of bottler execution against set standards. To encourage impulse purchases--which provided much higher margins than purchases for future consumption--bottlers invested in coolers that were strategically placed in retail outlets. These investments were significant: At one point, coolers represented about one-third of CCE’s annual capital expenditures. - 40 - These coolers were typically used to chill and display the Company’s beverages exclusively. About two-thirds of the System’s global sales were for immediate consumption, and coolers were essential in stimulating impulse purchases in warmer climates. The bottlers owned all Coca-Cola coolers in retail stores. Larger cooler capacity became necessary as the Company’s product line grew to include many non-CSD beverages. To incentivize investment in coolers, the Company provided financial support to bottlers through its “Jump Start” program, under which it paid a percentage of the coolers’ cost. When negotiating marketing budgets with the Company, bottlers generally viewed their costs of purchasing coolers (net of the Company’s subsidy) as marketing expenses on their side of the ledger. Bottlers also negotiated financial incentives to push sales. Price promotions for the Company’s beverages were a sensitive subject, and such decisions were generally made jointly by bottlers and ServCo marketing personnel. For large retailers with greater market power, relationship managers negotiated discounts on targeted product lines. Bottlers regularly engaged in trade promotions to encour- age retailers to give the Company’s products optimal shelf space. For restaurants and mom-and-pop retailers, bottlers promoted Coca-Cola products by supplying in-kind benefits, such as coolers and Coca-Cola-branded awnings and napkins. - 41 - Bottlers reflected their marketing expenses in different ways, depending on local accounting conventions. Such expenses might be shown as “marketing de- ductions from revenue” or as “direct marketing expenses,” or they could be in- cluded among “selling, delivery, and administrative” costs. However character- ized, they were significant. During 2008 CCE had “marketing deductions from revenue” of $2.5 billion, an amount equal to 11.5% of its net revenue. Other bott- lers showed marketing deductions as high as 18% of their net revenue. III. Contractual Relationships Understanding the rights and obligations of entities within the Coca-Cola System requires an examination of both written contracts and the parties’ course of dealing. TCCC operated synergistically with its supply point and ServCo affili- ates, and it had aligned financial interests with its independent bottlers. The par- ties often did not spell out the details of their relationships in formal contracts but left these details to be governed by mutual understanding. In some cases, System participants operated under outmoded contracts that included terms inconsistent with their actual behavior. A. Supply Point Agreements TCCC was the ultimate parent of the supply points, and the contracts it exe- cuted with them often seem terse and incomplete. (Indeed, petitioner could not lo- - 42 - cate any written agreement with the Egyptian supply point.) The agreements that existed during 2007-2009 reflected an amalgamation of several (often overlap- ping) prior contracts and amendments thereto. Over time the text of most con- tracts converged, making it possible to generalize about the parties’ rights and obligations. We discuss below the prevailing terms of these agreements, noting deviations where appropriate. 1. Rights and Obligations The agreements granted the supply points the rights to produce and sell concentrate in accordance with TCCC’s specifications. The supply points were authorized to use TCCC’s intangible property in connection with their production and selling rights. They generally lacked any contractual ownership interests in TCCC’s trademarks or other intangible property, and they owned little or no intan- gible property of their own.14 14 Atlantic, which owned the Costa Rican and Swazi supply points (as disre- garded CFCs) and the Egyptian and Irish supply points (as branches), was the reg- istered owner of some trademarks in some jurisdictions with respect to Canada Dry, Crush, and Dr. Pepper beverages. Atlantic was also the registered owner of the Schweppes and Cosmos trademarks in most jurisdictions. But none of these supply points had any ownership interest (direct or indirect) in any trademarks relating to the Company’s core brands. The Brazilian supply point at one time had rights to sublicense TCCC’s trademarks to select bottlers. See infra p. 46. - 43 - a. Production and Sale of Concentrate Supply point production rights consisted of the right to produce intermedi- ary “Products,” variously defined as “concentrate,” “syrups,” and/or “beverage base.” We use the terms “Products” and “concentrate” interchangeably. The agreements distinguish “Products” from “Beverages,” which were produced by bottlers using Products as an ingredient. At no time did any supply point produce finished beverages.15 The supply points agreed to undertake production of concentrate in accord- ance with TCCC’s standards and instructions. TCCC ensured compliance with its standards by reserving the right to inspect “the methods of preparation and pack- aging on the premises of [the supply point] at all reasonable times.” Compliance with TCCC’s standards required the supply points to obtain secret ingredients, formulas, and specifications from TCCC. Most contracts expressly granted the supply point the right to purchase secret ingredients, but no agreement specified any maximum price that TCCC could charge therefor. Most of the agreements included a covenant requiring the supply point to protect the secrecy of TCCC’s production know-how: 15 TCCC’s agreements with its Irish and Mexican supply points included a provision nominally authorizing them to manufacture finished beverages. In prac- tice neither they nor any other supply point ever did this. - 44 - [The supply point] shall not at any time reveal any information with reference to the formulae or ingredients of the Products without the prior written approval of the Company, and shall keep confidential all such formulae, specifications, standards and instructions. The contracts also authorized the supply points to sell concentrate. As a rule, however, they were permitted to sell concentrate only to bottlers that had an existing contract with TCCC.16 The contracts with the Mexican, Chilean, and Costa Rican supply points permitted them to sell concentrate only as “requested by the Company and at prices set and/or revised by the Company.” The contracts themselves did not specify any formula or guidelines for pricing concentrate; we discuss that subject in connection with TCCC’s agreements with its bottlers. See infra pp. 61-66. Each supply point agreed to “keep a full and accurate account” of “all Products sold by it” and to make that account and relevant invoices available for inspection by TCCC “at all reasonable times.” b. Trademarks Except in the case of the Brazilian affiliate, the agreements granted the sup- ply points no rights or ownership interest in TCCC’s trademarks. The agreements identified TCCC as the “owner” or “registered proprietor” of the trademarks, and 16 The Irish and Swazi supply points were also permitted to sell concentrate to “other parties authorized by the Company to use the Products and the Trade- marks in connection therewith.” - 45 - TCCC expressly “reserve[d] the right to control all things and acts related to or involving the use of [the] Trademarks.” The supply point agreed “not to do any act or thing which may impair the ownership and protection” of the trademarks owned by the Company. The supply point, in short, received only a limited right to use the trademarks in connection with its production and sales activities. Unlike the other supply points, the Brazilian supply point was originally al- lowed to contract with bottlers, and to that end it was permitted to sublicense the use of TCCC’s trademarks.17 The Brazilian supply point was authorized, with “the approval of the Company and * * * Export,” to make contracts with bottlers “in which the right to bottle the Beverage is granted, but only in conformity with the specifications, formulae, instructions and standards given from time to time by the Company.” Upon termination of the Brazilian supply point agreement, all con- tracts and sublicenses executed with bottlers involving the use of the Company’s trademarks were to “vest and inure to the benefit of the Company.” The Brazilian supply point explicitly acknowledged that a sublicense “will not in any way affect 17 TCCC and the Brazilian supply point executed a number of agreements (and amendments thereto) beginning in 1963. The terms of these agreements are mutually inconsistent in some respects. In the text we express our understanding of the salient terms prevailing during the tax years in issue. - 46 - the property rights of the Company concerning its * * * trademarks, which contin- ue to be the Company’s exclusive property.” The Brazilian supply point was the only supply point that executed agree- ments sublicensing to bottlers the use of TCCC’s trademarks. In each case, TCCC was listed in the agreement as a “Parte Interveniente” or “intervening party,” thus acknowledging its consent to the sublicense. In October 2007 TCCC executed new agreements with all bottlers that held outstanding contracts showing the Bra- zilian supply point as a counterparty. These new agreements, which show TCCC as the sole counterparty, appear to have displaced those earlier agreements and thus effectively canceled the Brazilian supply point’s sublicensing authority. 2. Term Length and Exclusivity The Brazilian supply point agreement ran indefinitely but could be terminat- ed by TCCC’s unilateral action or either party’s breach of contract. The other sup- ply point agreements had an initial 12-month term (except the Costa Rica agree- ment, which had an initial two-month term), and all of them renewed automatical- ly for one-year periods absent prior notice from TCCC or the supply point. Agree- ments with three of the supply points (Mexico, Swaziland, and Ireland) provided that, during any 12-month term, either party could terminate the agreement, for any reason, upon giving 30 or 60 days’ notice to the other party. - 47 - No supply point was granted exclusive territorial rights. Each agreement described a territory--usually the supply point’s domestic market--in which the supply point was expected to operate.18 But during the tax years at issue (and for many years previously) no supply point limited its concentrate sales to the geo- graphical territory in which its manufacturing facility was located. Supply points regularly sold concentrate to bottlers in other supply points’ domestic markets. And due to the Company’s aggressive consolidation of concentrate production, the seven supply points during 2007-2009 sold concentrate to bottlers doing business in 150 different countries and autonomous regions (such as Hong Kong). No supply point was granted any right, express or implied, to guaranteed production of Coca-Cola products. The record reflects dozens of production shifts among supply points between 1980 and 2011. In hardly any cases was the entity that lost production compensated--by TCCC or by the supply point that took over its production--for the loss of income it thus suffered. 3. Remuneration Although TCCC used the 10-50-50 method to compute royalties payable by the supply points, it never incorporated any aspect of that formula into its written 18 Only the Swazi agreement described a multinational territory, covering much of sub-Saharan Africa. In practice, bottlers in that region purchased concen- trate from the Irish and Egyptian supply points as well. - 48 - supply point agreements. Agreements with the Chilean and Costa Rican supply points included no discussion of payment whatever. The Mexican supply point agreement specified a royalty computed as a percentage of operating profit. The Irish and Swazi supply point agreements specified a royalty computed as a per- centage of concentrate sales. The Brazilian supply point had agreements that in- consistently recited a one-time royalty of $100 (this version was registered with the Brazilian trademark office) and an ongoing de facto royalty embedded in the cost of ingredients purchased from TCCC. It does not appear that TCCC or the supply points paid much if any attention to these remuneration clauses. Several supply points paid petitioner a headquarters fee, dubbed “pro-rata.” To calculate these payments petitioner quantified all HQ expenses that supported multiple foreign affiliates.19 Petitioner then allocated these expenses to participat- ing supply points under a complex formula, subject to the proviso that no supply point would be allocated pro-rata in excess of the amount that would be tax-de- ductible in its local jurisdiction. The Brazilian and Egyptian supply points did not participate in the pro-rata regime at all. The Irish supply point paid about $1 billion, and the other four sup- 19 Headquarters expenses that supported a specific foreign affiliate were generally excluded from pro-rata and charged directly to that entity. - 49 - ply points collectively paid about $500 million, of pro-rata during the tax years at issue. Petitioner credited all of these payments against the supply point’s royalty obligation under the 10-50-50 method, as had been permitted under its 1996 clos- ing agreement with the IRS. The details of the pro-rata arrangement were not spelled out--and sometimes were not even mentioned--in the supply points’ agree- ments with TCCC. B. Service Company Agreements TCCC contracted (typically through Export) with at least 60 ServCos doing business throughout the world. The ServCos performed local consumer marketing and supervised relationships with local bottlers. TCCC or Export generally exe- cuted with each ServCo a written agreement employing a standard template that was modified slightly over the years. Neither party disputes that these contracts reflected arm’s-length terms and compensation. 1. Standard Terms Virtually all of the agreements run between the ServCo and TCCC or Ex- port.20 The standard template for these agreements included a boilerplate pream- 20 The only apparent exception to this rule involved the Costa Rican supply point, which had agreements with eight ServCos through the end of 2009. One of its counterparties, the Costa Rican ServCo, subcontracted to provide services to the Ecuadorian ServCo and to receive services from the Colombian ServCo. - 50 - ble, a generic description of services provided, and a confidentiality clause. The preamble typically stated that TCCC or Export engaged the ServCo because of its “expertise and know-how on the production and marketing of the Beverages, in- cluding sales, advertising, promotion and business development.” Most agree- ments specified a one-year term, which was renewed indefinitely absent notice from either party of its intent to terminate. The ServCo typically agreed to supply services that included advice regard- ing “marketing, advertising and sales promotion.” Most agreements executed after 2006 stated explicitly that the ServCo would discharge these tasks by “working with third party marketing service providers.” ServCos agreed to make recom- mendations as to whether the Company should participate in (i.e., make a financial contribution to) bottlers’ trade marketing expenditures, and to perform research concerning “regulatory, technical and marketing conditions” that might affect beverage sales in the local jurisdiction. They also agreed to perform a variety of computer-related and other back-office functions. The standard agreement included two remuneration clauses, which together provided ServCos with cost-plus compensation. The first clause generally stated that the service recipient (typically Export) would “reimburse or cause to be reim- bursed at cost the expenses incurred by * * * [the ServCo] attributable to the ser- - 51 - vices under this Agreement.” Generally speaking, expenses were netted against any income of similar character before being reimbursed. Reimbursable expenses were determined in accordance with local accounting principles and generally excluded any income taxes incurred by the ServCo. The second remuneration clause stated that the ServCo would be paid a markup on certain expenses described in the first clause. These percentage mark- ups varied among the agreements from a low of 5% to a high of 12%, with the average markup being between 6% and 7%. These marked-up expenses, when charged to Export or other service recipient, were typically denominated “fees and commissions.” Most agreements provided that the ServCo would be paid no markup on “direct marketing expenses,” which included amounts paid to third-party market- ing professionals such as advertising agencies, media companies, and creative de- sign firms. The effect of this provision was generally to deny the ServCo any markup on third-party marketing costs, which typically constituted its largest cate- gory of expenses. For reasons not explained in the record, this provision is absent from many Latin American ServCo agreements. In 2008 the Company contracted with Ernst & Young (E&Y) to analyze the services provided by ServCos to Export. E&Y agreed to prepare a “master plat- - 52 - form document” that would provide a basis for transfer pricing reports that Serv- Cos were required to file with their local taxing jurisdictions. E&Y ultimately produced two master platform documents from which it prepared about 30 local transfer pricing reports. E&Y concluded in these documents that the cost-plus compensation out- lined in the ServCo agreements was within an arm’s-length range. In support of this conclusion E&Y noted that TCCC controlled the ServCos’ annual budgets, provided major inputs to their marketing efforts, and supplied final approval for all business plans. At trial an E&Y partner testified that all of these transfer pricing reports “were written based on the [ServCo] contract[s] and the cost-plus nature of the service provided” by the ServCos, which he described as “the exact standard required [under the] transfer pricing analysis paradigm in effect in every country at the time.” 2. Other Provisions Shortly before the tax years in issue, several new provisions were intro- duced into ServCo agreements, chiefly in Europe. Petitioner attributed these variations to local tax planning undertaken by the Company. Many agreements executed after 2003 include a new clause explaining the level of risk assumed by the ServCo and clarifying the ownership of assets gener- - 53 - ated by its marketing efforts and those of the third-party marketing professionals with whom it contracted. A typical version of the clause read as follows: ServCo acknowledges that it does not take entrepreneurial risk in developing marketing concepts because the marketing advice provided by ServCo is within the strategic guidelines established by Export for the brands. ServCo also acknowledges that any marketing concepts developed by third party vendors are the property of Export. A variation of the first sentence, appearing in the more recent agreements, states that the ServCo assumed no entrepreneurial risk “because the marketing is con- tracted for by ServCo with third party service providers and is within [TCCC’s] strategic guidelines.” Petitioner’s witnesses testified that this reservation clause was added to the agreements in order to minimize the risk that the ServCo would be treated by local tax authorities as creating, in that country, a “permanent establishment” of TCCC or a foreign supply point. Whatever its purpose, this reservation clause ultimately appeared in 29 of the 37 ServCo agreements executed after 2004. The Company added another layer of tax planning to agreements executed with ServCos in the EU. Those companies were generally subject to value added tax (VAT) in their home country and were required to include VAT on their in- voices to Export (a U.S. company). Export would generally be eligible for refund of the VAT, but such refunds could often be delayed for months or years. - 54 - To mitigate this problem Export internalized its intra-EU service transac- tions by interposing a Belgian affiliate, S.A. Coca-Cola Services N.V. (CCS), be- tween it and other ServCos in the EU. Export executed a “master service agree- ment” with CCS, and CCS executed subcontracts with the ServCos doing business in the EU. Steven Whaley, the Company’s general tax counsel during 1996-2008, testified that the interposition of CCS between the ServCos and Export allowed ServCos to “zero rate” their services, thus avoiding the need to file VAT refund claims. Export’s master service agreement with CCS generally resembled TCCC’s standard ServCo contract. However, CCS was allowed no markup on the fees it paid to the local ServCos for their services. And the master agreement included a robust reservation clause concerning ownership of intangible assets generated by the local ServCos’ marketing efforts and by the Belgian R&D unit: ServCo [CCS] * * * acknowledges that any marketing concepts dev- eloped by third party vendors or any affiliate of Coca-Cola that pro- vides services to ServCo * * * are the property of EXPORT. * * *. Any intangibles arising out of the research and development activities of ServCo are the property of EXPORT. 3. Invoicing Petitioner employed a complicated (and not entirely transparent) system to make inter-company charges on account of services rendered by the ServCos. - 55 - Most ServCo agreements stated that the ServCo “shall invoice” the service recipi- ent--typically Export--in the former’s local currency. The agreements specify no deadlines, and it is unclear whether any actual invoices were ever prepared. In practice, BU leadership and finance personnel initiated inter-company charges that placed on the books of each supply point, as they determined to be ap- propriate, an allocated portion of the amounts that the ServCos (including CCS) charged to Export. Supply points were thus charged an allocated share of the ServCos’ “fees and commissions” (marked-up costs) plus an allocated share of the ServCos’ third-party marketing expenses. Petitioner has pointed to no document in the record by which any supply point (except perhaps the Irish supply point) ex- plicitly agreed to bear financial responsibility for these charges.21 21 The record includes a January 1, 1998, agreement whereby Atlantic agreed, on behalf of the Irish supply point (its branch), “to make available funds to * * * [Export] for reimbursement of the expenses of the ServCos and for payment of the service fees charged by the ServCos.” The agreement also stated that “Atlantic shall act as paymaster for defraying expenses such as marketing, ad- vertising and promotional expenses incurred or to be incurred within the territory serviced.” There is no evidence establishing that this agreement, which had a one- year term, remained in effect during 2007-2009. As petitioner notes, the agree- ment “is less than two pages long and [is] composed largely of WHEREAS clauses.” Petitioner acknowledges that the agreement “does little to explain * * * [the parties’] relationship or Atlantic Industries’ role” and asserts that it “was not a valid contract because it lacked consideration.” - 56 - The method for allocating ServCo fees and DME to supply points is not explained in any document. Petitioner’s witnesses testified that allocations were based on “the matching principle,” i.e., on the principle that expenses should be matched to revenues. In theory, a supply point was supposed to be allocated fees and DME charged by a particular ServCo depending on how much concentrate that supply point sold to bottlers in the geographic market(s) for which that Serv- Co was responsible. Thus, if a supply point sold concentrate to bottlers in 30 geographic markets, it might be allocated fees and DME charged to Export by 30 separate ServCos. In practice, the allocations of “fees and commissions” and DME to the seven supply points, as percentages of their gross revenue, varied widely. See infra pp. 74-75. The record does not explain these discrepancies. One way or another, most ServCo charges eventually found their way onto the books of one or more supply point. But there is no evidence that the supply points received invoices for these services, reviewed the propriety of the amounts they were charged,22 or had any role in selecting or evaluating the services for 22 Petitioner cites only one instance of a supply point’s exercise of review over ServCo charges billed to it. In that case the supply point had been billed for charges from the Russian ServCo even though it sold no concentrate in Russia. As one witness noted, this “really stood out and caused them to question.” - 57 - which they were made financially responsible. In essence, the supply points were passive recipients of charges that HQ and BU leadership put on their books. C. Bottler Agreements Petitioner executed formal agreements with hundreds of Coca-Cola bottlers throughout the world. In virtually all of the agreements TCCC is shown as the legal counterparty to the bottler.23 These agreements, like the supply point agreements, were based on templates that reflected standard terms and conditions. The principal variations among the bottler agreements involved the length of the contract term, notice periods, choice of law, and the exclusivity of rights granted. Unlike the supply point agreements, TCCC’s contracts with its bottlers explicitly granted them long-term and generally exclusive rights to produce and sell TCCC’s products within their respective territories. 1. Rights and Obligations a. Production and Sale of Finished Beverages Through the bottler agreements TCCC licensed bottlers to use its trade- marks and other intangible property to produce, sell, and distribute finished bever- 23 As noted supra pp. 45-46, the Brazilian supply point was shown as the counterparty in certain agreements executed with Brazilian bottlers before October 2007, with TCCC appearing as a “Parte Intervenente.” - 58 - ages.24 Like the supply points, bottlers covenanted to adhere strictly to TCCC’s production standards and to grant TCCC access to their facilities for periodic qual- ity-assurance inspections. Like the supply points, bottlers were required to buy in- gredients from TCCC affiliates or TCCC-approved suppliers. And like the supply points, bottlers enjoyed no right to purchase these inputs at any predetermined price. Whereas the supply points were permitted to sell concentrate only to TCCC- approved bottlers, bottlers had complete freedom to sell finished beverages to any wholesaler or retailer within their respective territories. The bottler agreements granted TCCC the right to review and approve bottlers’ annual business plans, which were usually developed in coordination with the local BU. Once a business plan was approved by HQ in Atlanta, the bottler agreed to “prosecute diligently” the details of the plan and to update TCCC regularly on plan implementation (e.g., by submitting sales reports in a format specified by TCCC). Bottlers also made softer commitments, e.g., “to satisfy fully the demand for each of the Beverages within the [bottler’s] Territory” and “to spend such funds for the advertising and 24 Although some bottlers were authorized to produce “syrups,” such syrups were used by the bottler internally in the course of producing finished beverages. Bottlers invariably covenanted not to sell syrups or concentrate to third parties. - 59 - marketing of the Beverages as may be required to maintain and to increase the demand * * * in the Territory.” b. Trademarks Bottlers had limited trademark rights similar to those granted to the supply points. While bottlers could use TCCC’s trademarks in connection with the pro- duction, sale, and distribution of finished beverages, they expressly acknowledged that TCCC owned the trademarks together with any goodwill generated by the bot- tlers’ use of the trademarks. TCCC reserved the right to control most aspects of trademark use, and bottlers covenanted to seek approval from TCCC for most ad- vertising, promotions, or other marketing that employed these trademarks. In practice the local ServCo generally supplied such approval. 2. Term Length and Exclusivity The specified term of most bottler agreements was between five and ten years. The largest independent bottlers, including CCE, Coca-Cola FEMSA, Hel- lenic, and Coca-Cola Amatil (which did business in Australia), had agreements with ten-year terms. Explicit approval by TCCC was required to renew a bottler agreement at the expiration of its stated term; the agreements generally precluded automatic renewal based on tacit approval. As with supply point agreements, TCCC reserved rights that allowed it to terminate bottler agreements on no more - 60 - than a few months’ notice. Bottlers would have preferred longer term contracts granting TCCC more limited termination rights, but TCCC consistently refused to agree to such modifications. In practice, the mutual dependence between the Company and its bottlers ensured that bottler agreements were almost always renewed. When a bottler per- formed badly or encountered financial difficulties, TCCC’s solution typically was not to terminate the bottler, but to acquire it, put it into the “bottler hospital,” and supervise its operations from Atlanta until it had recovered its footing financially and operationally. See supra pp. 17-18. TCCC would then divest the bottler to new owners with its bottler contract intact. Many of the Company’s major bottlers were public companies required to disclose financial information in annual reports and public filings. CCE, one of the top three Coca-Cola bottlers, described its relationship with TCCC as follows: While the [bottler] agreements contain no automatic right of renewal * * * we believe that our interdependent relationship with TCCC and the substantial cost and disruption to that company that would be caused by nonrenewals ensure that these agreements will continue to be renewed. For this reason most major bottlers, including CCE, Coca-Cola FEMSA, and Hellenic, assigned to their bottling contracts an indefinite useful life for account- ing and financial statement purposes. - 61 - Bottler agreements also differed from supply point agreements in the ex- clusivity of the rights they granted. Supply points enjoyed no exclusivity what- ever: They were always at risk of having TCCC shift their production to another supply point, which could then sell to bottlers in their home country. By contrast, TCCC’s agreements with most bottlers included a geographically defined market in which the bottler was granted exclusive rights to produce and sell beverages. The legal landscape was different in the EU and the European Economic Area, where the Treaty of Rome guaranteed the free movement of goods among member states. For that reason, explicit exclusivity clauses are generally absent from European bottler agreements. But in practice bottlers respected each other’s notional territories and rarely attempted to sell into them. As explained by John Brock, a longtime industry veteran who formerly led CCE, there was within the EU “an implied geographic exclusivity, but it was not spelled out.” 3. Remuneration The bottlers remunerated the Company through the price they paid for con- centrate. That price in effect bundled all of the Company’s valuable inputs into a single bill, ostensibly for concentrate. By paying this bill, bottlers secured not only the physical beverage base, but the entire package of rights and privileges they needed to operate efficiently as Coca-Coca bottlers. This package included - 62 - the right to use TCCC’s trademarks, access to TCCC-approved suppliers, access to critical databases and marketing materials, and the expectation of ongoing con- sumer marketing support from TCCC and the ServCos. TCCC reserved the unilateral right to set the concentrate price, which in theory enabled it to determine the bottler’s profitability. But “in the real world,” as petitioner notes, “concentrate prices were established through local negotia- tions.” These local negotiations “aimed to equitably share System operating pro- fit,” i.e., the total pre-tax operating profit accruing to the Company and the bottler from that bottler’s sales of the Company’s beverages. Generally, the parties’ goal was to achieve something like a 50%-50% split of the System profit. In practice, the division usually ranged between 45% and 55% in favor of one party or the other. The bottler might negotiate for a share near the high end of this range if (for example) it faced economic headwinds or expected to incur large capital expenditures. By using estimates of future reve- nues and expenses contained in budgets and business plans, TCCC and the bottler could negotiate a concentrate price that was expected to deliver the intended share of System profit to each party. Adjustment to the concentrate price was a major undertaking that required ultimate approval by HQ in Atlanta. An officer of one BU described it as “the - 63 - mother of all negotiations with a bottler.” Such negotiations were typically under- taken only once every few years. Between those revisions, unexpected fluctua- tions in consumer demand, local inflation rates, or currency exchange rates could occur. If those risks materialized, use of a fixed concentrate price could throw off the intended division of System profit. TCCC and its bottlers devised two solutions to this problem. One solution was some form of variable pricing. In Latin America and Eastern Europe in par- ticular, bottler agreements increasingly adopted “incidence pricing,” whereby the concentrate price was initially determined at a fixed price and then “trued up” to reflect actual sales (incidences) when more complete financial data became avail- able. In Western Europe, where currencies and inflation rates were generally less volatile, TCCC and its bottlers employed a subtler version of variable pricing, key- ed to bottlers’ prior-year sales or projected current-year revenues. A second solution was to adjust, as compared with the original business plan, the marketing expenditures that the Company and its bottlers were going to make. For example, if the System profit split moved unexpectedly in the Com- pany’s direction, it might agree to reimburse the bottler for certain trade marketing expenses. Or the Company might agree to increase its consumer marketing ex- penses in the bottler’s territory, which would be expected to increase the bottler’s - 64 - sales and profits. In 2005, for example, TCCC appeased calls by Latin American bottlers for lower concentrate prices by (among other things) agreeing to reinvest an additional 20% of concentrate revenues in mutually agreed marketing projects. Conversely, if the System profit split moved unexpectedly in the bottler’s direc- tion, the Company might reduce its support for local trade marketing, or the bottler might increase its marketing expenditures, e.g., by accelerating placement of cool- ers in retail stores. Generally speaking, bottlers paid the full concentrate price to the supply point(s) from which they purchased concentrate. In some markets, however, the Company engaged in “split invoicing.” Under this practice, the supply point in- voiced the bottler for a portion of the concentrate price, and the local ServCo is- sued a separate invoice to the bottler for the remainder of the concentrate price. Where split invoicing occurred, the ServCo wound up receiving a portion of the revenues that the supply point would otherwise have received as payments for concentrate. “Split invoicing” was used chiefly with bottlers in countries that were sus- ceptible to high inflation or exchange-rate volatility. By having the bottler direct a portion of the concentrate price to a ServCo in the same country, the Company - 65 - was able to mitigate the effects of currency controls, delayed VAT refunds, and related fiscal problems.25 ServCos used their “split invoicing” revenues to offset expenses that other- wise would have been reimbursed (with markup where applicable) by Export un- der a ServCo agreement. The Brazilian, Chilean, and Irish supply points, which supplied concentrate to the bottlers in question, lost revenue as a result of this practice. But they also avoided having the corresponding expenses of the ten ServCos charged to their books. During the tax years at issue, the total “split in- voicing” revenues received by the ServCos and the expenses they allocated to these revenues were as follows: Affected Revenue Total revenue Total expenses Markup on total supply point recipient (2007-2009) (2007-2009) expenses (%) Brazil Venezuelan ServCo $445,752,031 $158,607,901 181.04 Colombian ServCo 227,660,409 176,822,070 28.75 Ireland Mexican ServCo 420,224,666 424,563,141 -1.02 Turkish ServCo 84,028,435 45,405,027 85.06 Moroccan ServCo 70,298,612 66,359,121 5.94 Bulgarian ServCo 7,183,562 8,138,222 -11.73 Chile Peruvian ServCo-1 69,365,968 58,558,956 18.45 Peruvian ServCo-2 15,572,164 10,730,629 45.12 Ecuadorian ServCo 49,051,552 46,745,900 4.93 Bolivian ServCo 7,053,548 5,168,272 36.48 Total 1,396,190,946 1,001,099,239 39.47 25 Ten ServCos received “split invoicing” revenues during 2007-2009: two ServCos in Peru and the ServCos in Venezuela, Bolivia, Ecuador, Colombia, Mexico, Bulgaria, Turkey, and Morocco. - 66 - The markups that ServCos received from bottlers under split invoicing were significantly higher (on average) than the markups ServCos normally enjoyed un- der their contracts with Export. The average markup under Export’s contracts was 6% to 7%. And this markup generally did not apply to amounts ServCos paid for third-party marketing services. See supra p. 51. As shown in the table above, the average markup ServCos received under split invoicing was almost 40%. Five of the ServCos had agreements with the bottlers from which they re- ceived split-invoicing payments. These agreements required the ServCo to pro- vide the bottler with services resembling those specified in contracts that ServCos typically executed with Export. The agreements executed by the Venezuelan and Ecuadorian ServCos specified no compensation formula. The agreement between the Mexican ServCo and its bottler (Coca-Cola FEMSA) called for a 5% markup on expenses other than DME. The agreement between the Turkish ServCo and its bottler called for an 8% markup on expenses other than DME, plus a “success fee” calculated on increases in year-over-year sales. IV. Assets and Income In 2000 the Company began using the Data Collection, Consolidation and Reporting System (DACCARS) for its worldwide operations. DACCARS tracked the financial performance of each subsidiary, branch, or other entity that prepared - 67 - and submitted data to HQ for consolidation purposes. Income and assets reported in DACCARS were aggregated and reported under one or more data codes and submitted as financial statements for various managerial units. In the ordinary course of its business, the Company did not prepare financial statements for the supply points, the most relevant units for purposes of transfer pricing analysis. However, the DACCARS data codes can be manipulated to generate separate balance sheets and income statements for the supply points. The parties have prepared and stipulated pro forma balance sheets and income state- ments, for 2007-2009, for each of the seven supply points involved here. At the parent level, the relevant unit is a consolidation of TCCC and Export that excludes the operations of the BUs that conducted the U.S. and Canadian bev- erage businesses. We will refer to this consolidated unit as HQ. HQ owned the trademarks and other intangible property at issue in this case, and it received the royalties paid by the supply points. In the ordinary course of its business, the Company did not prepare distinct financial statements for HQ, but the DACCARS data codes can be manipulated to generate balance sheets and income statements for it. The parties have prepared and stipulated pro forma balance sheets and income statements for HQ for 2007-2009. - 68 - The ServCos presumably prepared financial statements in the ordinary course of their business. But the parties have not introduced any ServCo financial statements into evidence or made any stipulations concerning their assets or in- come (apart from income earned by ServCos that received split invoicing reve- nues). Most ServCos were compensated on a cost-plus basis, and it is a fair in- ference that their reported assets and income were generally quite modest. A. Assets 1. HQ During 2007-2009 HQ showed average book assets of about $15 billion. The bulk of these assets ($11.7 billion on average) consisted of investments in subsidiaries and other affiliates. HQ’s balance sheets showed trademarks and oth- er intangible assets of about $500 million. This figure does not reflect the market value of the Company’s self-developed intangibles and beverage brands. During 2007-2009 HQ was the registered owner of virtually all trademarks covering the Coca-Cola, Fanta, and Sprite brands and of the most valuable trade- marks covering the Company’s other products. HQ was the registered owner of nearly all of the Company’s patents, including patents covering aesthetic designs (such as bottle shapes and caps), packaging materials, beverage ingredients, and production processes. HQ owned all intangible property resulting from the Com- - 69 - pany’s R&D concerning new products, ingredients, and packaging. And most ServCo agreements executed after 2003 explicitly provided that “any marketing concepts developed by third party vendors are the property of Export,” thus ce- menting ownership in HQ of subsequently developed marketing intangibles. 2. Supply Points The table below shows the average book assets appearing on the pro forma balance sheets of the seven supply points during 2007-2009: Average Assets Per Book (US$ millions) Costa Swazi- Brazil Chile Rica Egypt Ireland Mexico land Cash and cash equivalents 724 102 27 31 196 61 76 Trade accounts receivable 183 57 24 37 348 56 122 Inventories 38 15 7 10 129 45 20 Prepaid exp. and other current assets 57 3 3 25 42 82 5 Investment in investees 53 479 -0- -0- -0- -0- -0- Investments in consolidated affiliates 320 7 -0- -0- -0- 5 -0- Other assets 82 -1 2 28 113 63 3 Property, plant & equipment 70 55 8 17 382 35 23 Trademarks and other IP 190 37 -0- -0- -0- 60 -0- Total assets 1,715 753 70 148 1,209 407 249 As shown in the table, all of the supply points held significant amounts of cash and trade accounts receivable. Virtually all of their trade receivables were from Coca-Cola bottlers. The risk of bottler default was very low, and the supply points on average reported allowances for doubtful accounts equal to 0.25% of - 70 - these receivables. The Brazilian, Chilean, and Irish supply points reported aver- age allowances for doubtful accounts of less than 0.1%. The Irish supply point showed an unusually large investment in property, plant, and equipment (PPE), apparently attributable to the construction of the Bal- lina plant in 1999. The Brazilian and Chilean supply points showed unusually large investments in affiliates and investees, apparently attributable to acquisitions they made in 2009. See supra notes 5 and 6. Four of the supply points--in Ireland, Costa Rica, Egypt, and Swaziland--showed no trademarks or other intangible pro- perty on their balance sheets. Only the Brazilian supply point showed significant intangible property, representing about 11% of its book assets. B. Income and Expenses The Company derived its share of System profit through bottlers’ payments for concentrate. The supply points received and retained the bulk of this income, remitting to TCCC only what was needed to satisfy their royalty obligations as de- termined under the 10-50-50 method. Most administrative and marketing expen- ses were incurred by HQ or the ServCos. These expenses were placed on the books of the supply points through inter-company charges. Five of the supply points were charged pro-rata, which reimbursed HQ for headquarters expense. All of the supply points were charged DME (incurred by - 71 - the ServCos) and most were charged “fees and commissions” (marked-up ServCo expenses). These inter-company charges reimbursed Export for amounts that the ServCos had billed to it. Although the supply points’ pro forma income state- ments show DME as a direct expense, petitioner has not identified any supply point that actually incurred out-of-pocket costs for DME. As far as the record re- veals, all of the DME shown on the supply points’ pro forma income statements reflects inter-company charges for DME incurred by the ServCos. 1. HQ HQ’s income stream reflected its role as brand owner and administrator. Its gross receipts for 2007-2009 consisted primarily of pro-rata and royalties for use of its intangible property. HQ’s gross receipts for these years (in U.S. dollars rounded to the nearest million) included the following: Year IP royalties Pro-rata 2007 $1,394 $501 2008 1,536 513 2009 1,473 497 Total 4,403 1,511 These figures include royalties paid by 11 foreign supply points not at issue in this case but exclude any dividends paid by supply points in partial satisfaction of their royalty obligations under the 10-50-50 method. - 72 - HQ incurred numerous operating expenses, most of which were typical of the costs one would expect to be incurred by a headquarters unit. After deduction of these expenses and adjustments for nonoperating income and taxes, HQ re- ported net income (in U.S. dollars rounded to the nearest million) as follows: Year Net income 2007 $1,684 2008 1,425 2009 1,202 Total 4,311 2. Supply Points The supply points showed fairly steady increases in revenue before and dur- ing the tax years in issue. That revenue consisted almost entirely of payments from bottlers for concentrate. (Occasionally supply points also sold concentrate to one another.) The table below shows the revenues reported by the supply points for 2001 through 2009: Supply point revenue (US$ millions) Costa Swazi- Year Brazil Chile Rica Egypt Ireland Mexico land Total 2001 $626 $177 $8 $111 $3,184 $935 $284 $5,324 2002 447 167 93 104 3,586 930 359 5,685 2003 409 159 119 96 4,510 752 478 6,523 2004 481 170 130 100 5,075 647 638 7,242 2005 646 186 135 115 5,334 689 690 7,795 2006 849 223 157 129 5,760 772 696 8,586 2007 1,138 261 186 147 6,596 883 800 10,011 2008 1,286 313 220 216 7,276 941 773 11,025 2009 1,306 345 231 265 6,799 872 863 10,680 - 73 - Against these revenues the supply points offset their “cost of goods and ser- vices” (COGS) and certain minor items. Generally speaking, their COGS was modest compared to their revenues: The supply points had relatively few manu- facturing employees, and the materials needed to produce concentrate were inex- pensive and often procured by the Company in bulk. After offsetting COGS and other items the supply points reported gross profits (in US dollars rounded to the nearest million) and gross profit margins for 2007, 2008, and 2009 as follows: 2007 2007 2008 2008 2009 2009 Supply point G/P Margin (%) G/P Margin (%) GP Margin (%) Brazil $930 81.7 $1,044 81.2 $1,028 78.7 Chile 217 83.3 254 81.1 278 80.7 Costa Rica 149 80.0 176 79.9 172 74.6 Egypt 98 66.3 154 71.5 193 72.9 Ireland 5,282 80.1 5,829 80.1 5,430 79.9 Mexico 668 75.6 707 75.2 631 72.4 Swaziland 725 90.7 699 90.4 780 90.3 Total 8,069 8,863 8,512 From these gross profits the supply points deducted their business expenses. These consisted of inter-company charges and direct expenses. Inter-company charges, which varied greatly among the supply points, included royalties, pro- rata, “fees and commissions,” and DME. Direct expenses, which were significant only for the Brazilian supply point, included general and administrative expenses (G&A), sales/service costs, and marketing expenses other than DME. The table - 74 - below shows the average annual business expenses, by category, reported by the supply points during 2007-2009: Average annual business expenses (US$ millions) Supply Direct Fees & Inter-co point expenses DME comms Pro-rata royalties Total Brazil $121 $150 -0- -0- -0- $271 Chile 16 29 -0- $8 $2 55 Costa Rica 2 53 $39 11 -0- 104 Egypt 27 47 83 -0- -0- 157 Ireland 85 1,104 777 350 807 3,123 Mexico -0- 170 82 46 114 412 Swaziland 11 2 326 46 123 508 Total 4,630 As shown in the table above, the Brazilian, Costa Rican, Chilean, and Egyp- tian supply points recorded minimal or no royalty payments to TCCC. Petitioner represents that they fully satisfied their royalty obligations under the 10-50-50 method in other ways (i.e., by paying dividends and/or pro-rata). The Brazilian and Egyptian supply points did not participate in the pro-rata regime, see supra p. 48, so they showed no payments in this category. The charges for “fees and commissions” and DME varied widely among the supply points, with no clear relationship to their gross revenues. The Egyptian and Swazi supply points during 2007-2009 were allocated “fees and commissions” that averaged 40% of their gross revenue, whereas the Brazilian and Chilean supply - 75 - points reported zero “fees and commissions.”26 The DME charged to the supply points during 2007-2009, as a percentage of their average gross revenues (GR), likewise ranged widely, from 0.3% to 24.8%, as follows: Supply point DME as % of GR Brazil 12.1 Chile 9.5 Costa Rica 24.8 Egypt 22.4 Ireland 16.0 Mexico 18.9 Swaziland 0.3 After deducting inter-company charges and direct expenses as shown above, the supply points reported operating profit for 2007, 2008, and 2009 as follows: Operating profit (US$ millions) Supply point 2007 2008 2009 2007-2009 Brazil $668 $762 $758 $2,188 Chile 167 197 220 584 Costa Rica 60 72 51 184 Egypt (45) 1 18 (25) Ireland 2,185 2,530 2,456 7,172 Mexico 254 267 248 769 Swaziland 189 190 302 680 Total 3,478 4,019 4,054 11,551 26 The allocation of zero “fees and commissions” to the Brazilian and Chile- an supply points might be explained in part by the local ServCos’ receipt of “split invoicing” revenues from Venezuelan and Colombian bottlers. See supra pp. 65- 66. Where “split invoicing” occurred, the supply point(s) that sold to those bot- tlers lost revenue, but they avoided having an equivalent amount of ServCo expen- ses charged to their books. Petitioner has not quantified these effects. - 76 - The seven supply points involved here had a weighted average income tax rate of 6.3%. After adjustments for taxes and nonoperating income, these seven supply points reported total net income of $11.36 billion for 2007-2009. That total (which excludes the income realized by the Company’s 11 other foreign supply points) equaled 264% of the net income of $4.31 billion recorded by HQ during 2007-2009 (which included all royalties paid by all foreign affiliates). C. Brazilian Trademarks TCCC initially did business in Brazil through a branch. It conducted branch operations in Brazil beginning in 1945 or earlier. Those branch operations in- cluded the manufacture of concentrate beginning in 1949 or earlier. Coca-Cola bottlers have done business in Brazil since at least 1942. TCCC registered its first Brazilian trademark in 1912. Between 1912 and 1962, when the Brazilian supply point was incorporated, TCCC registered nine trademarks in Brazil. Five related to Coca-Cola, covering the product names Coca-Cola and Coke, the stylized label, and the Spencerian script. Two related to Fanta and two to Sprite, covering those product names and their stylized labels. In February 1963 TCCC executed an agreement authorizing the Brazilian supply point to manufacture concentrate and to use TCCC’s trademarks in doing so. This agreement, which related solely to Coca-Cola products, stated that the - 77 - trademarks continued to be TCCC’s “exclusive property” and that TCCC had “the exclusive right and jurisdiction * * * to control the use” of the trademarks. The agreement did not require the Brazilian supply point to perform marketing activi- ties or incur marketing expenditures. Between 1963 and November 17, 1985, TCCC registered an additional six trademarks in Brazil. Five related to Coca-Cola, covering the dynamic ribbon and the product names Coke Light, Coca-Cola Light, and Coke Classic. The sixth was a seemingly duplicative trademark for Sprite. Between November 17, 1985, and the tax years at issue, TCCC registered at least 53 additional trademarks in Brazil. These covered the Coca-Cola contour bottle shape, secondary design features for TCCC’s core products, advertising slo- gans, and composites of existing trademark elements. They also covered dozens of newer products including Coke Zero, Diet Fanta, Dasani, Minute Maid, Power- ade, Kuat, and numerous other local Brazilian brands. The February 1963 agreement was amended often between 1981 and 1996 to refer to products other than Coca-Cola and to authorize the Brazilian supply point to use the other trademarks described above. These amendments made clear that all trademarks were TCCC’s “exclusive property” and that the Brazilian sup- ply point was granted only a limited right to use them to manufacture and distri- - 78 - bute concentrate. None of the agreements as thus amended required the Brazilian supply point to perform any marketing activities or incur any marketing expenses. V. Tax Reporting and IRS Examination During 2007-2009 petitioner used the 10-50-50 method to determine the royalty obligations of its supply points. Under that method, supply points were permitted to satisfy their royalty obligations by a combination of actual royalties, dividends, and pro-rata payments. The Brazilian and Chilean supply points remitted during these years, in satisfaction of their royalty obligations, aggregate dividends of about $887 million and $233 million, respectively. Atlantic, which operated the Costa Rican, Egyptian, Irish, and Swazi supply points as branches (directly or indirectly), remitted aggregate dividends of about $682 million in satisfaction of those supply points’ royalty obligations. For this purpose petitioner treated Atlantic’s four supply points as a consolidated entity. Although petitioner elected “dividend offset” treatment on timely filed returns for 2007-2009, it did not include in those returns explanatory statements as directed by Rev. Proc. 99-32, 1999-2 C.B. 296. The IRS selected petitioner’s 2007-2009 returns for examination. It deter- mined that the 10-50-50 method did not reflect arm’s-length pricing because that method overcompensated the supply points and undercompensated TCCC for the - 79 - use of its intangible property. The IRS retained an economist, Dr. Scott Newlon, to analyze petitioner’s inter-company pricing and determine the best method to reallocate income. Dr. Newlon concluded that TCCC, as the legal owner of virtually all the Company’s trademarks and intangible property, owned the vast bulk of its brand value. But he found that the supply points, which functioned essentially as con- tract manufacturers, retained most of the profits generated by sales of concentrate to foreign bottlers. He concluded that a reallocation of income was necessary in order to reflect clearly the income of TCCC and its supply-point affiliates. Concluding that no uncontrolled transaction could accurately capture the value of licensing the Company’s unique brands, Dr. Newlon rejected the “com- parable uncontrolled transaction” (CUT) method as a transfer pricing methodolo- gy. He likewise rejected a “profit split” method, finding it unreliable where one party (TCCC) owned valuable intangible assets and the other parties (the supply points) owned virtually none. Instead, he elected to apply a “comparable profits method” (CPM) using independent Coca-Cola bottlers as parties comparable to the supply points. In the initial report that he prepared for the IRS, Dr. Newlon selected 18 in- dependent Coca-Cola bottlers, headquartered in 10 different countries, that had - 80 - qualified auditors’ opinions for 2007-2009.27 He concluded that a “return on oper- ating assets” (ROA) derived from these bottlers’ operations would yield appropri- ate adjustments to the supply points’ income. He believed that such adjustments would be conservative because the bottlers, which “possessed distribution net- works and customer relationships,” had more bargaining power than the supply points, which could be (and often were) terminated by petitioner at will. Dr. Newlon began his analysis by calculating the 18 bottlers’ operating in- come and operating assets, all of which he stated in their local currencies. He then divided operating income by operating assets to determine an ROA for each bot- tler. His results appear in the following table:28 A B C Bottler Operating income Operating assets ROA% home Bottler (% of net revenue) (% of net revenue) (A÷B) Chile Embotelladora Andina S.A. 18.0 41.2 43.6 Mexico Coca-Cola FEMSA, S.A.B. de C.V. 17.5 43.1 40.6 Mexico Grupo Continental, S.A.B. 18.1 50.0 36.2 Chile Coca-Cola Embonor S.A. 19.5 60.2 32.5 Mexico Embotelladoras Arca S.A.B. de C.V. 18.8 59.1 31.8 Australia Coca-Cola Amatil Limited 18.4 67.1 27.3 Spain Compania Nortena de Bebidas Gaseosas, S.A. 9.3 38.2 24.5 Chile Embotelladoras Coca-Cola Polar S.A. 13.7 57.1 24.0 USA Coca-Cola Enterprises, Inc. 8.6 47.2 18.1 27 As discussed infra p. 136, Dr. Newlon in his expert witness report expanded his analysis to include six additional independent Coca-Cola bottlers. 28 To avoid showing results in ten different currencies, the table shows each bottler’s operating income and operating assets as a percentage of its net revenue. - 81 - Turkey Coca-Cola Icecek A.S. 12.3 68.4 17.9 Greece Coca-Cola Hellenic Bottling Company S.A. 11.1 66.2 16.8 USA Coca-Cola Bottling Co. Consolidated 6.1 42.4 14.4 Nigeria Nigerian Bottling Co. PLC 6.8 60.3 11.2 Japan Mikuni Coca-Cola Bottling Co., Ltd. 3.4 45.9 7.4 Japan Coca-Cola West Holdings Company, Ltd. 2.6 54.2 4.8 Japan Shikoku Coca-Cola Bottling Co., Ltd. 2.0 55.1 3.7 Thailand Haad Thip Public Company Ltd. 1.8 60.8 2.9 Japan Hokkaido Coca Cola Bottling Co., Ltd. 0.5 46.2 1.6 Dr. Newlon observed that the five East Asian bottlers had the lowest ROAs, suggesting that they might be subject to uniquely local market conditions. He also observed that Latin American bottlers tended to have very high ROAs. To test the sensitivity of his analysis to regional differences, he segmented the bottlers as fol- lows: (1) all 18 bottlers; (2) non-East Asian bottlers; (3) Latin American bottlers; and (4) non-East Asian bottlers outside Latin America. He determined interquar- tile range ROAs for the bottlers in each segment as follows: Interquartile range ROA (2007-2009) Bottler segment 25th Percentile Median 75th Percentile All bottlers (18) 7.4% 18.0% 31.8% Non-East Asian bottlers (13) 17.9% 24.5% 32.5% Latin American bottlers (6) 31.8% 34.3% 40.6% Non-East Asian bottlers outside Latin America (7) 14.4% 17.9% 24.5% Dr. Newlon then calculated ROAs for the supply points. He determined their operating assets in essentially the same manner as for the bottlers but added an imputed asset equal to an estimated average of the supply point’s inter-com- - 82 - pany receivables. He then divided operating income by operating assets to yield ROAs as follows: Return on operating assets (ROA) 2007-2009 Supply point 2007 2008 2009 Average Ireland 189.5% 227.3% 227.9% 214.4% Brazil 175.4% 198.4% 167.5% 179.7% Chile 150.7% 159.2% 138.9% 148.6% Costa Rica 128.0% 168.0% 132.7% 143.0% Swaziland 103.7% 118.5% 161.8% 128.5% Mexico 86.8% 100.2% 96.1% 94.1% Egypt -38.8% 2.5% 17.9% -4.3% Because the first six supply points had ROAs that dwarfed those of their bottling counterparts, Dr. Newlon concluded that the supply points had received compensation in excess of an arm’s-length amount. He accordingly recommended that the IRS: (1) adjust the income of the Brazilian, Chilean, Costa Rican, and Mexican supply points downward to reflect an ROA consistent with the ROAs of the Latin American bottler segment; (2) adjust the income of the Irish and Swazi supply points downward to reflect an ROA consistent with the ROAs of the bot- tlers generally; and (3) adjust the income of the Egyptian supply point upward for 2007 and 2008. The IRS implemented adjustments consistent with Dr. Newlon’s recommen- dations. It adjusted the income of the Brazilian, Chilean, Costa Rican, and Mexi- - 83 - can supply points downward to reflect the median ROA of the Latin American bottler segment. And it adjusted the income of the Irish and Swazi supply points downward to reflect the median ROA of the 13 non-East Asian bottlers. To the extent a supply point reported income that exceeded its benchmark, the IRS deter- mined that additional royalty income should be allocated to petitioner from that supply point. The IRS calculated the additional royalties due to petitioner (in millions of U.S. dollars) as follows: From supply point 2007 2008 2009 2007-2009 Ireland $1,862 $2,223 $2,105 $6,190 Brazil 535 629 604 1,768 Swaziland 146 150 257 554 Mexico 155 180 160 496 Chile 126 152 161 439 Costa Rica 42 59 41 141 Egypt (67) (28) -0- (95) Total 2,799 3,366 3,329 9,494 The IRS issued petitioner a timely notice of deficiency reflecting these ad- justments, and petitioner timely sought review in this Court. Following discovery, respondent amended his answer to assert additional deficiencies related to peti- tioner’s practice of “split invoicing.” The ServCos that benefited from split in- voicing received compensation from the participating bottlers at rates that were higher (on average) than the rates specified in the agreements those ServCos had executed with Export. See supra p. 66. Concluding that the ServCo agreements - 84 - with Export reflected arm’s-length norms, the IRS alleged that any “excess in- come” that a ServCo received from a bottler--i.e., compensation in excess of a modest markup on non-DME expenses--should be reallocated to petitioner. In his amended answer respondent asserted increased deficiencies with re- spect to six ServCos that had received split invoicing payments, were not branches of TCCC or Export, and had received “excess income” from bottlers.29 For the Turkish ServCo, respondent defined “excess income” as income in excess of the 8% markup specified in its bottler agreement. For the other five ServCos, which lacked written agreements with their bottlers, respondent defined “excess income” as income in excess of a 5% markup on non-DME expenses. Those adjustments yielded reallocations of income from the ServCos to petitioner as follows: Reallocations to TCCC ServCo 2007 2008 2009 Total Venezuelan ServCo $50,072,015 $75,126,507 $158,699,922 $283,898,445 Colombian ServCo 7,553,181 19,590,854 19,464,437 46,608,472 Turkish ServCo 17,096,848 23,312,865 N/A 40,409,713 Peruvian ServCo-1 5,634,297 2,190,786 1,310,070 9,135,154 Peruvian ServCo-2 N/A 2,994,835 1,656,940 4,651,774 Ecuadorian ServCo N/A 540,141 199,660 739,800 Total reallocation 80,356,341 123,755,988 181,331,029 385,443,358 29 Respondent made no adjustment on account of the Bolivian or Moroccan ServCos, presumably because those entities were branches of Export, a U.S. cor- poration, so that petitioner had already received any “excess income” the bottlers had paid them. - 85 - These adjustments produced additional deficiencies totaling $134,905,174 for the three years. OPINION I. Burden of Proof The Commissioner’s determinations in a notice of deficiency are generally presumed correct, and the taxpayer has the burden of proving them erroneous. See Rule 142(a); Blohm v. Commissioner, 994 F.2d 1542, 1548-1549 (11th Cir. 1993), aff’g T.C. Memo. 1991-636. Petitioner does not urge any shift in the burden of proof under section 7491. For purposes of assigning the burden of proof in a transfer pricing case, we have treated respondent’s “determination” as the aggre- gate section 482 adjustment appearing in the notice of deficiency. See Seagate Tech., Inc. v. Commissioner, 102 T.C. 149, 170-172 (1994). The presumption of correctness does not extend to, and respondent bears the burden of proof in respect of, “any new matter, increases in deficiency, and affirm- ative defenses” pleaded in his answer. Rule 142(a)(1). In his amended answer, re- spondent asserted increased deficiencies totaling $134,905,174 for 2007-2009, all attributable to petitioner’s use of “split invoicing.” Respondent bears the burden of proof with respect to these increases in deficiency. - 86 - II. Standard of Review Section 482 has its genesis in a provision of the Revenue Act of 1926 that authorized the Commissioner to “consolidate the accounts” of related parties. Sec- tion 240(f) of that Act provided that “the Commissioner may and at the request of the taxpayer shall” consolidate the accounts of related trades or businesses “if necessary in order to make an accurate distribution or apportionment of gains, profits, income, deductions, or capital between or among such related trades or businesses.” Revenue Act of 1926, ch. 27, sec. 240(f), 44 Stat. at 46; see Ray- mond Pearson Motor Co. v. Commissioner, 246 F.2d 509, 515 (5th Cir. 1957) (Hutcheson, C.J., concurring), rev’g T.C. Memo. 1955-260. The Commissioner’s determination to “consolidate accounts” was subject to judicial review. See Now- land Realty Co. v. Commissioner, 47 F.2d 1018, 1021 (7th Cir. 1931), aff’g 18 B.T.A. 405 (1929). But in drafting section 240(f) Congress left “some discretion * * * in the [C]ommissioner, and the exercise of that discretion c[ould] only be disturbed when an abuse of it [wa]s shown.” Ibid. In 1928 Congress repealed section 240(f) and replaced it with the predeces- sor of section 482. See Revenue Act of 1928, ch. 852, sec. 45, 45 Stat. at 806. Section 45 of the 1928 Act provided: - 87 - In any case of two or more trades or businesses * * * owned or controlled directly or indirectly by the same interests, the Commis- sioner is authorized to distribute, apportion, or allocate gross income or deductions between or among such trades or businesses, if he de- termines that such distribution, apportionment, or allocation is neces- sary in order to prevent evasion of taxes or clearly to reflect the in- come of any of such trades or businesses. Whereas the “consolidated account” provision had permitted the Commis- sioner to take action “if necessary” to apportion income accurately, section 45 au- thorized him to take action “if he determines that such distribution, apportionment, or allocation is necessary.” The Senate Finance Committee explained that, while section 45 was “based upon section 240(f) of the 1926 Act,” the provision was “broadened considerably in order to afford adequate protection to the Govern- ment.” S. Rept. No. 70-960 (1928), 1939-1 C.B. (Part 2) 409, 426. We addressed the proper interpretation of section 45 of the 1928 Act in Asi- atic Petroleum Co. v. Commissioner, 31 B.T.A. 1152 (1935), aff’d, 79 F.2d 234 (2d Cir. 1935).30 We emphasized that “[t]he statute authorizes the Commissioner to make an allocation of income or deductions ‘if he determines that such * * * allocation is necessary.’” Id. at 1157 (alteration in original). This statement indi- 30 We have treated as our own the precedent established by the Board of Tax Appeals, the predecessor of this Court. See Smith v. Commissioner, 91 T.C. 1049, 1053 (1988), aff’d, 926 F.2d 1470 (6th Cir. 1991); see also Tax Reform Act of 1969, Pub. L. No. 91-172, sec. 951, 83 Stat. at 730; Revenue Act of 1942, ch. 619, sec. 504(a), 56 Stat. at 957. - 88 - cated that an allocation of this sort was “a matter of discretion with the Commis- sioner.” Ibid. “In matters intrusted to the discretion of administrative officers,” we said, “there is a heavy burden on him who claims error in its exercise.” Ibid. In Asiatic Petroleum we analogized section 45 of the 1928 Act to section 22(c) of the same law. Section 22(c) provided, as section 471(a) currently pro- vides, that inventories shall be taken “[w]henever in the opinion of the Commis- sioner the use of inventories is necessary in order clearly to determine the income of any taxpayer.” See Hamill v. Commissioner, 30 B.T.A. 955, 958 (1934) (quot- ing sec. 22(c) of the 1928 Act). We concluded that the standard of review under section 45 should resemble that under section 22(c), citing a passage from an inventory case--Fin. & Guar. Co. v. Commissioner, 50 F.2d 1061, 1062 (4th Cir. 1931), aff’g 19 B.T.A. 1313 (1930)--as enunciating the appropriate test: Where a statute commits to an executive department of the govern- ment a duty requiring the exercise of administrative discretion, the decision of the executive department, as to such questions, is final and conclusive, unless it is clearly proven arbitrary or capricious, or fraudulent, or involving a mistake of law. [Asiatic Petroleum Co., 31 B.T.A. at 1157.] We held in Asiatic Petroleum that, because the Commissioner had “exercised the discretionary power vested in him and determined that allocation [wa]s neces- - 89 - sary,” the taxpayer had “the burden of showing that such determination was purely arbitrary.” Id. at 1158. That standard of review continues to apply today. Section 482 provides, similarly to section 45 of the 1928 Act, that “the Secretary may distribute, appor- tion, or allocate gross income, deductions, credits, or allowances between or a- mong * * * [related] organizations, trades, or businesses, if he determines that such distribution, apportionment, or allocation is necessary in order to prevent evasion of taxes or clearly to reflect the income of any such * * * [entities].” In order to set aside such discretionary action by the Commissioner, “a taxpayer must establish that the Commissioner abused his discretion by making allocations that are arbitrary, capricious, and unreasonable.” Guidant LLC v. Commissioner, 146 T.C. 60, 73 (2016); accord Amazon.com, Inc. & Subs. v. Commissioner, 148 T.C. 108, 150 (2017) (“The Commissioner has broad discretion in applying section 482, and we will uphold his determination unless the taxpayer shows it to be arbi- trary, capricious, or unreasonable.”), aff’d, 934 F.3d 976 (9th Cir. 2019); Bausch & Lomb, Inc. & Consol. Subs. v. Commissioner, 92 T.C. 525, 582 (1989) (ruling that the Commissioner’s “section 482 determination must be sustained absent a showing that he has abused his discretion”), aff’d, 933 F.2d 1084 (2d Cir. 1991). - 90 - Our determination whether the Commissioner has abused his discretion gen- erally turns upon questions of fact. See Amazon.com, Inc., 148 T.C. at 150 (citing cases); Paccar, Inc. & Subs. v. Commissioner, 85 T.C. 754, 787 (1985), aff’d, 849 F.2d 393 (9th Cir. 1988). “If the record before this Court fails to support the allo- cation, then we must conclude that the Commissioner abused his discretion.” Marc’s Big Boy-Prospect, Inc. v. Commissioner, 52 T.C. 1073, 1092 (1969), aff’d sub nom. Wis. Big Boy Corp. v. Commissioner, 452 F.2d 137 (7th Cir. 1971). “But if there is substantial evidence supporting the determination, it must be aff- irmed.” Ibid.; see Advance Mach. Exch., Inc. v. Commissioner, 196 F.2d 1006, 1007-1008 (2d Cir. 1952), aff’g 8 T.C.M. (CCH) 84 (1949). In considering whether the Commissioner abused his discretion, we have of- ten said that our review “focuses on the reasonableness of the [Commissioner’s] result and not on the details of the methodology” he employed. Guidant LLC, 146 T.C. at 73.31 A taxpayer may show that the Commissioner reached an unreasona- ble result by establishing that its income as reported reflects “the results that would have been realized if uncontrolled taxpayers had engaged in the same trans- 31 Accord Altama Delta Corp. v. Commissioner, 104 T.C. 424, 457 (1995); Sundstrand Corp. v. Commissioner, 96 T.C. 226, 354 (1991); Bausch & Lomb, Inc., 92 T.C. at 582; Leedy-Glover Realty & Ins. Co. v. Commissioner, 13 T.C. 95, 107 (1949), aff’d, 184 F.2d 833 (5th Cir. 1950). - 91 - action under the same circumstances.” Sec. 1.482-1(b)(1), Income Tax Regs. But that typically requires evidence of comparable uncontrolled transactions that sup- port the taxpayer’s return position. See Lufkin Foundry & Mach. Co. v. Commis- sioner, 468 F.2d 805, 807-808 (5th Cir. 1972), rev’g on other grounds T.C. Memo. 1971-101. In cases such as this, involving unique and extremely valuable intangible property, comparable uncontrolled transactions may not exist. In order to show that the Commissioner has reached an unreasonable result in such a case, the tax- payer typically will need to establish that the Commissioner employed an unrea- sonable methodology to reach his result. A taxpayer may do this by showing that the Commissioner’s methodology implicated significant legal error.32 Alternative- ly, the taxpayer may show that the Commissioner implemented his methodology in 32 See, e.g., Commissioner v. First Sec. Bank of Utah, 405 U.S. 394, 407 (1972) (finding Commissioner’s allocation of income to bank unwarranted be- cause bank’s receipt of that income would have violated banking laws); Ama- zon.com., Inc., 148 T.C. at 157-158 (finding Commissioner’s enterprise valuation method arbitrary because it included growth options and other residual business assets that were not “intangibles” for purposes of sec. 482); Hosp. Corp. of Am. & Subs. v. Commissioner, 81 T.C. 520, 595 (1983) (finding Commissioner’s 100% reallocation unreasonable because “section 482 does not authorize an allocation that would in effect disregard the separate corporate existence” of a related foreign corporation); L.E. Shunk Latex Prods., Inc. v. Commissioner, 18 T.C. 940 (1952) (finding Commissioner’s method arbitrary because it allocated income to an entity prevented by wartime price controls from receiving such income). - 92 - an unreasonable manner, e.g., by employing erroneous assumptions, incorrect data, or an analysis that is internally inconsistent.33 If the taxpayer demonstrates that the Commissioner’s allocation is arbitrary, capricious, or unreasonable, but fails to prove an alternative allocation that meets the arm’s-length standard, the Court, using its best judgment, “must determine from the record the proper allocation of income.” Sundstrand Corp. v. Commis- sioner, 96 T.C. 226, 354 (1991); see Hosp. Corp. of Am. & Subs. v. Commission- er, 81 T.C. 520, 596-597, 601 (1983); Nat Harrison Assocs., Inc. v. Commissioner, 42 T.C. 601, 617-618 (1964) (determining a proper allocation “without the benefit of any presumptions” after finding the Commissioner’s allocation unreasonable). We may make partial allocations to the extent “the evidence shows that neither side is correct.” Eli Lilly & Co. v. Commissioner, 856 F.2d 855, 860 (7th Cir. 1988), rev’g in part on other grounds 84 T.C. 996 (1985); see Amazon.com, Inc., 33 See, e.g., Veritas Software Corp. & Subs. v. Commissioner, 133 T.C. 297, 323-327 (2009) (finding allocations based on a discounted cashflow methodology unreasonable where the Commissioner “employed the wrong useful life, the wrong discount rate, and an unrealistic growth rate”); Altama Delta Corp., 104 T.C. at 466 (finding allocations unreasonable where the Commissioner implemented his cost-plus method by marking up operating profit margins instead of gross profit margins); Seagate Tech., Inc., 102 T.C. at 192 (rejecting expert’s pricing of com- ponent parts upon finding that his methodology “d[id] not meet the description of the cost-plus method” in the regulations); Achiro v. Commissioner, 77 T.C. 881, 900 (1981) (rejecting the Commissioner’s allocation where he made no “reason- able attempt[] to reflect arm’s-length transactions among the related entities”). - 93 - 148 T.C. at 163-214 (making partial allocations with respect to buy-in payment for taxpayer’s website technology, marketing intangibles, and customer information). III. Threshold Considerations A. The 1996 Closing Agreement At the outset petitioner urges that the IRS acted arbitrarily by deviating from the 10-50-50 method, to which the parties had agreed when executing the closing agreement in 1996. Generally, the Commissioner’s discretion to reallocate income under section 482 is limited only by the arm’s-length standard. But the Commis- sioner may voluntarily limit his discretion in certain ways, e.g., by entering into an advanced pricing agreement (APA). See Eaton Corp. v. Commissioner, 140 T.C. 410, 413 (2013). He may also restrict his discretion by executing a closing agree- ment under section 7121. In 1996 the parties settled a transfer pricing dispute in- volving petitioner’s 1987-1995 tax years and embodied the terms of that settle- ment in a closing agreement. Closing agreements are contracts and are “governed by the rules applicable to contracts generally.” United States v. Lane, 303 F.2d 1, 4 (5th Cir. 1962); see Long v. Commissioner, 93 T.C. 5, 10 (1989), aff’d, 916 F.2d 721 (11th Cir. 1990). Closing agreements are construed according to the intent of the parties when exe- cuting the agreement, and their intent will be inferred from the four corners of the - 94 - document unless it is ambiguous. Ibid. “Under section 7121 a court may not in- clude as part of the agreement matters other than the matters specifically agreed upon and mentioned in the closing agreement.” Analog Devices, Inc. v. Commis- sioner, 147 T.C. 429, 445 (2016) (quoting Zaentz v. Commissioner, 90 T.C. 753, 766 (1988)). The recitals to the 1996 closing agreement stated that: (1) petitioner owned “intangible property that is used by Supply Points in connection with the manufac- ture and marketing of concentrates”; (2) a dispute arose “regarding the allocation of Product Royalty income between Supply Points and [petitioner]”; and (3) the parties “have agreed on a method for computing the arm’s length amount of the Product Royalties.” The body of the agreement details the 10-50-50 method as the agreed formula for determining “Product Royalties” for tax years 1987-1995. The short and (we think) the complete answer to petitioner’s argument is that the closing agreement says nothing whatever about the transfer pricing meth- odology 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 fu- ture 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 evi- dence in the document that the IRS shared that desire or agreed to implement it. - 95 - Petitioner urges that the closing agreement was predicated on certain “fac- tual underpinnings,” including a “recogni[tion]” by the IRS that the supply points “were responsible for generating demand and were entitled to share in the result- ing profits related to the * * * [Company’s] intangibles.” These “factual underpin- nings,” petitioner says, are binding on the Commissioner unless he can show some material change in underlying fact. This argument is unpersuasive for at least two reasons. First, we do not dis- cern in the closing agreement the “factual underpinnings” that petitioner seeks to extract from it. The parties executed the closing agreement to settle a dispute. Parties agree to settlements for all sorts of reasons--to avoid the hazards of litiga- tion, to minimize litigation costs, or to seek other fish to fry. There is nothing within the four corners of the closing agreement to suggest that the Commissioner regarded the 10-50-50 method as the Platonic ideal of arm’s-length pricing for pe- titioner and its supply points. The 10-50-50 method was simply a formula to which the parties agreed in settling the dispute before them at that moment. The only mention of “arm’s length” in the closing agreement appears in a preliminary recital, which is not binding on the parties. See Analog Devices, Inc., 147 T.C. at 446; see also Rev. Proc. 68-16, sec. 6.05(3), 1968-1 C.B. 770, 779. - 96 - Second, even if we could confidently extract any factual underpinnings from the closing agreement, there is no evidence that the parties intended them to be binding for future years. Petitioner urges that “ordinary preclusion doctrines” prevent parties “from revisiting a [settlement] agreement’s factual underpinnings in later litigation when the parties intend their agreement to have that preclusive effect.” But in so contending petitioner assumes what it needs to prove--that the parties in 1996 intended to address the appropriate transfer pricing methodology for years after 1995 and embodied that intent in the closing agreement. See Ari- zona v. California, 530 U.S. 392, 414 (2000) (noting that “settlements ordinarily occasion no issue preclusion * * * unless it is clear * * * that the parties intend their agreement to have such an effect”), supplemented by 531 U.S. 1 (2000). Petitioner notes that the closing agreement was intended to have some pro- spective effect because it granted the Company penalty protection for future years, providing: 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. - 97 - We think this provision hurts, rather than helps, petitioner. It shows that the parties knew how to make the closing agreement conclusive for future years when they wished to do so. “[W]here the specificity and apparent comprehensiveness of an agreement’s enumeration of a category of things * * * implies that things not enumerated are excluded, we will apply the canon expressio unius est exclusio alterius.” BMC Software, Inc. v. Commissioner, 780 F.3d 669, 676 (5th Cir. 2015), rev’g on other grounds 141 T.C. 224 (2013).34 Indeed, the agreement spe- cifically recognizes the possibility that the IRS might make transfer pricing adjust- ments for years after 1995, because it gives petitioner penalty protection “with re- spect to the portion of any underpayment that is attributable to an adjustment of such Product Royalty.” Petitioner seeks to frame this issue as if the IRS has pulled the rug out from under it. After executing the 1996 closing agreement petitioner took steps to mini- mize its exposure to VAT and income tax in Europe and elsewhere. And it viewed the closing agreement as “facilitat[ing] the shift of concentrate supply among for- eign * * * [supply points] to meet [its] business exigencies.” Certain of these ac- 34 Accord, e.g., Smith v. United States, 850 F.2d 242, 245 (5th Cir. 1988) (finding that closing agreement that did not address penalties was not ambiguous and did not bar IRS from later demanding penalties); Analog Devices, Inc. & Subs. v. Commissioner, 147 T.C. 409, 455 (2016), overruling BMC Software, Inc. v. Commissioner, 141 T.C. 224 (2013). - 98 - tions are unhelpful to its central submission in this case--that the supply points owned immensely valuable off-book intangible assets that justified the extraordi- narily high profits they enjoyed. See infra pp. 157-158, 168, 186-187. In essence, petitioner urges that it relied to its detriment on a belief that the IRS would adhere to the 10-50-50 method indefinitely. But petitioner cannot es- top the Government on the basis of a promise that the Government did not make. See Union Equity Coop. Exch. v. Commissioner, 58 T.C. 397, 408 (1972) (“The mere fact that * * * [the taxpayer] may have obtained a windfall in prior years does not entitle it to like treatment for the taxable year[s] here in issue[.]”), aff’d, 481 F.2d 812 (10th Cir. 1973); see also ATL & Sons Holdings, Inc. v. Commis- sioner, 152 T.C. 138, 147 (2019). B. Relevant Parties and Transactions The section 482 regulations require that we determine the “true taxable in- come of a controlled taxpayer.” Sec. 1.482-1(b)(1), Income Tax Regs. “Taxpayer means any person, organization, trade or business, whether or not subject to any internal revenue tax.” Id. para. (i)(3). A “controlled taxpayer” is defined as “any one of two or more taxpayers owned or controlled directly or indirectly by the same interests, and includes the taxpayer that owns or controls the other taxpay- ers.” Id. subpara. (5). - 99 - In determining the true taxable income of a controlled taxpayer, we examine the “controlled transaction[s]” to which that taxpayer was a party. Id. para. (b)(1). A “controlled transaction” includes any transfer, between members of a controlled group, of any interest in or right to use any intangible property, “however such transaction is effected, and whether or not the terms of such transaction are form- ally documented.” Id. para. (i)(7) and (8). The Commissioner properly treated the supply points and the ServCos as distinct sets of “controlled taxpayers” that engaged in discrete sets of “controlled transactions” with TCCC, itself a “controlled taxpayer.” TCCC’s affiliates per- formed distinct economic functions: The supply points manufactured concentrate, and the ServCos arranged local consumer marketing and liaison with bottlers. These affiliates were separate legal entities (or branches of CFCs that were sepa- rate legal entities) and were treated by petitioner as such. Each supply point and ServCo executed, with TCCC, Export, or CCS as the counterparty, a separate agreement specifying each party’s rights and obligations. Determining that the supply points had paid insufficient compensation to petitioner for the rights to use petitioner’s intangible property, the Commissioner reallocated income to petitioner from the supply points (or from the CFCs of which they were branches). The Commissioner determined that the ServCos’ tran- - 100 - sactions with petitioner, generally priced on a cost-plus basis, were conducted at arm’s length. Except where “split invoicing” occurred, therefore, he made no transfer pricing adjustments with respect to the ServCos. Petitioner does not dispute (and could not plausibly dispute) that the supply points were “controlled taxpayer[s]” within the meaning of section 1.482-1(b)(1), Income Tax Regs. But petitioner urges that we focus more broadly on the activi- ties of its foreign “business units” (BUs). Petitioner’s economic experts common- ly refer, even more vaguely, to “the Field,” by which they mean an amalgamation of TCCC’s foreign affiliates in toto. They seek to frame the task before us as di- viding income between HQ and “the Field” on the basis of the “historical market- ing spend” by “the Field.” We reject these overtures because they ignore the separate taxable and legal entities involved. Each BU had responsibility for the Company’s economic per- formance within a geographic market, which could consist of one or more coun- tries. The BUs received reports and data from “management accounting units” (MAUs), and the BUs in turn reported to regional operating groups (OGs). The MAUs, BUs, and OGs were not legal entities; rather, they identified lines of mana- gerial reporting from smaller to larger geographical territories and ultimately to - 101 - HQ in Atlanta. During the years at issue, the ServCos employed all of the OG leadership and about 90% of the 200 officers who made up BU leadership. To the extent petitioner suggests that we should treat the BUs as the rele- vant “controlled taxpayers,” we reject that suggestion. The BUs were not legal en- tities and were not taxpayers. Essentially they were boxes on an organizational chart--groups of Company officials who formulated business plans and prepared financial reports for their territories, then forwarded those documents up the chain to Atlanta for review and approval. Unlike the supply points and the ServCos, the BUs engaged in no economic transactions that could be tested for compliance with arm’s-length norms. Petitioner’s suggestion, moreover, entails both duplication and inconsisten- cy. Petitioner would treat the ServCos as controlled taxpayers transacting with TCCC at arm’s length, while using the BUs as proxies or substitutes for the supply points. But the BUs consisted almost entirely of personnel employed by the Serv- Cos. Petitioner agrees that the ServCos were properly compensated on a cost-plus basis for their employees’ services; petitioner cannot simultaneously contend that the services of these employees justified supranormal returns for the supply points. We cannot endorse an approach that would count the contributions of the Serv- - 102 - Cos’ employees twice, as well as treat their contributions as having vastly differ- ent values depending on the entity with which they were deemed associated.35 In determining whether the Commissioner abused his discretion in reallocat- ing income to petitioner from the supply points, we consider the fact that they re- ported on their books most of the marketing and related costs that the ServCos in- curred and invoiced to TCCC or Export. See infra pp. 167-172. But we will not conflate the ServCos with the supply points, attribute the activities of the ServCos’ employees to the supply points, or otherwise combine them for purposes of our transfer pricing analysis. C. The “Best Method Rule” Section 482 authorizes the Secretary to “make allocations between or a- mong the members of a controlled group if a controlled taxpayer has not reported its true taxable income.” Sec. 1.482-1(a)(2), Income Tax Regs. A controlled tax- 35 In any event, petitioner has offered no alternative transfer pricing analysis using the BUs as the “controlled taxpayers.” See sec. 1.482-1(b)(1), Income Tax Regs. The BUs were managerial lines of reporting that crossed the lines of the un- derlying legal entities. Multiple supply points often sold concentrate into the geo- graphical territory for which a particular BU had reporting responsibility. One or more ServCos might provide services within that area. Petitioner reconfigured its OGs in 2008, combining Eurasia and Africa, and the scope of each BU’s reporting responsibility could likewise change at any time as petitioner saw fit. None of petitioner’s experts attempted to construct a transfer pricing analysis using these vague and uncertain parameters. - 103 - payer’s “true taxable income” is the income “that would have resulted had * * * [the controlled taxpayer] dealt with the other member or members of the group at arm’s length.” Id. para. (i)(9). In assessing the appropriateness of any allocation “the standard to be applied in every case is that of a taxpayer dealing at arm’s length with an uncontrolled taxpayer.” Id. para. (b)(1). The Department of the Treasury (Treasury) and the courts have applied this arm’s-length standard for decades. See, e.g., Essex Broadcasters, Inc. v. Commissioner, 2 T.C. 523, 529 n.2 (1943) (discussing the Revenue Act of 1938). The rules governing the choice of methodology for applying the arm’s- length standard, however, have changed significantly over time. When promulgat- ing detailed transfer pricing regulations in 1968, Treasury set forth a fixed hierar- chy of methods, with the “comparable uncontrolled price” (CUP) method being the most highly prized. See 26 C.F.R. sec. 1.482-2(e)(1)(ii), (2) (1969), T.D. 6952, 1968-1 C.B. 218, 235. The other methods specified in the 1968 regulations, in order of priority, were the resale price method and the cost-plus method. Id. subparas. (3) and (4). These regulations authorized the use of “other appropriate method[s], or variations of * * * [the specified] methods, for determining an arm’s length price.” Sundstrand Corp., 96 T.C. at 358. But use of an unspecified meth- od was allowed only if “none of the three [specified] methods of pricing * * * can - 104 - reasonably be applied under the facts and circumstances as they exist in a particu- lar case.” 26 C.F.R. sec. 1.482-2(e)(1)(iii) (1969).36 In cases governed by the 1968 regulations, courts searched assiduously for “comparable uncontrolled sales,” heeding the regulation’s injunction that, if such transactions existed, the CUP method “must be utilized because it is the method likely to result in the most accurate estimate of an arm’s length price.” Id. subdiv. (ii); see also id. para. (d)(2)(ii). In 1985 Congress expressed concern that courts had sometimes strained too far in this direction by approving use of the CUP method “even though there are significant differences in the volume and risks in- volved, or in other factors.” H.R. Rept. No. 99-426, at 424 (1985), 1986-3 C.B. (Vol. 2) 1, 424. Viewing this problem as especially “troublesome where transfers of intangibles are concerned,” Congress decided that a “statutory modification to the intercompany pricing rules regarding transfers of intangibles [wa]s necessary.” Ibid. 36 While the 1968 regulations had distinct rules for transfers of intangible property, 26 C.F.R. sec. 1.482-1(d) (1968), the “preferred method” was still the CUP, see T.D. 8470, 1993-1 C.B. 90, 91. The three-method hierarchy was under- stood to apply for purposes of transfer pricing generally. See Cym H. Lowell et al., U.S. International Transfer Pricing para. 5.05[1] (WG&L 2019) (stating that the regulations’ “priority of method approach” was “essentially the same” for transfers of tangible and intangible property). - 105 - Congress accordingly amended section 482 to require that, “in the case of any transfer (or license) of intangible property * * * , the income with respect to such transfer or license shall be commensurate with the income attributable to the intangible.” Tax Reform Act of 1986, Pub. L. No. 99-514, sec. 1231(e)(1), 100 Stat. at 2562-2563. Congress recognized that this legislation left unresolved many difficult and important issues. See H.R. Conf. Rept. No. 99-841, at II-638 (1986), 1986 U.S.C.C.A.N. 4075, 4726. It therefore directed the IRS to give “careful con- sideration * * * to whether the existing regulations could be modified in any re- spect.” Ibid. Responding to Congress’ concerns, Treasury in 1994 promulgated new reg- ulations under section 482 that supersede the 1968 regulations for transactions after their effective date. See T.D. 8552, 1994-2 C.B. 93; sec. 1.482-1(j)(4), In- come Tax Regs. These regulations eliminated the hierarchical approach of the 1968 regulations and replaced it with the “best method rule.” Sec. 1.482-1(c)(1), Income Tax Regs. The “best method rule” requires that “[t]he arm’s length result of a controlled transaction must be determined under the method that, under the facts and circumstances, provides the most reliable measure of an arm’s length result.” Ibid. “Thus, there is no strict priority of methods, and no method will in- variably be considered to be more reliable than others.” Ibid. - 106 - For controlled transfers of intangible property, the regulations require that the arm’s-length result be determined under one of four methods listed in section 1.482-4(a), Income Tax Regs. The four permissible methods are: (1) the “compa- rable uncontrolled transaction” (CUT) method, which succeeded the CUP method of the 1968 regulations; (2) the “comparable profits method” (CPM), which the Commissioner employed in this case; (3) the “profit split method”; and (4) an “un- specified method,” subject to constraints set forth in the regulations. Id. Detailed rules for applying the CPM are set forth in section 1.482-5, Income Tax Regs. Petitioner launches a threshold sally against respondent’s methodology by urging that the CPM is inferior, in some generic sense, to other methods for pric- ing transfers of intangible property. For that proposition petitioner relies chiefly on a statement in the preamble to the 1994 final regulations, where Treasury re- ferred to the CPM as “a method of last resort.” T.D. 8552, 1994-2 C.B. at 109. Petitioner’s argument pays insufficient heed to the context in which that statement was made. The preamble notes that, “[g]iven adequate data, methods that determine an arm’s length price (e.g., the CUP method) * * * generally achieve a higher degree of comparability than the CPM.” Ibid. For that reason, results based on compar- able uncontrolled transactions “will be selected unless the data necessary to apply - 107 - * * * [the CUT method are] relatively incomplete or unreliable.” Ibid. “In this re- gard,” Treasury said, “the CPM generally would be considered a method of last resort.” Ibid.37 Treasury’s reference to the CPM as a “method of last resort” is predicated on the assumption that “adequate data” are available to apply the CUT method. The 1968 regulations had directed use of the CUP method so long as there existed uncontrolled transactions involving the “same or similar intangible property under the same or similar circumstances.” 26 C.F.R. sec. 1.482-2(d)(2)(ii) (1969). The current regulations, by contrast, indicate that the CUT method has an especially high degree of reliability only “[i]f an uncontrolled transaction involves the trans- fer of the same intangible under the same, or substantially the same, circumstances as the controlled transaction.” Sec. 1.482-4(c)(2)(ii), Income Tax Regs. (emphasis added). Petitioner has identified no pricing data for transactions with unrelated par- ties that “involve[] the transfer of the same intangible”--viz., the trademarks, brand names, patents, logos, secret formulas, and proprietary manufacturing processes 37 Treasury noted that methods based on gross margin (e.g., the resale price method) may likewise offer a high degree of reliability, again assuming the avail- ability of adequate data. T.D. 8552, 1994-2 C.B. 93, 109. Neither party suggests that a method based on gross margins is the best method in this case. - 108 - used to produce Coca-Cola, Fanta, Sprite, and the Company’s other branded bev- erage products. Thus, the circumstances that caused Treasury to refer to the CPM as a “method of last resort” do not exist here. See sec. 1.482-5(e), Example (4), Income Tax Regs. (treating the CPM as “the best method” for determining an arm’s-length royalty for the transfer of intangibles to a foreign affiliate that per- forms routine manufacturing functions). In short, as the preamble elsewhere explains, “[t]he final regulations make it clear that the CPM is subject to the same considerations as any other method.” T.D. 8552, 1994-2 C.B. at 109. “[T]here is no strict priority of methods, and no method will invariably be considered to be more reliable than others.” Sec. 1.482- 1(c)(1), Income Tax Regs. The reliability of any particular method depends on “the facts and circumstances” of each case, especially on “the quality of the data and assumptions used in the analysis” and “the degree of comparability between the controlled transaction (or taxpayer) and any uncontrolled comparables.” Id. subpara. (2). We accordingly proceed to evaluate respondent’s application of the CPM in the light of those considerations, with no thumb on the scale in favor of or against that methodology. - 109 - IV. Respondent’s Bottler CPM Each party relies on expert testimony to establish an arm’s-length price for the transfer of petitioner’s intangibles. Expert testimony is admissible where it assists the Court to understand the evidence or to determine a fact in issue. See Fed. R. Evid. 702; ASAT, Inc. v. Commissioner, 108 T.C. 147, 168 (1997). The Court has broad discretion to evaluate the cogency of an expert’s analysis. See Gibson & Assocs., Inc. v. Commissioner, 136 T.C. 195, 229-230 (2011). We are not bound by any particular expert’s opinion, and we will reject expert testimony to the extent it is contrary to the judgment we form on the basis of our understand- ing of the record as a whole. See id. at 230. In support of his position, the Commissioner relies chiefly on the expert report prepared by Dr. Newlon. He determined that the supply points (other than the Egyptian supply point) enjoyed levels of profitability unjustified by the eco- nomic functions they performed. They engaged almost exclusively in manufac- turing, and petitioner’s experts agreed that this was a routine activity that could be benchmarked to the activities of contract manufacturers. Two of petitioner’s ex- perts, Drs. Cragg and Unni, applied an 8.5% markup on costs to determine an appropriate arm’s-length return for the supply points’ concentrate manufacturing function. - 110 - The Brazilian, Chilean, and Egyptian supply points employed personnel who engaged in other activities, including marketing, sales, and finance. To the extent the supply points performed nonmanufacturing activities, they discharged functions similar to those performed by ServCo employees. The ServCos were compensated for their employees’ services on a cost-plus basis, with an average markup of 6% to 7%. Petitioner does not question the arm’s-length character of the ServCos’ compensation. The arm’s-length compensation for the totality of the services performed by the supply points would thus seem to be somewhere between 6% and 8.5% above their costs. But the profits the supply points enjoyed vastly exceeded that range. One needs no more than a back-of-the-envelope calculation to make this clear. The seven supply points for 2007-2009 reported total revenues of roughly $31.71 billion, or an average of $10.57 billion annually. See supra p. 72. They re- ported total gross profits for those years of $25.44 billion, or an average of $8.48 billion annually, after offsetting COGS and other costs of about $2.09 billion an- nually. See supra p. 73. With a few exceptions (chiefly for the Egyptian supply point) their gross profit margins ranged between 75% and 90% each year. See id. The seven supply points for 2007-2009 reported average business expenses --consisting mostly of expenses incurred by the ServCos and assigned to the sup- - 111 - ply points--of $4.63 billion annually. See supra p. 74. Adding those expenses to their COGS and other costs, we derive average total costs of $6.72 billion per year ($4.63 billion + $2.09 billion). Their average annual revenues thus exceeded their average annual costs by $3.85 billion ($10.57 billion ! $6.72 billion). They thus enjoyed, on average, a markup on costs of about 57% ($3.85 billion ÷ $6.72 billion). That return is almost seven times higher than the 8.5% return that repre- sents the arm’s-length value of the manufacturing activities they performed. A test conducted by Dr. Newlon confirms that the supply points’ net profit levels were economically inexplicable. He compared their profitability to that of companies classified by Standard & Poor’s as manufacturers of food, beverages, and related products. After eliminating smaller entities and those with insufficient data, he examined a universe of 996 companies that included the Company, the Pepsi-Cola Company, Nestlé, major brewing companies, bottlers, and virtually every other major food or beverage manufacturer worldwide. Dr. Newlon compared the average ROAs for these 996 companies to the supply points’ average ROAs for 2007-2009. He plotted his results on a histo- gram, whose horizontal axis shows the companies’ ROAs and whose vertical axis shows the number of companies that fall into each 5% increment of ROAs: - 112 - - 113 - As this figure shows, the Company as a whole, with an ROA of 55%, was highly profitable; it outperformed 968 (or 97%) of the 996 tested companies. But the supply points had much higher ROAs than the Company, reflecting the shift of intra-Company profits to them. The Irish, Brazilian, Chilean, and Costa Rican supply points, with average ROAs of 215%, 182%, 149%, and 143%, respectively, had ROAs higher than any of the 996 companies in the comparison group--literal- ly off the high end of the bell curve. The Swazi and Mexican supply points, with ROAs of 129% and 94%, respectively, had ROAs higher than 99% of the compa- nies in the comparison group. These data prompt two obvious questions: Why are the supply points, engaged as they are in routine contract manufacturing, the most profitable food and beverage companies in the world? And why does their profit- ability dwarf that of TCCC, which owns the intangibles upon which the Compa- ny’s profitability depends? Properly concluding that these results did not clearly reflect income, the Commissioner reallocated income between the “controlled taxpayers”--viz., be- tween the supply points and petitioner--employing a CPM that treated independent Coca-Cola bottlers as comparable parties. The CPM is a specified method for de- termining “[t]he arm’s length consideration for the transfer of an intangible.” Sec. 1.482-4(a), Income Tax Regs. Under the CPM, the determination of an arm’s- - 114 - length result is based on “the amount of operating profit” that a controlled taxpay- er (the “tested party”) would earn if its “profit level indicator were equal to that of an uncontrolled comparable.” Id. sec. 1.482-5(b)(1). An “uncontrolled compara- ble” is an unrelated taxpayer that engages in similar business activities under sim- ilar circumstances. See id. para. (a). Reported operating profits in excess of the benchmark profit level indicator are allocated to the other controlled party (here, petitioner). See id. sec. 1.482-5(e), Example (4). Respondent’s expert, Dr. Newlon, employed a CPM that benchmarked the supply points’ profits to the profits earned by independent Coca-Cola bottlers. In the initial report that he prepared for the IRS, Dr. Newlon selected 18 independent Coca-Cola bottlers, headquartered in 10 different countries, that had qualified au- ditors’ opinions for 2007-2009. In the expert witness report that he prepared for trial, he expanded his list to 24 bottlers. He concluded that the best profit level in- dicator (PLI) was an ROA, i.e., “the ratio of operating profit to operating assets.” See id. para. (b)(4)(i). The regulations list this ratio, also called the “rate of return on capital employed,” as a specified PLI for comparing operating profits of the tested party and the uncontrolled comparables. Ibid. We conclude that the Commissioner did not abuse his discretion by using the bottler ROA to reallocate income between petitioner and the supply points. - 115 - First, a CPM analysis was appropriate given the nature of the assets owned and the activities performed by the controlled taxpayers. Second, the Commissioner sel- ected appropriate comparable parties. Third, the Commissioner computed and ap- plied his ROA using reliable data, assumptions, and adjustments. We find that the bottlers in many respects enjoyed an economic position superior to that of the sup- ply points, which would justify for the bottlers a higher relative return. Dr. New- lon’s choice of methodology was thus conservative. A. Reasonableness of CPM Analysis “It has been shown that the choice of transfer pricing method is largely driv- en by the question: Who in the group owns the valuable non-routine intangibles?” Marc M. Levey, U.S. Taxation of Foreign Controlled Businesses, para. 8.06[1][e] (WG&L 2019). Intangible assets, especially unique intangibles of the kind involv- ed here, tend to defy easy valuation for at least three reasons. First, they derive their high value from their ability to exclude comparable external transactions. See United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 392 (1956).38 38 Comparable external transactions involving high-value intangibles may occasionally exist. See Amazon.com, Inc., 148 T.C. at 167 (finding a reliable CUT where the taxpayer built for an unrelated party a website that included all the features of the taxpayer’s own website and enabled the other party to sell its retail products online). But external transactions involving a company’s “crown jewels” are quite rare. - 116 - Second, the value of such intangibles is not easily delimited from the taxpayer’s complementary business assets. Third, the profit potential of such intangibles is not reliably tethered to any cost or other input. See Yariv Brauner, “Value in the Eye of the Beholder: The Valuation of Intangibles for Transfer Pricing Purposes,” 28 Va. Tax Rev. 79, 89 (2008). Where controlled transactions implicate high-value intangibles, therefore, the most reliable transfer pricing method is often one that avoids any direct valua- tion of those intangibles. See Levey, supra, para. 8.06[2][f]. And the choice among indirect methods typically depends on who owns the intangibles. The “profit split” method will often be appropriate where both controlled taxpayers contribute significant intangible property. See sec. 1.482-8(b), Example (8), In- come Tax Regs. By contrast, the CPM will often be preferred where only one of two entities contributes meaningful intangible property. See id. Example (9). This case is particularly susceptible to a CPM analysis because petitioner owned virtually all the intangible assets needed to produce and sell the Company’s beverages. Petitioner was the registered owner of virtually all trademarks cover- ing the Coca-Cola, Fanta, and Sprite brands and of the most valuable trademarks covering the Company’s other products. Petitioner was the registered owner of nearly all of the Company’s patents, including patents covering aesthetic designs, - 117 - packaging materials, beverage ingredients, and production processes. Petitioner owned all rights to the Company’s secret formulas and proprietary manufacturing protocols. Petitioner owned all intangible property resulting from the Company’s R&D concerning new products, ingredients, and packaging. Petitioner was the counterparty to all bottler agreements, giving it ultimate control over the distribu- tion system for the Company’s beverages. And most ServCo agreements executed after 2003 explicitly provided that “any marketing concepts developed by third party vendors are the property of Export,” thus cementing petitioner’s ownership of marketing intangibles subsequently developed outside the United States. The supply points, by contrast, owned few (if any) valuable intangibles. Their agreements with petitioner explicitly acknowledged that TCCC owned the Company’s trademarks, giving the supply points only a limited right to use peti- tioner’s IP in connection with manufacturing and distributing concentrate. Four supply points, including the Irish supply point, showed zero trademarks or other intangible assets on their balance sheets. See supra p. 69. None of the other sup- ply points developed significant intangibles in-house. Only the Brazilian supply point showed meaningful IP, but its IP of $190 million, attributable largely to locally owned beverage brands, bottler franchise rights, and goodwill, represented just 11% of its total assets. - 118 - The supply points’ agreements with petitioner granted them rights to pro- duce and sell concentrate, and petitioner refers to these rights as “franchise rights.” But these agreements were terminable by petitioner at will. No supply point en- joyed any form of territorial exclusivity, and no supply point was granted any right, express or implied, to guaranteed production of concentrate. The imperma- nence of their rights was demonstrated in practice: From 1986 through 2009 TCCC closed (or shifted production away from) 18 supply points, but no supply point received any compensation for this loss of production and income. We find no factual support for the assertion that the supply points owned valuable intangi- ble assets in the form of “franchise rights.” The Coca-Cola System is an extremely sophisticated and complex opera- tion, but nearly all of its complexity is external to the supply points. They engag- ed in routine manufacturing, mixing ingredients specified by petitioner according to manufacturing protocols supplied by petitioner. In essence, they were wholly- owned contract manufacturers. The CPM is ideally suited to this scenario: The CPM evaluates the profitability only of the tested party--here, the supply points-- and it can thus determine an arm’s-length profit range for the supply points with- out attempting a direct valuation of the Company’s hard-to-value intangible assets. - 119 - There is a possible wrinkle to the CPM’s application where controlled trans- actions involve multiple “controlled taxpayers” apart from the tested party. In that event, determining an arm’s-length profit range for the tested party does not end the inquiry, for the remaining profit must then be divided between the other en- tities. This further step is not difficult here. Putting aside the “split invoicing” allocations, petitioner and respondent agree that the ServCos--the other group of “controlled taxpayers”--received arm’s-length compensation for the functions they performed. Thus, once the CPM determines an arm’s-length profit level for the supply points, their residual income is necessarily reallocated to petitioner, the owner of the valuable intangibles. Petitioner has identified no pricing data for transactions with unrelated par- ties that “involve[] the transfer of the same intangible”--viz., the trademarks, brand names, logos, patents, secret formulas, and proprietary manufacturing processes used to produce Coca-Cola, Fanta, Sprite, and the Company’s other branded bev- erages. See sec. 1.482-4(c)(2)(ii), Income Tax Regs. The reliability of any CUT method is thus considerably reduced here.39 Once it has been ruled out, the CPM is preferable in principle to the “profit split” method because petitioner owns 39 We address infra pp. 191-197 petitioner’s contention that “master fran- chising transactions” entered into by companies like McDonald’s provide a reli- able CUT. - 120 - virtually all of the relevant intangibles. See sec. 1.482-8(b), Examples (8) and (9), Income Tax Regs. B. Selection of Bottlers as Comparable Parties Our conclusion that the controlled activity was highly susceptible to a CPM analysis does not suffice to show that the Commissioner’s CPM is “the best meth- od” for ascertaining an arm’s-length price. See sec. 1.482-1(c)(1), Income Tax Regs. We must determine next whether the independent Coca-Cola bottlers that the IRS selected are reasonably treated as “comparable” to the supply points for this purpose. “The determination of the degree of comparability between the tested party and the uncontrolled taxpayer depends upon all the relevant facts and circum- stances.” Id. sec. 1.482-5(c)(2)(i). Like other transfer pricing methods, the CPM requires consideration of the general comparability factors set forth in section 1.482-1(d), Income Tax Regs. These factors include (i) functions performed, (ii) contractual terms, (iii) risks, (iv) economic conditions, and (v) property employed or transferred. Id. subpara. (1). “Each of these factors must be considered in de- termining the degree of comparability between * * * taxpayers and the extent to which comparability adjustments may be necessary.” Id. subpara. (3). - 121 - The CPM is keyed to operating profit, which “represents a return for the in- vestment of resources and assumption of risk.” Id. sec. 1.482-5(c)(2)(ii). For this reason, “comparability under this method is particularly dependent on resources employed and risks assumed.” Ibid. “Moreover, because resources and risks usu- ally are directly related to functions performed,” an analysis of functions per- formed is important “in determining the degree of comparability between the tested party and an uncontrolled taxpayer.” Ibid. Other relevant factors include the parties’ respective “lines of business, the product or service markets involved, the asset composition employed (including the nature and quantity of tangible assets, intangible assets, and working capital), the size and scope of operations, and the stage in a business or product cycle.” Id. subdiv. (i). We agree with the Commissioner’s conclusion that independent Coca-Cola bottlers serve as appropriate comparable parties for purposes of a CPM/ROA anal- ysis. The bottlers are comparable to the supply points because they operated in the same industry, faced similar economic risks, had similar (but more favorable) con- tractual and economic relationships with petitioner, employed in the same manner many of the same intangible assets (petitioner’s brand names, trademarks, and lo- gos), and ultimately shared the same income stream from sales of petitioner’s bev- erages. - 122 - The “functions performed” by the bottlers resembled those performed by the supply points, except that the bottlers performed those functions at a greater scale. See id. secs. 1.482-1(d)(3)(i), 1.482-5(c)(2)(ii). Both sets of companies engaged in the manufacture and distribution of products in the NARTD beverage business. Both sets of companies thus engaged in the same “line of business” and ultimately served the same “product * * * markets.” See id. sec. 1.482-5(c)(2)(i). The manu- facturing activity in both cases was routine, consisting largely of mixing ingredi- ents according to detailed protocols supplied by petitioner. See id. para. (e), Ex- ample (4)(ii) (concluding that “the ratio of operating profit to operating assets is an appropriate profit level indicator” where the tested party “engages in relatively routine manufacturing activities”); id. sec. 1.482-8(b), Example (8) (same).40 Quality control was important to both sets of companies, and they engaged in comparable forms of quality control at the plant level, subject to standards pre- scribed and inspections conducted by petitioner. The bottlers’ distribution fun- ction was more complex and benefited from greater economies of scale. The bot- 40 Petitioner does not appear to dispute the application of an ROA for the supply points’ manufacturing activity. Indeed, it agrees that an ROA “is most re- liable when operating assets play a significant role in generating operating income of the tested party.” But petitioner urges that the bottlers are noncomparable to the supply points because the latter supposedly possessed immensely valuable off- book assets in the form of “marketing intangibles.” We address that contention infra pp. 150-172. - 123 - tlers were responsible for securing, retaining, and distributing to thousands of re- tail customers, whereas the supply points distributed to a relatively small number of bottlers that had preexisting contracts with petitioner. In that respect the bot- tlers would be deserving of a higher ROA than the supply points, so Dr. Newlon’s selection of the bottlers as comparable parties was to that degree conservative. With respect to “contractual terms,” see id. sec. 1.482-1(d)(1)(ii), Dr. New- lon’s selection of the bottlers as comparables was appropriate, but conservative. Formally speaking, the bottlers and the supply points operated under similar con- tracts with petitioner. Both sets of contracts recognized that TCCC owned the trademarks and other IP needed to produce the beverages. The bottlers and the supply points enjoyed limited (and essentially equivalent) rights to use petitioner’s IP in carrying our their manufacturing and distribution functions. And TCCC reserved the rights to terminate both types of agreements on no more than a few months notice. Practically speaking, however, bottlers enjoyed more favorable contract terms. The supply points generally had very short-term contracts that petitioner could (and frequently did) terminate at will. The contracts gave the supply points no territorial exclusivity, and other supply points routinely sold into their “home” markets. Their contracts afforded them no guaranteed production: The record - 124 - reflects dozens of production shifts among supply points in the decades after 1980, and TCCC closed (or shifted production away from) at least 18 supply points be- tween 1986 and 2009. The supply points had no say in this, with all production shifts being dictated by CPS in Atlanta. The largest bottlers, which Dr. Newlon included in his sample, had ten-year contracts with petitioner. Although TCCC reserved the rights to terminate their contracts on short notice, the mutual dependence between the Company and its bottlers ensured that bottler agreements were almost always renewed. As CCE, one of the top three bottlers, stated in its public filings: “We believe that our inter- dependent relationship with TCCC and the substantial cost and disruption to that company that would be caused by nonrenewals ensure that these agreements will continue to be renewed.” For this reason most major bottlers, including CCE, Hellenic, and Coca-Cola FEMSA, assigned to their bottling contracts an indefinite useful life for accounting and financial statement purposes. The bottlers’ contracts with petitioner also afforded them a high degree of territorial exclusivity. TCCC’s agreements with most bottlers included a geo- graphically defined market in which the bottler was granted exclusive rights to produce and sell beverages. For legal reasons explicit exclusivity clauses were generally absent from European bottler agreements, but in practice bottlers res- - 125 - pected each other’s notional territories and rarely attempted to sell into them. Be- cause the bottlers in practice enjoyed more favorable contract terms both as to dur- ation and exclusivity, the bottlers would be deserving of a higher ROA than the supply points. In this respect Dr. Newlon’s selection of the bottlers as compar- ables was again conservative. In many respects the bottlers and the supply points faced comparable “econ- omic conditions.” Sec. 1.482-1(d)(1)(iv), (3)(iv), Income Tax Regs. Both sets of companies engaged in the NARTD business, with their revenues ultimately depen- dent on retail sale of the Company’s beverages. They thus faced similar risks from economic cycles and the ebb and flow of consumer demand. The bottlers selected by Dr. Newlon, like the supply points, did business on multiple continents, and Dr. Newlon segmented the bottlers and supply points into corresponding geographic regions, thus minimizing the risk of noncomparability owing to uniquely local market factors. See supra p. 81. In determining comparability we must consider in particular any “economic conditions that could affect the prices that would be charged or paid, or the profit that would be earned” by the tested party and the uncontrolled comparable. Id. para. (d)(3)(iv). In this regard, we consider (among other things) “[t]he extent of - 126 - competition in each market” and “[t]he alternatives realistically available to the buyer and seller.” Id. subdiv. (iv)(F) and (H). The supply points sold concentrate to preordained buyers in a global market where competition was robust. Bottlers contracted with TCCC and agreed to pur- chase concentrate from whichever supply point TCCC dictated. Most supply points had unused production capacity. There were no substantial barriers to pre- vent TCCC (or others) from quickly building more capacity, and none of the sup- ply points enjoyed any significant local advantage. In other words, the supply points were easily replaceable--and regularly were replaced--at relatively low cost to petitioner. Indeed, when petitioner shifted concentrate production from one supply point to another, without building a new manufacturing plant, the cost of replacing the losing supply point was virtually zero. In an arm’s-length negotiation over price, therefore, the supply points would have occupied a very weak bargaining position vis-à-vis TCCC. See E.I. du Pont de Nemours & Co., 351 U.S. at 392 (“It is inconceivable that price could be con- trolled without power over competition or vice versa.”). In this respect the bottlers occupied a vastly superior position. Because they enjoyed de facto territorial exclusivity, they could restrict the “extent of competi- tion” for sale of Coca-Cola beverages in their markets. Sec. 1.482-1(d)(3)(iv)(F), - 127 - Income Tax Regs. And because the bottlers created and managed their local dis- tribution networks, TCCC had few “alternatives realistically available” to secure local trade marketing and product placement in stores. See id. subdiv. (iv)(H). Collectively these economic conditions gave the bottlers market power in their respective territories that enabled them to secure, in negotiations with the Com- pany, an equitable split of System profit, sometimes as high as 55% in the bottler’s favor. See supra p. 62. The supply points had no comparable bargaining power. We find that the bottlers and the supply points were reasonably (albeit not perfectly) comparable in terms of “resources employed.” See id. sec. 1.482- 5(c)(2)(ii). Both sets of companies manufactured and distributed products in the NARTD beverage business, and they used a similar mix of resources to discharge those functions. The bottlers were generally larger companies and their distribu- tion activities were more complex; they accordingly had many more employees than the supply points, and PPE represented a larger percentage of their operating assets. But the two sets of companies were similar in terms of the relationship be- tween their operating assets and their total resources. The bottlers were highly comparable to the supply points in terms of “risks assumed.” Ibid.; see id. sec. 1.482-1(d)(1)(iii). Both sets of companies faced the same risk to capital employed, because they used similar resources to perform sim- - 128 - ilar functions uniquely demanded by the NARTD market. Although Coca-Cola bottlers occasionally bottled other companies’ beverages, their diversification in this respect was minimal. As was true for the supply points, therefore, their pro- fitability ultimately depended on income generated by retail sale of the Company’s products. Because their employed capital could not easily be redirected to non- Company products, let alone to products outside the NARTD market, the bottlers and the supply points faced essentially the same risks associated with the success or failure of the NARTD beverage business and the Company’s products in particular. See id. subpara. (3)(iii)(A)(1), (2), (6). Dr. Newlon observed (and we agree) that the bottlers in some respects as- sumed greater risks than the supply points. The supply points’ operating assets in- cluded large amounts of cash and accounts receivable from creditworthy bottlers. The bottlers’ operating assets, by contrast, consisted chiefly of illiquid and diffi- cult-to-repurpose PPE. And their receivables from small retailers were subject to greater collection risk. See id. subdiv. (iii)(A)(4), (6). The bottlers also bore greater risk as a consequence of their more robust dis- tribution functions. The supply points in essence had guaranteed customers--the bottlers--and TCCC dictated the bottlers to which their concentrate was sold. The bottlers, by contrast, had to acquire and retain retail customers, secure desirable - 129 - product placement in stores, and negotiate favorable product pricing, particularly with smaller retailers. Cf. id. subdiv. (iii)(A)(2). The bottlers’ distribution func- tion, as compared with the supply points’ distribution function, entailed greater uncertainty and thus greater risk. 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 ServCo mar- keting 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 re- ceived (by shifting concentrate production among them) and how much expense each supply point was charged (by way of DME and “fees and commissions” al- located to it). Since the flow of revenue and marketing expenses to the supply points was controlled by TCCC, 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. - 130 - Dr. Newlon acknowledges (and we agree) that the economic position of the supply points and the bottlers differed in a few respects. Petitioner emphasizes that the two sets of companies occupied different points in the Company’s supply chain and did business at different “levels of the market”: Supply points sold con- centrate to bottlers, and bottlers sold finished beverages to distributors and retail- ers. See id. subdiv. (iv)(C). But petitioner has failed to show how these distinc- tions affect the comparability of the functions the two sets of companies dis- charged or the operating profit they could earn. The supply points and the bottlers both performed manufacturing and distribution functions in the NARTD beverage business, and both sold their products at the “wholesale” level. See ibid. Petitioner contends that the bottlers were “marketing-light businesses,” whereas the supply points assertedly had “marketing-intensive operations.” As explained more fully below, we disagree with the former contention: The evi- dence at trial showed that bottlers, in the aggregate, paid about as much for trade marketing annually as TCCC and its affiliates together paid for consumer market- ing. See infra p. 188. And the latter contention is simply inaccurate. The supply points did not have “marketing-intensive operations” because (apart from the Bra- zilian supply point) they engaged in no marketing operations. All consumer mar- keting activities were undertaken by TCCC and the ServCos; the supply points - 131 - simply had these costs charged to their books via inter-company accounting. In assessing the comparability of the bottlers and the supply points, we focus on the economic functions they actually performed and the risks they actually assumed, not on inter-company charges made by their parent. See id. subdiv. (iii)(B). Petitioner’s experts developed various financial ratios--involving revenue, inventory, asset turnover, fixed assets, and marketing expenses--in an effort to show that the supply points were not comparable to the bottlers. Some of these computations relied on questionable assumptions. For example, Dr. Timothy Luehrman calculated that the supply points’ receivables represented 33% of their adjusted total assets, whereas the comparable figure for the bottlers was 15%. But as Dr. Newlon noted, the receivables shown by the supply points “may include items that were not trade-related and/or settled in the ordinary course of business under normal commercial terms.” See infra pp. 138-139. Dr. Luehrman calculated that marketing expenses as a percent of net reve- nue were 30% for supply points and only 20% for bottlers. But as noted previous- ly, this ratio is not meaningful for the supply points because they did not them- selves engage in significant marketing activities; they simply had marketing costs assigned to them. Because petitioner had complete discretion in deciding how - 132 - many dollars of marketing costs would be assigned to each supply point, these numbers have no grounding in operational reality. Dr. Luehrman also calculated that net revenue, as a percentage of adjusted assets and PPE, was much higher for the supply points than for the bottlers. But these ratios have a question-begging aspect to them. The supply points’ net reve- nue consisted chiefly of the concentrate price paid by bottlers. But by paying the concentrate price the bottlers secured, not only the physical beverage base, but also the entire package of rights and privileges they needed to operate efficiently as Coca-Coca bottlers. Cf. id. sec. 1.482-1(d)(3)(v) (noting that transferred pro- perty may include intangible property embedded in tangible property). This pack- age included the right to use TCCC’s trademarks, access to TCCC-approved sup- pliers, access to critical databases and marketing materials, and the expectation of on-going consumer marketing support from TCCC and the ServCos. In short, the net revenue received by the supply points was artificially in- flated because it represented compensation to petitioner for its intangible contribu- tions as well as to the supply points for their manufacturing activity. Since the supply points’ net revenue was artificially inflated, financial ratios with net reve- nue in the numerator do not furnish useful evidence in judging the supply points’ comparablity to the bottlers. Indeed, the objective of this transfer pricing exercise - 133 - is to ascertain an arm’s-length division of concentrate revenue between petitioner and the supply points. If Dr. Luehrman had used “net revenue less arm’s-length royalty” as the numerator in his ratios, those ratios would look very different. Financial ratios can be useful. But depending on which ratios one chooses and what assumptions and adjustments one makes, they can be used to support ra- ther different conclusions. At the end of the day, we did not find the financial ra- tios constructed by petitioner’s experts to have much probative value in determin- ing the bottlers’ comparability to the supply points. Comparability under the regulations is principally judged on the basis of functions performed, contractual terms, risks assumed, economic conditions, and assets employed. See id. subpara. (1). We have found the bottlers highly com- parable to the supply points in all five respects. Although concededly there are differences between the two sets of companies, we find on balance that these dif- ferences tend to make the bottlers deserving of a higher ROA than the supply points. To that extent Dr. Newlon’s CPM will tend to overcompensate rather than to undercompensate the supply points, and it is therefore conservative. C. Data, Assumptions, and Comparability Adjustments Respondent has gone a long way toward establishing the reasonableness of his transfer pricing methodology by showing that the tested activity is highly sus- - 134 - ceptible to a CPM and that the tested parties (supply points) and uncontrolled comparables (independent Coca-Cola bottlers) engaged in similar business activi- ties under similar circumstances. However, we cannot conclude that his results are reasonable without reviewing the selection and quality of his data and the assump- tions employed to bridge any gaps in those data. See sec. 1.482-5(c)(2)(iv), (3), Income Tax Regs. We must also ensure that the Commissioner applied the CPM in a manner that accounts for all of the supply points’ relevant business activity. 1. Selection of Bottlers In preparing his audit report Dr. Newlon searched multiple databases for companies whose primary function was the production and bottling of Coca-Cola trademarked beverages. He began by searching for companies classified under either of two Standard Industrial Classification (SIC) Codes: 2086 (Bottled and Canned Soft Drinks and Carbonated Waters) and 2087 (Flavoring Extracts and Flavoring Syrups, Not Elsewhere Classified). He supplemented that search by reviewing 284 additional companies identified in bottler agreements with TCCC or listed as members of bottling associations (e.g., the American Beverage Associ- ation and the Coca-Cola Bottlers Association). This search generated 508 bottlers for consideration. - 135 - Dr. Newlon eliminated companies that primarily bottled non-Coca-Cola products, that produced alcoholic beverages, or in which TCCC held a controlling interest. He also eliminated companies that lacked audited financial statements for 2007-2009 or otherwise had insufficient data to perform an ROA. After reviewing business descriptions of each company, Forms 10-K, annual reports, and other fi- nancial information, Dr. Newlon ultimately selected 18 independent Coca-Cola bottlers, headquartered in 10 countries, but with operations on every continent ex- cept Antarctica. See supra pp. 80-81. His list included the four largest Coca-Cola bottlers in the world: CCE, Hellenic, Coca-Cola FEMSA, and Coca-Cola Amatil. Observing that bottlers doing business in Latin America tended to have high ROAs and that bottlers doing business in East Asia tended to have low ROAs, Dr. Newlon segmented the 18 bottlers into geographic regions to avoid bias when making comparison with the supply points. In his audit report he calculated ROAs for those 18 bottlers as follows: Interquartile range ROA (2007-2009) Bottler segment 25th Percentile Median 75th Percentile All bottlers (18) 7.4% 18.0% 31.8% Non-East Asian bottlers (13) 17.9% 24.5% 32.5% Latin American bottlers (6) 31.8% 34.3% 40.6% Non-East Asian bottlers outside Latin America (7) 14.4% 17.9% 24.5% - 136 - After retaining Dr. Newlon as an expert witness for trial, respondent asked him to consider possible refinements to his methodology as new information be- came available during discovery (e.g., supply point sales data and additional bot- tler financial statements). After reviewing these data Dr. Newlon selected six ad- ditional bottlers and calculated for them ROAs as follows: A B C. Bottler Operating income Operating assets ROA% home Bottler (% of net revenue) (% of net revenue) (A÷B) China Beijing Coca-Cola Beverage Co. Ltd. 6.1 47.4 13.0 Japan Coca-Cola Central Japan Co. Ltd. 2.0 43.6 4.7 Spain Bebidas Gaseosas del Noroeste SA 9.8 47.1 20.9 Spain Compania Asturiana de Bebidas Gaseosas SA 11.5 77.3 14.8 Spain Compania Levantina de Bebidas Gaseosas SA 7.6 52.0 14.6 Spain Refrescos Envasados del Sur SA. 6.0 45.0 13.3 Adding these results to the results for the initial 18 bottlers Dr. Newlon calculated geographically segmented median ROAs as follows: Median ROA (2007-2009) Bottler segment Expert witness report (%) Audit report (%) All bottlers (24) 15.8 18.0 Non-East Asian bottlers (17) 20.9 24.5 Latin American bottlers (6) 34.3 34.3 Non-East Asian bottlers outside Latin America (11) 16.7 17.9 Petitioner does not contend that Dr. Newlon erred in the search process he employed. Nor does petitioner challenge the representativeness of Dr. Newlon’s 24-bottler sample or the reliability of the data he extracted from bottler financial statements. We conclude that Dr. Newlon’s 24-bottler sample is reliable and fairly - 137 - represents the universe of independent bottlers engaged principally in the business of bottling and distributing TCCC-branded products. 2. Computational Adjustments To implement the CPM Dr. Newlon needed to calculate “the ratio of operat- ing profit to operating assets” for the supply points and the bottlers in his sample. See sec. 1.482-5(b)(4)(i), Income Tax Regs. Financial statements provide the logical starting point for calculating these numbers. But adjustments are typically required for differences between the tested party and an uncontrolled taxpayer, e.g., in terms of their respective accounting practices. See id. para. (c)(2)(iv), (3)(ii), para. (e), Examples (5) and (6). a. Operating Assets “The term operating assets means the value of all assets used in the relevant business activity of the tested party, including fixed assets and current assets (such as cash, cash equivalents, accounts receivable, and inventories).” Id. para. (d)(6). Assets not used in the relevant business activity “include investments in subsidiar- ies, excess cash, and portfolio investments.” Ibid. Operating assets “may be mea- sured by their net book value or by their fair market value,” provided that the same method is applied consistently. Ibid. - 138 - As permitted by the regulations, Dr. Newlon calculated operating assets by using the net book value figures reported by the bottlers and the supply points. For each year, he averaged prior-year and current-year operating assets as reported by each entity. Petitioner does not contend that Dr. Newlon applied the book val- ue approach inconsistently, and we conclude that this approach was reasonable.41 When preparing his initial audit report, Dr. Newlon did not have complete information concerning the supply points’ receivables from affiliated entities. He accordingly increased each supply point’s operating assets by an estimate of its intra-Company accounts receivable, assuming that such revenues were payable to the supply points within 30 days. He computed net operating assets for the supply points and the bottlers by subtracting, from total operating assets, items they iden- tified on their financial statements as nonoperating assets. Before finalizing his expert witness report Dr. Newlon received data con- cerning the supply points’ intra-Company receivables as shown on petitioner’s books. He concluded that the receivables as thus shown “may include items that 41 The book value approach may undervalue long-held assets that have ap- preciated. This effect is nonexistent or negligible for cash and accounts receiv- able. Although land, buildings, and other PPE may have appreciated, it would be difficult to determine the FMV of these items for 24 foreign bottlers and 7 foreign supply points for each of three separate years. Since the bottlers had relatively larger investments in PPE, any understatement of their operating assets owing to use of book value would increase their ROA and thus benefit petitioner. - 139 - were not trade-related and/or settled in the ordinary course of business under nor- mal commercial terms.” In other words, Dr. Newlon concluded that a portion of the supply points’ reported receivables did not constitute “operating assets” be- cause they were not “used in the relevant business activity of the tested party.” See sec. 1.482-5(d)(6), Income Tax Regs. He accordingly adhered to the approach of his initial audit report, estimating receivables from affiliates by assuming 30- day payment terms for the supply points’ receipt of intra-Company revenue. We consider that approach to be reasonable. In his expert witness report Dr. Newlon stated that his updated computa- tions (following inclusion of six additional bottlers) incorporated “minor differ- ences in the definition of the Supply Points’ operating assets” as compared with his original audit report. Specifically, the operating assets of the Brazilian supply point were increased by about 3%; the operating assets of the Mexican supply point were decreased by about 2%; and the operating assets of the other five sup- ply points were unchanged. Because Dr. Newlon did not explain the basis or jus- tification for these revisions, we cannot evaluate them. We will accordingly re- solve doubts in petitioner’s favor by adopting Dr. Newlon’s increased operating asset figure for the Brazilian supply point, but not his reduced operating asset figure for the Mexican supply point. - 140 - b. Operating Profit Operating profit means gross profit (sales revenue less COGS) reduced by operating expenses. See sec. 1.482-5(d)(2), (4), Income Tax Regs. Operating ex- penses include “all expenses not included in * * * [COGS] except for interest ex- pense, * * * income taxes, and any other expenses not related to the operation of the relevant business activity.” Id. subpara. (3). Operating expenses generally in- clude “expenses associated with advertising, promotion, sales, marketing, ware- housing and distribution, administration, and a reasonable allowance for deprecia- tion and amortization.” Ibid. Certain items reported as operating income or ex- pense--e.g., “extraordinary gains and losses,” gain or loss on the sale of PPE, or embedded restructuring expenses--often do not relate to continuing operations and are thus generally excluded when performing these calculations. Id. subpara. (4); see Steven D. Felgran et al., “Treatment of Restructuring Expenses in the Applica- tion of CPM,” 15 Tax Mgmt. Transfer Pricing Rept. 755, 757-758 (Feb. 21, 2007). For the most part Dr. Newlon computed operating profit by adopting the classification of items as reported on the supply points’ and the bottlers’ income statements. For the supply points he reclassified some items and made certain ad- - 141 - justments to earnings and profits.42 For the bottlers he generally excluded from operating expenses any separately stated amortization, stock option expense, and other items that do not necessarily reflect operational costs. Petitioner has not as- signed error to these calculations, and we find them reasonable. Dr. Newlon also added to operating income, both for the supply points and for the bottlers, imputed interest on their non-interest-bearing liabilities (NIBLs), such as accounts payable. As Dr. Newlon observed, NIBLs “can represent a source of financing for which there is no explicit interest charge.” He also ob- served that the bottlers and the supply points had differing practices regarding payment terms, both as to their volume of accounts payable and the lag time be- fore they made payment. The regulations recognize the need for adjustments to account for such differences. See sec. 1.482-1(d)(3)(ii)(A)(7), Income Tax Regs. (requiring “an adjustment to reflect the difference in payment terms”); see also id. sec. 1.482-5(e), Example (5) (requiring adjustment to operating profit “for differ- ences in accounts receivable”). See generally Compaq Computer Corp. v. Com- 42 Specifically Dr. Newlon: (1) reclassified as nonoperating income (or ex- pense) any net gain (or loss) on the sale of PPE or other assets and any separately stated restructuring expense; (2) reclassified minor amounts of “fees and commis- sions income” and certain rental income as operating income; and (3) reclassified reported royalty expense, license fee expense, Company-owned property rental expense, and “fees and commissions expense” as operating expenses. - 142 - missioner, T.C. Memo. 1999-220, 78 T.C.M. (CCH) 20, 31 (approving adjustment for differences in payment terms). To avoid distorting the operating income comparison between the bottlers and the supply points, Dr. Newlon estimated the total cost of carrying NIBLs, for each bottler and supply point, and added that amount to each entity’s operating income. In effect, he treated all entities as if they had operated with zero-day pay- ment terms, then added to their operating income the imputed interest savings. The regulations specifically permit this manner of adjustment. See sec. 1.482- 5(c)(2)(iv), Income Tax Regs. (“[W]here there are material differences in accounts payable among the comparable parties and the tested party, it will generally be ap- propriate to adjust the operating profit of each party by increasing it to reflect an imputed interest charge on each party’s accounts payable.”). To calculate the imputed interest charge, Dr. Newlon applied a local cost of funds rate to each entity’s average volume of outstanding NIBLs. For purposes of determining the supply points’ average inter-Company payables, he assumed 30- day payment terms, an assumption that we approve. See supra p. 138. We con- clude that Dr. Newlon’s calculations for determining NIBL adjustments to operat- - 143 - ing profit, as set forth in his expert witness report, are reasonable as applied both to the bottlers and to the supply points.43 3. Implementation of the CPM/ROA Dr. Newlon made two further refinements in implementing his CPM/ROA. First, he observed that the ROAs earned by the bottlers varied somewhat by geo- graphy: Most Latin American bottlers had ROAs above 30%, whereas most East Asian bottlers had ROAs below 10%. See supra pp. 81, 135-136. Rather than computing a single median ROA for all bottlers, Dr. Newlon calculated geograph- ically segmented bottler ROAs, which the IRS applied to supply points from simi- lar regions. We agree with respondent that applying geographically segmented ROAs improves the reliability of the CPM. Bottlers in East Asia (and Japan in particu- lar) appear to have been subject to local economic conditions or financial report- ing rules that reduced their ROAs as compared to bottlers outside East Asia. 43 In his expert witness report Dr. Newlon made several technical revisions to his NIBL computations for the supply points, e.g., by computing the Irish sup- ply point’s cost of funds using a reference rate for Ireland in particular instead of an EU-wide rate. These adjustments generated very slight upward revisions to annual operating profits for the Chilean and Irish supply points, and somewhat larger downward revisions to annual operating profits for the Brazilian, Costa Rican, and Swazi supply points. We find no fault with these adjustments, which were slightly favorable to petitioner in the aggregate. - 144 - Bottlers in Latin America tended to have much higher ROAs, attributable in part to local accounting rules and (apparently) to having a more atomized retail custo- mer base. To avoid distortion, the IRS implemented Dr. Newlon’s recommendations by using the median ROA of the Latin American bottler segment to adjust the in- come of the Brazilian, Chilean, Costa Rican, and Mexican supply points. And the IRS eliminated the East Asian bottlers, with their low ROAs, when calculating the median ROA for adjusting the income of the Irish and Swazi supply points. As compared with using a uniform ROA for all supply points, these refinements were conservative in that they benefited petitioner. All in all, we conclude that respond- ent’s geographically segmented ROAs were reasonable. Dr. Newlon’s second set of refinements addressed the question of non- TCCC-branded beverages. Although the vast bulk of the supply points’ activity consisted of manufacturing concentrate for TCCC-branded beverages, several sup- ply points--in particular, the Brazilian supply point--owned the rights to other products, many of which were sold locally. As compared to TCCC’s global brands, the supply points often made significant contributions to these local brands and owned the intangible assets used in producing such beverages. - 145 - Dr. Newlon correctly concluded that the CPM should not be used to reallo- cate to TCCC any supply point income attributable to these local brands, because no TCCC intangibles were employed in generating this income. Dr. Newlon thus agreed that the supply points were entitled to keep the income generated by the intangible assets connected to brands they owned. When he prepared his initial reports, however, he lacked sufficient data to extract reliably from his CPM the income and assets attributable to brands legally owned by the supply points. Two of petitioner’s experts, Dr. Cragg and Mr. Robert Wentland, supplied the needed data. Mr. Wentland identified and extracted assets and income attri- butable to beverages whose trademarks were owned by Atlantic, which held the Costa Rican, Egyptian, Irish, and Swazi supply points.44 Similarly, Dr. Cragg identified and extracted from the revenue base of the Brazilian supply point reve- nues from brands of which it (or other non-U.S. entity) was the registered owner.45 Dr. Newlon ultimately accepted these data. In extracting revenue attribut- able to trademarks owned by the supply points, his final calculations assume that 44 These brands included A&W, Canada Dry, Diet Canada Dry, Crush, Dr. Pepper, Hires, Roses, Schweppes, and Diet Schweppes, as well as three “Cosmos” products identified as Pop, Sarsi, and Jazz. 45 These brands included Dr. Frescolita, Chinotto, Chinotto Light, Guarana Jesus, Matte Leao, Gladiador, and Vittalev. - 146 - the excluded brands (which were mostly local and less competitive) had the same profitability as TCCC’s highly profitable brands. This assumption appears reason- able, indeed conservative. We will accordingly instruct the parties, in their Rule 155 computations, to adjust the allocations of income set forth in the notice of de- ficiency to exclude income attributable to trademarks owned by the supply points, as identified by petitioner’s experts and accepted by Dr. Newlon. In sum, we conclude that the Commissioner did not abuse his discretion in reallocating income from the supply points to petitioner by use of the bottler CPM. Petitioner has not carried its “burden of showing that such determination was purely arbitrary.” Asiatic Petroleum Co., 31 B.T.A. at 1158. And because “there is substantial evidence supporting the determination, it must be affirmed.” Marc’s Big Boy-Prospect, Inc., 52 T.C. at 1092. We will accordingly sustain the realloca- tions of income determined in the notice of deficiency, subject to the adjustments noted in the preceding pages and to our holdings elsewhere in this Opinion. We are hopeful that the parties can work out the granular-level calculations in their Rule 155 computations. To the extent they are unable to do so they may seek fur- ther guidance from the Court. - 147 - V. “Split Invoicing” We now turn to respondent’s “split invoicing” allocations. In his amended answer respondent reallocated an additional $385 million of income to petitioner from six ServCos that benefited from split invoicing during 2007-2009. These reallocations produced additional deficiencies totaling about $135 million. See supra p. 6. Respondent bears the burden of proof on this issue. See supra p. 85. Where split invoicing was employed, a supply point invoiced the bottler for a portion of the concentrate price, and a local ServCo issued a separate invoice to the bottler for the remainder of the concentrate price. As a result of this practice the participating supply points lost revenue, which was diverted to the ServCos. Respondent determined that six ServCos, doing business in South America and Turkey, received “excess income” as a result of this practice--that is, income in excess of the cost-plus compensation to which they were entitled under their normal contracts with Export or TCCC. Under their normal contracts with TCCC and Export, ServCos typically received reimbursement of (but no markup on) third-party expenses, e.g., DME, and received an average markup of 6% to 7% on costs other than DME. To cal- culate the arm’s-length compensation that the six ServCos should have received from the bottlers, Dr. Newlon initially assumed that each ServCo was entitled to a - 148 - 7% markup on its non-DME costs. He further estimated that 31.7% of the ServCos’ expenses were non-DME costs attracting the 7% markup. In post-trial briefing respondent appears to have adjusted these computations to reflect the markups actually appearing in the ServCo agreements or (if no markup was specified) an estimated markup of 5% on non-DME costs. He also made certain revisions to reflect updated financial data. The services for which the ServCos were compensated via split invoicing were indistinguishable from the services that they rendered directly to TCCC and Export. Petitioner agrees that the ServCos’ contracts with Export and TCCC re- flected arm’s-length terms. Petitioner accordingly concedes that the six ServCos received excess income from the bottlers, and it does not appear to dispute the amounts of excess income as ultimately calculated by respondent. In its opening brief petitioner agreed that “split invoicing must be accounted for in determining the arm’s-length royalties” payable by the supply points to TCCC. In effect, petitioner contended that the supply points should be deemed to have received (and should be allowed to retain) the excess income derived by the ServCos. In its answering brief petitioner “agree[s] that the Court should address split invoicing in the manner in which the Court addresses the main royalty issue.” - 149 - In deciding the “main royalty issue,” we have held that the arm’s-length compensation for the supply points is capped at the amounts calculated under the Commissioner’s bottler CPM. All income the supply points received in excess of that benchmark must be reallocated to petitioner as additional royalties. To the extent the supply points are deemed to have received the excess income diverted to the six ServCos, that additional income must likewise be reallocated to petition- er as additional royalties. Or alternatively, as respondent contends, the ServCos’ excess income should be allocated to petitioner as royalties paid by the bottlers for use of petitioner’s intangible property. On either theory, in short, the bottom line is the same. Since the extra in- come received by the six ServCos was in excess of their arm’s-length income, and since it cannot be allocated to the supply points without exceeding their arm’s- length income, it must necessarily be reallocated to petitioner as the owner of the valuable intangibles. We thus hold that respondent has carried his burden of proof with respect to split invoicing. We leave it to the parties, in their Rule 155 compu- tations, to calculate the exact amount of the additional deficiencies in a manner consistent with the evidence. - 150 - VI. Petitioner’s Arguments A. Supposed “Marketing Intangibles” Petitioner’s principal contention is that the supply points owned immensely valuable off-book assets, in the form of “marketing intangibles,” that Dr. Newlon neglected to consider when performing his CPM analysis. From this premise peti- tioner derives two conclusions. First, it urges that the CPM was not an appropriate transfer pricing methodology because, with respect to each transaction, both con- trolled parties--each supply point as well as petitioner--allegedly contributed non- routine intangibles to the production of TCCC-trademarked products. Second, if a bottler CPM were deemed appropriate, petitioner contends that these supposed marketing intangibles would have to be included among the supply points’ “oper- ating assets” for purposes of performing the ROA computations. Petitioner acknowledges that TCCC was the registered legal owner of virtu- ally all of the trademarks and other intangible assets needed to manufacture and sell Coca-Cola, Fanta, Sprite, and other TCCC-trademarked beverages in foreign markets. But petitioner contends that these were in effect “wasting assets.” What kept TCCC’s products fresh in consumers’ minds, petitioner says, were the bil- lions of dollars spent annually on television advertisements, social media, and other forms of consumer marketing. Without ongoing consumer advertising, the - 151 - trademarks, secret formulas, and proprietary manufacturing processes owned by TCCC would allegedly have suffered over time a hollowing out of value. Cf. Nestlé Holdings, Inc. v. Commissioner, T.C. Memo. 1995-441, 70 T.C.M. (CCH) 682, 696 (“Trademarks lose substantial value without adequate investment, man- agement, marketing, advertising, and sales organization.”), supplemented by T.C. Memo. 2000-374, aff’d in part, rev’d in part and remanded on other grounds, 152 F.3d 83 (2d Cir. 1998). Petitioner accordingly urges that the Court should look beyond TCCC’s legal ownership of these assets and focus instead on supposed “marketing intangibles” generated by the expenditure of advertising dollars. Dr. Cragg, one of petitioner’s experts, refers to these supposed assets as “marketing- related IP” or “IP associated with trademarks.” The Company’s consumer marketing in foreign markets was undertaken by the ServCos, with assistance from third-party media and creative design firms. Global marketing campaigns were conceived by HQ in Atlanta, which distributed the material to the ServCos for customization to their local markets. Under the “charter model,” HQ arranged for certain ServCos to take the lead in designing advertising content around holidays and sports events, then shared this material with other ServCos. - 152 - Most ServCos had contracts with Export, a domestic subsidiary of TCCC, and they invoiced Export for the marketing costs they incurred. BU leadership and finance personnel then initiated inter-company charges that placed on the books of each supply point, as they deemed appropriate, an allocated portion of the sums that the ServCos had invoiced to Export. As explained supra p. 74, the amounts thus allocated to the supply points varied widely, with no apparent rela- tionship to their gross revenues. Petitioner offered no clear explanation as to how this allocation methodology actually worked. Through this inter-company accounting, the supply points were charged al- located shares of the ServCos’ “fees and commissions” and third-party marketing expenses. The supply points played no role in arranging consumer marketing and had no voice in selecting or evaluating the services for which they were thus made financially responsible. In essence, they were passive recipients of charges that HQ and BU leadership put on their books. But because petitioner saw fit to put those charges on their books, petitioner’s experts assert that the supply points thereby acquired “marketing intangibles” worth tens of billions of dollars. We find no support for petitioner’s argument in law, fact, economic theory, or common sense. The regulations “explicitly stat[e] that legal ownership is the test for identifying the intangible.” DHL Corp. v. Commissioner, 285 F.3d 1210, - 153 - 1221-1222 (9th Cir. 2002) (citing section 1.482-4(f)(3)(ii)(A), Income Tax Regs.), aff’g in part, rev’g in part T.C. Memo. 1998-461. To the extent the ServCos’ consumer advertising expenditures added value, those expenditures did not create new intangible assets owned by the supply points. Rather, the advertising en- hanced the value of the trademarks and other intangible assets that were legally owned by TCCC. In urging that its unilateral allocation of costs generated valuable intangible assets owned by the supply points, petitioner is attempting to create, retroactively, something resembling a “cost sharing arrangement” of the sort permitted by other provisions of the regulations. See Amazon.com, Inc., 148 T.C. at 191-192 (dis- cussing useful life of intangibles contributed to a qualified cost-sharing arrange- ment). But petitioner and its supply points did not enter into a “qualified cost sharing arrangement.” See sec. 1.482-7A(b), Income Tax Regs. And this is sim- ply not a cost sharing case. See id. para. (a)(3); see also Ciba-Geigy Corp. v. Commissioner, 85 T.C. 172, 230 n.38 (1985) (noting that a taxpayer cannot claim to have entered into a qualified cost-sharing arrangement “unless the arrangement has been reduced to writing”). - 154 - 1. Legal Ownership In August 2006 Treasury adopted final and temporary regulations address- ing (among other things) the allocation of income from intangibles under section 482. See T.D. 9278, 2006-2 C.B. 256. After considering public comments, Trea- sury concluded that “legal ownership provides the appropriate framework for de- termining ownership of intangibles” for transfer pricing purposes. Id., 2006-2 C.B. at 267. Treasury explained that, under these regulations: [T]he “legal owner” * * * will be the controlled party that possesses title to the intangible, based on consideration of the facts and circum- stances. This analysis would take into account applications filed with a central government registry (such as the U.S. Patent and Trademark Office or the Copyright Office in the United States), any contractual provisions in effect between the controlled parties, and other legal provisions. Legal ownership, understood in this manner, provides a practical and administrable framework for determining ownership of intangibles for purposes of section 482. [Id., 2006-2 C.B. at 267- 268.] Treasury adopted in T.D. 9278 a temporary regulation governing this sub- ject. See sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., 71 Fed. Reg. 44476, 44484 (Aug. 4, 2006). The temporary regulation applied for all taxable years beginning after December 31, 2006. See id. subpara. (7), 71 Fed. Reg. 44486. The temporary regulation was replaced by a final regulation with identical - 155 - text. See sec. 1.482-4(f)(3)(i)(A), Income Tax Regs. The final regulation applies for all taxable years beginning after July 31, 2009. See id. para. (h)(1). The regulation provides that “the sole owner” of intangible property for pur- poses of section 482 will be “[t]he legal owner * * * pursuant to the intellectual property law of the relevant jurisdiction” or “the holder of rights constituting an intangible [property] pursuant to contractual terms (such as the terms of a license) or other legal provision.” Sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra. Legal or contractual ownership is not dispositive, however, if “such ownership is inconsistent with the economic substance of the underlying trans- actions.” Ibid. For the latter proposition the regulation cross-refers to section 1.482- 1(d)(3)(ii)(B), Income Tax Regs. That provision, which had been in place since 1994, sets forth general rules for analyzing contractual terms when performing a comparability analysis. It provides that contractual terms “agreed to in writing before the transactions are entered into will be respected if such terms are consis- tent with the economic substance of the underlying transactions.” Id. subdiv. (ii)(B)(1). If that is not true, the IRS “may disregard such terms and impute terms that are consistent with the economic substance of the transaction.” Ibid. The pre- - 156 - amble to T.D. 9278 confirmed that this preexisting rule continued to apply under the final and temporary regulations adopted in 2006: The Treasury Department and the IRS anticipate that own- ership of an intangible as determined under the legal owner standard will not conflict with the simultaneous requirement that ownership under section 482 be determined in accordance with the economic substance. For example, if the economic substance of the controlled parties’ dealings conflicts with treatment of the legal owner as the owner under section 482, the Commissioner may determine owner- ship by reference to the economic substance of the transaction. * * * [Id., 2006-2 C.B. at 268.] Petitioner’s assertion that the supply points owned immensely valuable assets in the form of “marketing intangibles” is unsustainable under these regula- tions. TCCC was the registered legal owner of virtually all trademarks and other intangible assets used to manufacture and produce TCCC-branded beverages. The ServCos’ consumer marketing activities presumably enhanced the value of TCCC’s intangibles. But petitioner does not contend (and could not plausibly contend) that the supply points were the legal owners of any distinct marketing intangibles “pursuant to the intellectual property law of the relevant jurisdiction.” See sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra. Nor were the supply points “holder[s] of rights constituting an intangible [property] pursuant to contractual terms (such as the terms of a license) or other legal provision.” Ibid. TCCC’s agreements with the supply points explicitly - 157 - provided that the latter were granted no rights or ownership interest in TCCC’s intangible property. The agreements identified TCCC as the “owner” or “regis- tered proprietor” of the trademarks, and TCCC expressly “reserve[d] the right to control all things and acts related to or involving the use of [the] Trademarks.” The supply point agreed “not to do any act or thing which may impair the ownership and protection” of TCCC’s trademarks. The supply points, in short, received only a limited right to use TCCC’s intangibles in connection with their production and sales activities.46 Petitioner can point to no provision of any supply point agreement that makes the supply points “holder[s] of rights constituting an intangible [property].” Ibid. And even if the agreements contained contractual terms granting the supply points rights in TCCC’s intangible property, such rights would be illusory because petitioner could revoke the agreements. Four of the agreements could be canceled by TCCC unilaterally at any time, and no agreement could persist more than one 46 Before the years at issue the Brazilian supply point was authorized to exe- cute contracts with bottlers and was granted limited rights to sublicense TCCC’s trademarks for that purpose. However, upon termination of the Brazilian supply point agreement, all contracts and sublicenses executed with bottlers involving TCCC’s trademarks were to “vest and inure to the benefit of the Company.” TCCC effectively canceled the Brazilian supply point’s sublicensing authority in October 2007. See supra p. 46. - 158 - year without TCCC’s acquiescence. See supra p. 46. TCCC exercised its termination rights in actual practice: From 1986 through 2009 it closed (or shifted production away from) 18 supply points, but no supply point received any comp- ensation for this loss of production and income. Indeed, to the extent petitioner’s contracts with its affiliates actually ad- dressed “marketing intangibles,” those contracts are altogether hostile to its posi- tion. At least 29 ServCo agreements executed after 2004 included a new clause clarifying the ownership of assets generated by the ServCos’ marketing efforts and those of the third-party marketing professionals with whom the ServCos con- tracted. A typical version of the clause read as follows: ServCo acknowledges that it does not take entrepreneurial risk in developing marketing concepts because the marketing advice pro- vided by ServCo is within the strategic guidelines established by Ex- port for the brands. ServCo also acknowledges that any marketing concepts developed by third party vendors are the property of Export. Export executed with CCS, the Belgian ServCo, a “master service agree- ment” whereby CCS acted as an intermediary between Export and other ServCos operating in Europe. That agreement included a robust reservation clause con- cerning ownership of intangible assets generated by the local ServCos’ marketing efforts and by the Belgian R&D unit: - 159 - ServCo [CCS] * * * acknowledges that any marketing concepts dev- eloped by third party vendors or any affiliate of Coca-Cola that pro- vides services to ServCo * * * are the property of EXPORT. * * *. Any intangibles arising out of the research and development activities of ServCo are the property of EXPORT. These contractual terms cannot be reconciled with petitioner’s assertion that the ServCos’ consumer marketing generated intangible assets owned by the supply points. The agreements explicitly state that any marketing concepts developed by the ServCos and the third-party marketing professionals with whom they contract- ed “are the property of Export,” a domestic subsidiary of TCCC. These reserva- tion clauses reflect the Company’s consistent strategy for protection of its “crown jewels”--centralizing ownership of all intangible assets under the U.S. parent to ensure protection under U.S. law.47 2. Economic Substance For the reasons outlined above, the supply points were neither “legal own- er[s]” of marketing intangibles under relevant intellectual property law nor owners 47 Petitioner relies on IRS private letter rulings for the proposition that cer- tain supply points inherited goodwill and similar assets from their predecessors. But it cites no authority for the proposition that ordinary corporate goodwill should be treated as an “intangible asset” for purposes of analysis under sec. 1.482-4(b), Income Tax Regs. Cf. Canterbury v. Commissioner, 99 T.C. 223, 249 (1992) (finding that goodwill of McDonald’s franchise system “inheres in the McDonald’s trade name and trademarks” and that franchisee had no goodwill except through its franchise agreement). - 160 - of intangible rights “pursuant to contractual terms.” See sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra. Petitioner is thus forced to rely on the provi- so to this regulation, which states that legal or contractual ownership is dispositive “unless such ownership is inconsistent with the economic substance of the under- lying transactions.” Ibid. Petitioner’s reliance on this proviso is misplaced for two reasons. First, only the Commissioner, and not the taxpayer, may set aside contractual terms as inconsistent with economic substance. Second, even if peti- tioner could set aside the terms of its own contracts, it has failed to establish that the economic substance differs from the contractual form. a. Setting Aside Contract Terms In defining ownership of intangible property for section 482 purposes, the regulation cross-refers to the general rules for “identifying contractual terms.” See sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra (cross-referring to section 1.482-1(d)(3)(ii)(B), Income Tax Regs.). The latter regulation specifies rules applicable when there is a “written agreement” and when there is “no written agreement.” Sec. 1.482-1(d)(3)(ii)(B), Income Tax Regs. Contractual terms set forth in a written agreement antedating the transac- tions at issue “will be respected if such terms are consistent with the economic substance of the underlying transactions.” Id. subdiv. (ii)(B)(1). “If the contract- - 161 - ual terms are inconsistent with the economic substance,” the IRS “may disregard such terms and impute terms that are consistent with the economic substance.” Ibid. Similarly, in the absence of a written agreement, the IRS “may impute a contractual agreement between the controlled taxpayers consistent with the econ- omic substance of the transaction.” Id. subdiv. (ii)(B)(2).48 Notably absent from this regulation is any provision authorizing the taxpay- er to set aside its own contract terms or impute terms where no written agreement exists. That is not surprising: It is recurring principle of tax law that setting aside contract terms is not a two-way street. In a related-party setting such as this, the taxpayer has complete control over how contracts with its affiliates are drafted. There is thus rarely any justification for letting the taxpayer disavow contract terms it has freely chosen. But because the terms of such contracts may be self- 48 This regulation, which was promulgated in 1994, mentions “the district director” as the IRS official authorized to set aside or impute contract terms. Sec. 1.482-1(d)(3)(ii)(B), Income Tax Regs. The IRS eliminated the office of district director pursuant to the IRS Restructuring and Reform Act of 1998 (RRA), Pub. L. No. 105-206, 112 Stat. 685. The RRA includes a saving provision that “applies to keep in effect regulations that refer to officers whose positions no longer exist.” Grunsted v. Commissioner, 136 T.C. 455, 461 (2011). - 162 - serving and tax-motivated, the regulations regularly authorize the Commissioner to set contract terms aside if they do not reflect economic reality.49 This Court and others have repeatedly recognized that disregarding contract terms on the basis of economic substance is generally the prerogative of the Com- missioner, not of the taxpayer. “A taxpayer is free to adopt such organization for his affairs as he may choose and * * * [having done so] must accept the tax disad- vantages.” Higgins v. Smith, 308 U.S. 473, 477 (1940). “A taxpayer cannot elect a specific course of action and then when finding himself in an adverse situation extricate himself by applying the age-old theory of substance over form.” Legg v. Commissioner, 57 T.C. 164, 169 (1971), aff’d, 496 F.2d 1179 (9th Cir. 1974); accord Woodruff v. Commissioner, 131 F.2d 429, 430 (5th Cir. 1942) (holding that “the method pursued is determinative for tax purposes” even though different tax results would attach if an alternative procedure had been followed), aff’g 46 49 See, e.g., sec. 1.141-14(a), Income Tax Regs. (authorizing the Commis- sioner to “take any action to reflect the substance of the transaction”); id. sec. 1.446-3(g)(1) (providing that “[t]he Commissioner may recharacterize all or part of a [notional principal contract] transaction”); id. sec. 1.467-1(a)(5) (providing that “the substance-over-form doctrine * * * may be applied by the Commission- er”); id. sec. 1.672(f)-4(e) (providing that “the Commissioner may depart from the rules of this section and recharacterize * * * the transfer in accordance with its form or its economic substance:); id. sec. 1.988-2(f)(1) (providing that a nonfunc- tional currency transaction “may be recharacterized by the Commissioner in ac- cordance with its substance”). - 163 - B.T.A 727 (1942); Norwest Corp. v. Commissioner, 111 T.C. 105, 145 (1998) (“[W]hen a taxpayer seeks to disavow its own tax return treatment of a transaction by asserting the priority of substance only after the Commissioner raises questions with respect thereto, this Court need not entertain the taxpayer’s assertion of the priority of substance.”). Absent agreement to the contrary, appeal of this case would lie to the U.S. Court of Appeals for the Eleventh Circuit. See sec. 7482(b)(1)(B). That court has adopted the so-called Danielson rule. See Spector v. Commissioner, 641 F.2d 376, 384-386 (5th Cir. 1981) (adopting Commissioner v. Danielson, 378 F.2d 771 (3d Cir. 1967), vacating and remanding 44 T.C. 549 (1965)), rev’g 71 T.C. 1017 (1979)); see also Bonner v. City of Prichard, 661 F.2d 1206, 1209 (11th Cir. 1981) (adopting Fifth Circuit case law established before October 1, 1981). In Daniel- son, the Third Circuit held that “a party can challenge the tax consequences of his agreement as construed by the Commissioner only by adducing proof which in an action between the parties to the agreement would be admissible to alter that construction or to show its unenforceability because of mistake, undue influence, fraud, duress, et cetera.” Danielson, 378 F.2d at 775. The Eleventh Circuit recently reiterated its agreement with this principle. See Peterson v. Commis- - 164 - sioner, 827 F.3d 968, 987-988 (11th Cir. 2016) (quoting Plante v. Commissioner, 168 F.3d 1279, 1280-1281 (11th Cir. 1999)), aff’g T.C. Memo. 2013-271.50 The supply points’ agreements with TCCC endow them with no ownership rights in marketing intangibles, and the ServCos’ contracts with Export dictate that any rights to marketing intangibles “are the property of Export.” Under Danielson petitioner cannot disregard these contract terms unless it can show that they would be judicially unenforceable. Petitioner has not attempted to make such a showing. Petitioner appears to believe that Danielson and its progeny do not apply to determinations made under section 482. But it cites no Tax Court or appellate au- thority for that proposition. And the transfer pricing regulations provide absolute- ly no support for the notion that a taxpayer can set aside the unambiguous terms of a related-party agreement on the theory that they do not comport with economic substance. The regulations state that legal or contractual ownership is dispositive in determining who owns an intangible, but they authorize the Commissioner to impute terms consistent with economic substance when the taxpayer’s contract 50 The Danielson rule generally prevents taxpayers from disavowing the terms of their own contracts. It does not preclude a taxpayer from disputing the tax consequences that flow from a contract’s actual terms. See United States v. Fort, 638 F.3d 1334, 1338 (11th Cir. 2011). - 165 - does not reflect such terms. See sec. 1.482-1(d)(3)(ii)(B), Income Tax Regs. The regulations grant no reciprocal authority to taxpayers.51 As Treasury stated in 2006, the regulations embrace the principles that [C]ontrolled taxpayers have substantial freedom to adopt contractual terms, and * * * such contractual terms are given effect under section 482, provided they are in accord with the economic substance of the controlled parties’ dealings. An important corollary of these prin- ciples, however, applies where controlled parties fail to specify con- tractual terms, or specify terms that are not consistent with economic substance. In such cases, the Commissioner may impute contractual terms to accord with the economic substance of the controlled parties’ activities. * * * [T.D. 9278, 2006-2 C.B. at 269.] Here, TCCC exercised its freedom to adopt contract terms that furthered a central element of its corporate strategy--to centralize ownership of its “crown jewels” in the U.S. parent to ensure their protection under U.S. law. Petitioner ac- cordingly granted neither the supply points nor the ServCos any ownership rights 51 The only judicial decision petitioner cites for a contrary proposition was issued by a North Carolina bankruptcy court. See In re DeCoro USA, Ltd., 113 A.F.T.R. 2d 2014-1434 (Bankr. M.D.N.C. 2014). The IRS claim in that bankrupt- cy case turned on whether title to certain furniture had transferred to a related enti- ty and (if so) when. According to an inter-company agreement, title transferred at “the time risk of loss or damage so transfers.” The bankruptcy court concluded that title never in fact transferred because the inter-company agreement, when “[r]ead in its entirety,” indicated that risk of loss never transferred. The court rel- ied in part on testimony to corroborate that conclusion. The court did not cite Danielson or otherwise suggest that the Third Circuit’s opinion had been brought to its attention. And In re DeCoro had nothing to do with allocation of income from intangibles under section 482. To the extent the bankruptcy court’s opinion were thought to have any relevance here, we are unpersuaded by its analysis. - 166 - in its intangible assets, including what petitioner calls “marketing intangibles.” “[W]hile a taxpayer is free to organize his affairs as he chooses, nevertheless, once having done so, he must accept the tax consequences of his choice.” Commis- sioner v. Nat’l Alfalfa Dehydrating & Milling Co., 417 U.S. 134, 149 (1974).52 52 Whereas respondent urges that legal ownership of intangibles is disposi- tive here, petitioner asserts that the Commissioner took an inconsistent position in notices of proposed adjustments (NOPAs) issued to its Canadian service company (Canadian ServCo) in 2012. In those NOPAs the Commissioner proposed transfer pricing adjustments based in part on the theory that trademarks legally owned by the Canadian ServCo were owned in economic substance by TCCC. After consul- tations with the U.S. and Canadian competent authorities, the parties settled that dispute in 2016, shortly after the IRS issued the notice of deficiency in this case. We do not see how this line of argument helps petitioner. Because the sec- tion 482 regulations authorize the Commissioner--but not the taxpayer--to set aside contract terms as inconsistent with economic substance, the Commissioner’s position in the Canadian dispute was not necessarily inconsistent with his position here. In any event respondent is free, in order to avoid the risk of whipsaw, to plead inconsistent positions in separate disputes against related parties. See Mag- gie Mgmt. Co. v. Commissioner, 108 T.C. 430, 446-448 (1997) (holding that the IRS’ taking inconsistent positions was substantially justified to avoid whipsaw); Jacklin v. Commissioner, 79 T.C. 340, 344 (1982) (acknowledging that the IRS may take inconsistent positions regarding payments received by former spouses incident to divorce “in order to protect the revenue and to insure consistent treat- ment”). And because the parties settled the Canadian dispute before it got to court, petitioner does not contend that judicial estoppel could have any application here. Cf. Huddleston v. Commissioner, 100 T.C. 17, 28 (1993). - 167 - b. Consistency With Economic Substance Even if petitioner were permitted to disavow the terms of its contracts, it has not carried its burden of proving that it would be consistent with economic sub- stance to treat the supply points as owning “marketing intangibles” worth billions of dollars. According to petitioner, these intangible assets arose from the Serv- Cos’ expenditures for consumer marketing, which they invoiced to Export. The supply points had nothing to do with consumer marketing, but petitioner’s person- nel made inter-company charges transferring those costs to the supply points’ books. Petitioner’s first problem is its failure to explain how and why particular costs were allocated to particular supply points. The “fees and commissions” and DME allocated to the supply points varied widely, with no apparent relationship to their gross revenues. The Egyptian and Swazi supply points during 2007-2009 were allocated “fees and commissions” that averaged 40% of their gross revenue, whereas the Brazilian and Chilean supply points reported zero “fees and commis- sions.” The DME charged to the supply points during 2007-2009, as a percentage of their average gross revenues, likewise ranged widely, from 0.3% to 24.8%. See supra p. 75. A party urging that expenditures generated intangible assets has the - 168 - burden of showing how those assets came to exist. Petitioner has not met that burden.53 Second, petitioner has cited no authority for the proposition that spending money on consumer advertising, without more, gives rise to freestanding intangi- ble assets as a matter of economic substance. In calculating the supply points’ net operating profit, Dr. Newlon gave them full credit for the consumer marketing expenses that petitioner’s personnel placed on their books. But petitioner is con- tending that those expenses should not only be treated as offsets to revenue, but should also be capitalized--for transfer pricing purposes anyway--as intangible as- sets worth billions of dollars. If a debtor attempted to capitalize its ordinary ad- vertising expenses as intangible assets, we suspect a bank would be hesitant to lend against that security. Third, even if advertising expenses could properly be capitalized as intangi- ble assets, petitioner has failed to show that this treatment would comport with 53 Several of petitioner’s witnesses testified--at a very high level of general- ity--that TCCC employed “the matching principle” to assign to each supply point a share of consumer marketing costs corresponding to its concentrate revenues. On the basis of the evidence submitted at trial, this testimony struck the Court as ipse dixit. As explained in the text, the DME and “fees and commissions” charged to the seven supply points did not “match” their revenues in any discernible manner. The Court offered petitioners’ witnesses the opportunity to clarify the details of how these charges were determined, but they did not do so. - 169 - economic substance when the advertising supports somebody else’s product. TCCC owned virtually all the intangibles relating to manufacture and sale of its branded beverages. To the extent the ServCos’ advertising expenditures added value, they added value to the trademarks and brands that TCCC owned. These expenditures did not create new, freestanding intangible assets in the hands of the ServCos or the supply points. Petitioner’s own past practice shows that the supply points, in economic substance, did not own “marketing intangibles.” Petitioner’s agreements with the supply points were terminable (and frequently were terminated) by TCCC at will. For example, during 2007-2009 petitioner closed its Peruvian supply point, which had been producing concentrate for years, and transferred its production to other supply points in Latin America. During the many years of its existence, the Peru- vian supply point on petitioner’s theory would have amassed a treasure trove of “marketing intangibles.” What happened to those assets when the Peruvian affil- iate was terminated? The Court posed that question to petitioner’s experts, but they had no an- swer. The supply points that took over the Peruvian concentrate production paid nothing for these supposed intangibles when the Peruvian business was transferred to them. Any value traceable to the Peruvian supply point’s expenditures for con- - 170 - sumer advertising necessarily reverted to TCCC, the owner of the brands that the advertising supported. This shows that the Peruvian supply point never owned any “marketing intangibles” to begin with. The value derived from consumer ad- vertising for petitioner’s brands belonged to petitioner--and exclusively to it--all the while. Petitioner’s experts assert that the supply points’ actions were inexplicable unless they had, in economic substance, some long-term right to exploit TCCC’s intangibles. The supply points had no legal obligation to defray the cost of the ServCos’ consumer marketing. That obligation fell to Export, whose contracts with the ServCos required that it reimburse them. Petitioner’s experts assert that the supply points, acting at arm’s length, would never have acquiesced in being subjected to these charges unless they expected, as a quid pro quo, a long-term ownership interest in TCCC’s intangibles. That conclusion is said to follow from the principle that an arm’s-length actor will not work for the benefit of another un- less that conduct furthers its own interests. This argument ignores the supply points’ actual economic position. Assum- ing for the moment that the supply points dealt with petitioner at arm’s length, pet- itioner offered them a very attractive deal. It offered to vastly undercharge them for royalty expense--allowing them to use TCCC’s trademarks, secret formulas, - 171 - and proprietary manufacturing processes for a small fraction of their value--so long as the supply points agreed to pick up some consumer marketing expenses. After paying the consumer marketing expenses that petitioner allocated to them, the Irish, Brazilian, Chilean, and Costa Rican supply points enjoyed during 2007- 2009 average ROAs of 215%, 182%, 149%, and 143%, respectively--higher than any of the 996 international food and beverage companies in Dr. Newlon’s com- parison group. See supra p. 113. The Swazi and Mexican supply points enjoyed ROAs of 129% and 94%, respectively--higher than 99% of the companies in Dr. Newlon’s comparison group. Since petitioner permitted the supply points to enjoy astronomical levels of profitability, their agreement to the quid pro quo of absorb- ing marketing costs is perfectly explicable in economic terms: Because the for- gone royalty expense would vastly exceed the marketing costs, any rational econ- omic actor would have accepted this deal. And this conclusion holds true after respondent’s section 482 allocations are given effect. Under Dr. Newlon’s bottler CPM, the four Latin American supply points would have ROAs of 34.3%--higher than those enjoyed by 863 of the 996 food and beverage companies in Dr. Newlon’s comparison group. The Irish and Swazi supply points would have ROAs of 20.9%--higher than those enjoyed by 794 of the 996 companies in Dr. Newlon’s comparison group. - 172 - As petitioner’s experts conceded, the supply points’ manufacturing activity was a routine activity, benchmarkable to the activities of contract manufacturers and meriting no more than a cost-plus return. But after respondent’s reallocations are given effect, these contract manufacturers find themselves in the top quartile of food and beverage companies worldwide in terms of profitability. This level of profitability is more than sufficient to explain their willingness to absorb the con- sumer marketing costs petitioner allocated to them. B. Supposed “Long-Term Licenses” Petitioner puts an alternative spin on the preceding argument by asserting that the supply points’ contracts with petitioner endowed them with intangible as- sets in the form of “long-term licenses.” As explained supra pp. 115-118, we find no factual support for that argument. These agreements were terminable (and fre- quently were terminated) by petitioner at will. No supply point enjoyed any form of territorial exclusivity, and no supply point was granted any right, express or im- plied, to guaranteed production of concentrate. The impermanence of their rights was demonstrated in practice by petitioner’s closure of supply points and repeated shifts of production from one supply point to another. In short, the supply points were neither the “legal owner[s]” of long-term licenses nor “holder[s] of rights constituting an intangible [property] pursuant to - 173 - contractual terms * * * or other legal provision.” See sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra. The supply points enjoyed none of the privileges or protections that a genuine long-term licensee enjoys. Although petitioner asserts that their rights were “substantively exclusive and long-term,” petitioner cannot urge economic substance as a ground for setting aside the terms of its own contracts. See supra pp. 159-166. And even if petitioner could properly urge substance over form, the argument would not be persuasive. Petitioner contends that the supply points had “leverage in hypothetical arm’s-length negotiations” that gave them de facto long-term rights. But TCCC, not the supply points, had nearly all the bargaining power in the market for con- centrate production. The supply points were contract manufacturers that perform- ed routine functions; they competed with other supply points, most of which had unused production capacity. TCCC’s repeated shifts in concentrate production show that no supply point had any significant local advantage; TCCC shifted so much production to the Irish supply point that it ultimately sold concentrate to bottlers in more than 90 countries. In short, the supply points were easily replace- - 174 - able--and regularly were replaced--at relatively low or zero cost to petitioner. A party thus circumstanced would have little leverage in arm’s-length negotiations.54 Petitioner contends (and Dr. Newlon acknowledged) that “TCCC would not have had the capacity to supply its markets had it terminated the six * * * [supply points] suddenly during the years in issue.” But this does not mean that any sup- ply point individually had de facto long-term rights. The record reflects dozens of production shifts among supply points in the decades after 1980, but CPS effected these changes incrementally, with the precise goal of avoiding any disruption of concentrate supply to bottlers. Every supply point faced the perpetual risk that its production would be shifted, in whole or in part, to a competing supply point. On CPS’ recommenda- tion the Company in 2008 began construction of a new concentrate plant in Singa- pore, seeking to leverage free trade agreements and take advantage of tax and tariff incentives. The Singapore supply point ultimately supplied concentrate to 54 The supply points arguably might have increased bargaining power if they negotiated in concert (as petitioner appears to assume they could). But petitioner has not explained how the supply points, if independent economic actors, could legally have engaged in cartel-like behavior under the laws of their respective jur- isdictions. In any event, cartels are unstable and can be doomed by any partici- pant’s acting in its own self-interest. See, e.g., Volvo N. Am. Corp. v. Men’s Int’l Prof. Tennis Council, 857 F.2d 55, 67 (2d Cir. 1988) (discussing the inherent in- stability of cartels). - 175 - bottlers in 16 markets that had previously been served by 14 supply points in Asia and elsewhere. This is but one of many examples showing that no supply point, as a matter of “economic substance,” held de facto long-term license rights. The fact that CPS effected these production shifts prudently, in a manner that avoided dis- ruption of supply, does not support a contrary conclusion. C. Royalties Payable by Brazilian Supply Point 1. Ownership of Brazilian Trademarks Petitioner advances an additional argument regarding ownership of intangi- bles that is specific to the Brazilian supply point, which was formed in 1962. Like the other supply points, it defrayed consumer marketing expenses when those costs were charged to its books. Petitioner contends that the Brazilian supply point, by virtue of having defrayed such expenses, should be treated as the devel- oper (and thus as the owner) of 15 trademarks covering the Coca-Cola, Fanta, and Sprite brands that TCCC registered in Brazil before 1986. Petitioner urges that the Brazilian supply point was not required to pay any royalties for use of this intangi- ble property. This result is said to follow under the “developer-assister” rules of the 1968 regulations, 26 C.F.R. sec. 1.482-2(d) (1969), as applicable through a grandfather clause in the current regulations. See sec. 1.482-1(j)(4), Income Tax Regs. (pro- - 176 - viding generally that “[t]hese regulations will not apply with respect to transfers made or licenses granted to foreign persons before November 17, 1985”). Assum- ing without deciding that the “developer-assister” rules could have some applica- tion here, we reject petitioner’s argument. The 1968 regulations generally provide that the Commissioner may make appropriate allocations “where intangible property or an interest therein is trans- ferred * * * or otherwise made available in any manner by one member of a [con- trolled] group * * * to another member of the group * * * for other than an arm’s length consideration.” 26 C.F.R. sec. 1.482-2(d)(1)(i) (1969). The regulation then creates a special rule for the situation where (in the absence of a bona fide cost sharing arrangement) “one member of a group * * * undertakes the development of intangible property as a developer.” Id. subdiv. (ii)(a). In that event, “no al- location with respect to such development activity shall be made * * * until such time as any property developed, or any interest therein, is or is deemed to be transferred * * * or otherwise made available * * * by the developer to a related entity.” Ibid. The regulation then addresses the possibility that another group member may have “render[ed] assistance * * * to a developer in connection with an attempt to develop intangible property.” Id. subdiv. (ii)(b). If the assisting member is not - 177 - compensated at arm’s length for its assistance, the Commissioner may make an al- location or other adjustment. Ibid. Because the regulation addresses the section 482 consequences both for the developer and for the assister, these provisions are commonly called the “developer-assister” rules. Under these rules, “[t]he determination as to which member of a group * * * is a developer and which members of the group are rendering assistance * * * shall be based upon all the facts and circumstances of the individual case.” Id. subdiv. (ii)(c). “Of all the facts and circumstances to be taken into account in making this determination,” the regulation explains: [G]reatest weight shall be given to the relative amounts of all the direct and indirect costs of development and the corresponding risks of development borne by the various members of the group, and the relative values of the use of any intangible property of members of the group which is made available without adequate consideration for use in connection with the development activity, which property is likely to contribute to a substantial extent in the production of intan- gible property. * * * [Ibid.] For this purpose, “the risk to be borne with respect to the development activity is the possibility that such activity will not result in the production of intangible property or that the intangible property produced will not be of sufficient value to allow for the recovery of the costs of developing it.” Ibid. “A member will not be considered to have borne the costs and corresponding risks of development unless - 178 - such member is committed to bearing such costs in advance of, or contempora- neously with, their incurrence.” Ibid. For many reasons we conclude that the developer-assister rules are of no help to petitioner. First, these rules address the scenario where a group member undertakes to develop intangible property that does not currently exist. The regu- lation refers to “an attempt to develop intangible property,” in circumstances where there is a risk that the development activity “will not result in the produc- tion of intangible property.” Id. subdiv. (ii)(b) and (c). By way of example, the regulation cites a developer’s effort “to develop a new patentable product” or to “develop a new machine which will function effectively in the climate in which * * * [the developer’s] factory is located.” Id. subdiv. (ii)(a), (d), Example (1). Petitioner asserts that the Brazilian supply point, by virtue of defraying con- sumer marketing costs after 1962, should be deemed “the developer” of 15 trade- marks that TCCC registered in Brazil before November 17, 1985. But TCCC had fully developed and secured Brazilian registration for the nine most important marks--covering the product names and core design features for Coca-Cola, Coke, Fanta, and Sprite--between 1912 and 1962, before the Brazilian supply point exist- - 179 - ed. The Brazilian supply point could not possibly have “developed” or contrib- uted to the development of this preexisting intangible property.55 Between 1962 and November 17, 1985, TCCC registered an additional six trademarks in Brazil. Five related to Coca-Cola, covering the dynamic ribbon and the product names Coke Light, Coca-Cola Light, and Coke Classic. The sixth was a seemingly duplicative trademark for Sprite. Coke Classic was a new name for traditional Coca-Cola, a product that TCCC had developed and trademarked in Brazil before 1962. The dynamic ribbon was an element of the Spencerian script that TCCC had developed and trade- marked in Brazil before 1962. TCCC had registered a Brazilian trademark for Sprite in 1961. Coke Light and Coca-Cola Light were new products that relied heavily on preexisting intangible assets that TCCC owned. Petitioner has supplied no evidence that the Brazilian supply point acted in any sense as “the developer” of these six trademarks. In determining whether a group member was the developer, a key question is whether it bore the “risks of 55 Under the 1968 regulations the development period concluded when a group member “acquire[d] an interest in the property developed by virtue of obtaining a patent or copyright, or by any other means.” 26 C.F.R. sec. 1.482- 2(d)(1)(ii)(a) (1969). Here, TCCC acquired the nine original trademarks before the Brazilian supply point came into existence. As to those intangibles, there was no development period during which the developer-assister rules could apply. - 180 - development.” Id. subdiv. (ii)(c). For this purpose, “the risk to be borne * * * is the possibility that such activity will not result in the production of intangible property or that the intangible property produced will not be of sufficient value to allow for the recovery of the costs of developing it.” Ibid. Because TCCC almost certainly had registered trademarks for Coke Light, Coca-Cola Light, and Coke Classic in the United States before it registered them in Brazil, petitioner has not shown that the Brazilian supply point incurred any meaningful risk that these intangibles would not be developed (or would not be developed successfully). In determining which group member was the developer, another important question concerns “the relative values of the use of any intangible property” that is “made available without adequate consideration for use in connection with the development activity.” Ibid. Petitioner made obvious contributions of intangible property used in developing the Coke Classic, Coke Light, and Coca-Cola Light trademarks in Brazil. These contributions included the Coca-Cola secret formula, the flavorings and ingredients used in producing these brands, and new low- calorie sweeteners created in petitioner’s U.S. laboratories. Petitioner has supplied no facts to establish “the relative values of the use of * * * intangible property” contributed by the two controlled parties. Ibid. The only “intangible property” that the Brazilian supply point supposedly brought to - 181 - development of the pre-1986 Brazilian trademarks was funding consumer adver- tising in Brazil. Even if consumer advertising were thought to generate intangible property, petitioner has not established its value relative to the intangibles of ob- vious value that petitioner contributed.56 Under the 1968 regulations, the determination as to which group member is a developer and which members are rendering assistance “shall be based upon all the facts and circumstances of the individual case.” Ibid. The greatest weight is given to “the relative amounts of all the direct and indirect costs of development,” “the corresponding risks of development borne by the various members of the group,” and the relative value of intangible assets that group members contribute to the development activity. Ibid. Petitioner bears the burden of proof, and it has 56 Under the developer-assister rules a member “will not be considered to have borne the costs and corresponding risks of development unless such member is committed to bearing such costs in advance of, or contemporaneously with, their incurrence.” 26 C.F.R. sec. 1.482-2(d)(1)(ii)(c) (1969). Needless to say, the Bra- zilian supply point made no contemporaneous commitment to bear the costs of developing the nine Brazilian trademarks that TCCC registered before the Brazili- an supply point was created in 1962. Moreover, the agreements it executed with TCCC beginning in February 1963 did not require it to engage in any marketing activities or bear any consumer marketing costs. Thus, assuming arguendo that defraying advertising costs could be construed as costs of developing trademarks, the Brazilian supply point made no contemporaneous commitment to bear those costs either. - 182 - failed to offer any credible evidence that the Brazilian supply point should be deemed “the developer” of the pre-1986 trademarks under these standards. Petitioner errs in relying, in support of its argument, on DHL Corp. v. Commissioner, 285 F.3d 1210. Construing the developer-assister rules of the 1968 regulations, the Ninth Circuit concluded that DHLI, a foreign affiliate of DHL, should be treated either as the developer of DHL’s foreign trademarks or as rendering assistance to DHL as the developer of the foreign marks. Id. at 1224. DHLI was formed in Hong Kong “shortly after DHL began operations.” Id. at 1223. DHLI “register[ed] the ‘DHL’ trademark under DHLI’s name in various foreign countries,” “bore essentially all related costs,” and “had the exclusive right to use and sublicense” the DHL trademark overseas. Id. at 1214, 1223. “Con- versely, it was undisputed at trial that DHL bore none of the costs and risks in developing the foreign trademark rights.” Id. at 1223. The facts of the instant case are entirely different. The Brazilian supply point was created 76 years after the Company opened for business. During that 76-year period, TCCC incurred all the costs and risks of developing the Coca- Cola, Fanta, and Sprite brands that are covered by the trademarks in question. TCCC incurred the costs of registering the nine original trademarks in Brazil be- tween 1912 and 1962. And it incurred all the costs and risks of developing the - 183 - value of those marks through its branch operations in Brazil between 1945 and 1962. The Ninth Circuit suggested that it would have reached a different result on facts such as these: “[I]n the trademark context, if a company with a product already recognized in the target market incorporated a local subsidiary, the sub- sidiary’s expenditures might be presumed to be exploiting this trademark rather than developing its value.” Id. at 1222. That is precisely the situation here. By the time the Brazilian supply point was created in 1962, the Coca-Cola, Fanta, and Sprite brands were well recognized in the Brazilian market and the trademarks in Brazil had been fully developed. At the end of the day, petitioner’s Brazil-focused argument is really just a variation on its central theme. Its principal contention is that its inter-company allocation of consumer advertising costs to the supply points created “marketing intangibles” in their hands. Under the developer-assister rules it contends that its inter-company allocation of those same consumer advertising costs, to the Brazili- an supply point in particular, caused the latter to be the developer (and thus the owner) of intangibles (trademarks) that TCCC itself had registered and developed - 184 - in Brazil long before the Brazilian supply point was even created. In neither incarnation is this argument persuasive.57 2. Brazilian “Blocked Income” Petitioner alternatively contends that, if TCCC owned the Brazilian trade- marks, Brazilian law would have prevented the Brazilian supply point from pay- ing, for use of those trademarks, royalties anywhere close to the amounts deter- mined in the notice of deficiency. During 2007-2009 Brazilian law restricted the amount of trademark royalty and technology transfer payments that a Brazilian 57 If petitioner could succeed in establishing that the Brazilian supply point was “the developer” of one or more of the 15 pre-1986 Brazilian trademarks, peti- tioner would have several additional hurdles to overleap. TCCC would necessari- ly be treated as having “render[ed] assistance” to the Brazilian supply point in that endeavor, and petitioner has not supplied the facts necessary to establish “the rela- tive amounts” of development costs incurred by TCCC and the Brazilian supply point or “the relative values of the use of any intangible property” that each contri- buted. 26 C.F.R. sec. 1.482-2(d)(1)(ii)(b) and (c) (1969). Moreover, TCCC re- gistered at least 53 additional trademarks in Brazil after 1985, covering the Coca- Cola contour bottle shape, secondary design features for TCCC’s core products, advertising slogans, composites of existing trademark elements, and product names for numerous additional beverages, including Coke Zero, Diet Fanta, Dasani, Minute Maid, Powerade, Kuat, and other local Brazilian brands. TCCC was the legal owner of those 53 trademarks, and the Brazilian supply point would have to pay royalties for use of this intangible property. Finally, the IP for which the Brazilian supply point was required to pay compensation was not limited to TCCC’s trademarks, but also included (for example) TCCC’s secret formulas and proprietary manufacturing processes. Because we hold that petitioner has failed to prove that the Brazilian supply point was the developer of any of the pre-1986 trademarks, we need not address such royalty allocation questions. - 185 - entity could pay to a foreign parent. The parties have stipulated that those maxi- mum amounts were approximately $16 million for 2007, $19 million for 2008, and $21 million for 2009. Relying on what is commonly called the “blocked income” regulation, res- pondent contends that these Brazilian legal restrictions should be given no effect in determining the arm’s-length transfer price. See sec. 1.482-1(h)(2), Income Tax Regs. The regulation generally provides that foreign legal restrictions will be taken into account only if four conditions are met. See id. subdiv. (ii). Petitioner contends that this regulation does not apply here or that the necessary conditions were met. Alternatively, it contends that the blocked income regulation is invalid under the Administrative Procedure Act and/or Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984). As the parties have observed, the validity of section 1.482-1(h)(2), Income Tax Regs., has been challenged by the taxpayer in 3M Co. & Subs. v. Commis- sioner, T.C. Dkt. No. 5816-13 (filed Mar. 11, 2013). The Court has granted a motion to submit the 3M case for decision without trial under Rule 122, and the case is still pending. We will accordingly reserve ruling on the parties’ arguments concerning the blocked income regulation until an opinion in the 3M case has been issued. - 186 - D. Bottlers’ Ownership of Intangibles Even if petitioner could convince us that the supply points should be deemed to own valuable “marketing intangibles,” it has failed to establish the other half of its argument--namely, that the supply points for that reason were not comparable to the bottlers on which Dr. Newlon predicated his CPM. The evi- dence at trial convinced us that the bottlers, in this respect as in others, occupied a position superior to that of the supply points. The bottlers owned and controlled genuine intangible assets in the form of retail distribution networks, sales forces, and customer lists--each deriving from the bottlers’ relationships with tens of thousands of wholesale and retail custom- ers. The supply points held no comparable assets: They sold concentrate to bottlers as directed by CPS, and the identity of those bottlers changed regularly as CPS shifted concentrate production among supply points. Because the supply points had no real customers of their own, their “distribution network” consisted simply of packaging concentrate and shipping it to bottlers as CPS instructed. As compared with the supply points, the bottlers also had a much stronger claim to intangibles in the form of “long-term franchise rights.” The largest bot- tlers, which Dr. Newlon included in his sample, had ten-year contracts with peti- tioner. Although TCCC could terminate these contracts, the mutual dependence - 187 - between the Company and its bottlers ensured that bottler agreements were almost always renewed. For this reason most major bottlers, including CCE, Hellenic, and Coca-Cola FEMSA, assigned to their bottling contracts an indefinite useful life for accounting and financial statement purposes. The supply points had no comparable leverage over the Company and thus no plausible claim to “long-term franchises.” The bottlers’ contracts with TCCC also afforded them a high degree of ter- ritorial exclusivity that the supply points lacked. As Drs. Kusum Ailawadi and Paul Farris, two of respondent’s experts, noted: “Exclusive territories motivate bottlers to invest in the development of production and distribution capacity as well as in retailer development, even if these investments require time and effort to become profitable.” Such investments by bottlers inevitably helped to promote and develop the Company’s brands. By contrast, the supply points’ lack of a consistent territorial market precluded them from making targeted investments or trusting that they could capture the benefits of investment before TCCC shifted production to another supply point. And if the supply points were deemed to have “marketing intangibles” be- cause they funded consumer marketing, the bottlers would also have “marketing intangibles” because they not only funded, but actually carried out, trade market- - 188 - ing. During the tax years at issue the System expended billions of dollars annually for marketing, split about evenly between the Company and its bottlers. TCCC and its bottlers implemented an informal “true up” strategy to ensure that overall marketing costs were split roughly 50-50 between them. The evidence at trial showed that bottlers, in the aggregate, paid about as much for trade marketing annually as TCCC and its affiliates paid for consumer marketing. During 2008, for example, CCE had “marketing deductions from revenue” of $2.5 billion, an amount equal to 11.5% of its net revenue. Other bottlers showed marketing deductions as high as 18% of their net revenue. And apart from incurring costs explicitly classified as “marketing,” the bottlers’ sales personnel engaged in other activities, such as “category management,” that enhanced the value of TCCC’s brands in consumers’ minds.58 Petitioner’s experts opined that the Company’s consumer marketing, con- sisting mostly of television ads and social media, was intrinsically more valuable than the bottlers’ trade marketing and was more likely to enhance the value of TCCC’s brands. Respondent’s experts disputed that proposition, explaining that 58 As Drs. Ailawadi and Farris explained, too much inventory leads to stale products while too little inventory leads to disappointed consumers and an oppor- tunity for a competitor to steal a loyal customer. Ensuring that retailers get the balance right--known as “category management”--was an important activity under- taken by bottlers’ sales personnel. - 189 - the two forms of marketing are essential and mutually reinforcing. We found the latter testimony more persuasive. As Dr. Ailawadi and Prof. Farris observed, “If advertising’s role is sometimes overestimated, our opinion is that the difficulty of building an effective and efficient distribution system is often underappreciated.” The bottlers’ trade marketing activities were extensive and varied. Bottlers regularly engaged in promotions to encourage retailers to give TCCC’s products optimal shelf space. Bottlers were responsible for arranging point-of-sale promo- tions (such as floor decals and end-of-aisle displays), and offering in-store samples of new products. Bottlers managed most price promotions (including coupons, product discounts, and digital redemption codes), which often keyed off holidays and sporting events. Bottlers often integrated these retail promotions with the Company’s global sponsorship campaigns and consumer marketing themes--a good example of how consumer marketing and trade marketing were “mutually reinforcing.” In Europe, where third-party distributors delivered most beverages to retailers, bottlers sent merchandisers into stores to assure proper product place- ment and point-of-sale displays. To encourage impulse purchases--which provided much higher margins than purchases for future consumption--bottlers invested in coolers that were stra- tegically placed in retail outlets. These investments were significant: At one - 190 - point, coolers represented about one-third of CCE’s annual capital expenditures. Bottlers purchased coolers and negotiated with retailers for the requisite floor space--ideally near the checkout counter. By securing exclusive rights to stock the coolers with the Company’s products, bottlers perpetuated TCCC’s dominance in impulse sales and simultaneously advertised the Company brands. TCCC’s consumer marketing aimed to keep the Company’s products “top of mind” for consumers, but we find that the bottlers’ trade marketing achieved simi- lar benefits. Hundreds of thousands of coolers reminded customers of Coca-Cola as they approached checkout counters in retail stores. Tens of thousands of deli- very trucks emblazoned with the Coca-Cola logo threaded their way through cities, towns, and villages worldwide. Bottler representatives, clad in Coca-Cola uniforms, acted as mobile advertisers for the Company’s brands. For restaurants and mom-and-pop retailers, bottlers promoted Coca-Cola products by supplying Coca-Cola-branded awnings, Coca-Cola branded napkins, and plasma TVs that constantly advertised the Company’s beverages. All in all, we find that the bot- tlers, as compared with the supply points, owned far more intangible assets and invested at least as much in marketing that directly benefited petitioner’s brands. - 191 - E. Proposed Alternative Transfer Pricing Methodologies Petitioner submitted transfer pricing reports from three experts: Dr. Unni, Dr. Cragg, and Keith Reams. Each admits that TCCC owned the trademarks, secret formulas, and proprietary manufacturing processes needed to produce the Company’s beverages abroad. But each concludes that the supply points, at arm’s length, would be entitled to receive the vast bulk of the income that the Company derives from foreign markets. We found none of their analyses persuasive. 1. Proposed CUT Method Dr. Unni opines that “the CUT method represents the best method for determining TCCC’s arm’s length income.” He derives his supposed CUT from “master franchising transactions” that companies like McDonald’s and Domino’s Pizza execute with regional franchisees abroad. The regional franchisees typically operate some fast-food facilities themselves and execute subfranchise agreements with numerous owners of individual restaurants. Dr. Unni in fact located only one actual master franchising agreement for the years at issue. However, through an extremely complex series of calculations and assumptions, he purports to extract from the master franchise transactions a royalty rate payable to the franchisor for use of its IP. When the dust settles, he concludes that the master franchisees paid McDonald’s and Domino’s a median - 192 - average royalty equal to 2.2% of gross retail sales to consumers. He applies that 2.2% rate to retail revenues from sale of the Company’s products in the relevant foreign markets, then works back to compute an average royalty rate of 12.3% payable by the supply points. He concludes, in other words, that the supply points at arm’s length would be entitled to keep 87.7% of the Company’s total revenues from the markets the supply points served. Dr. Unni begins with the premise that “[t]he Foreign Licensees [i.e., supply points] are responsible for the Company’s foreign businesses, including managing and overseeing the franchise bottlers,” and that the supply points “are responsible * * * for consumer marketing activities and expenditures to exploit and develop * * * [TCCC’s] intangibles.” This premise is incorrect. The supply points did not manage or oversee bottlers and were not responsible for creating or implementing consumer marketing. All of those responsibilities were discharged by TCCC and the ServCos. Dr. Unni then adopts a sleight of hand by conflating the supply points and the ServCos into a concept called “the Field.” By “the Field” he appears to mean an aggregation of the functions performed and workers employed by all of TCCC’s foreign affiliates. He analogizes “the Field” to a “master franchisee” that licenses TCCC’s intangibles, then assumes responsibility for “maintain[ing] and - 193 - develop[ing] the value of the intangibles” and “managing the franchise in its territory.” To complete this imaginary universe he turns to the bottlers, casting them in the role of “subfranchisees” who “engage in the foregoing relationship with the Field, and bear responsibility for operating the franchises in the local market and delivering finished product to customers.” There are so many flaws in Dr. Unni’s construct that it is difficult to know where to begin. But to start with the basics: The regulations require that income be properly allocated among “controlled taxpayer[s].” Sec. 1.482-1(b)(1), Income Tax Regs. There are three sets of “controlled taxpayer[s]” here: petitioner, the supply points, and the ServCos. Putting “split invoicing” aside, the parties agree that the ServCos transacted with petitioner at arm’s length. We accordingly must determine whether the Commissioner abused his discretion in reallocating income to petitioner from the supply points. See supra p. 100. In seeking to calculate an arm’s-length royalty payable to petitioner by “the Field,” Dr. Unni has posited a controlled taxpayer that does not exist. Nor can the supply points--the relevant “controlled taxpayers”--plausibly be analogized to “master franchisees.” Dr. Unni’s master franchisees appear to have had long-term agreements (ranging from 10 to 50 years) that endowed them with exclusive rights within a specified territory. The supply points had short-term - 194 - contracts that petitioner could (and often did) terminate at will. And they enjoyed no territorial exclusivity whatsoever. The supply points manufactured concentrate. They played no role in “man- aging the franchise,” in “selecting subfranchisees,” or in overseeing bottlers in any geographic territory. Those responsibilities were discharged exclusively by TCCC and the ServCos. Nor can the bottlers be analogized to “subfranchisees” of the supply points. The bottlers’ contracts invariably ran with TCCC, not with the sup- ply points. The bottlers received direction and marketing assistance from the ServCos, not from the supply points. And as manufacturers and distributors the bottlers cannot be analogized to owners of restaurants that serve consumers. The regulations provide that the CUT method has an especially high degree of reliability only “[i]f an uncontrolled transaction involves the transfer of the same intangible under the same, or substantially the same, circumstances as the controlled transaction.” Id. sec. 1.482-4(c)(2)(ii) (emphasis added). Neither Dr. Unni nor petitioner identified any pricing data for transactions with unrelated par- ties that “involve[] the transfer of the same intangible”--viz., the trademarks, brand names, logos, secret formulas, and proprietary manufacturing processes used to produce Coca-Cola, Fanta, Sprite, and the Company’s other branded beverage products. - 195 - Instead, Dr. Unni derived his CUT using data from an entirely different industry--the fast-food restaurant business. But “[i]n order for intangibles in- volved in controlled and uncontrolled transactions to be comparable, both intangi- bles must be ‘used in connection with similar products or processes within the same general industry or market.’” Amazon.com, Inc., 148 T.C. at 164 (quoting section 1.482-4(c)(2)(iii)(B)(1)(i), Income Tax Regs.). Concentrate and ham- burgers are not “similar products,” and beverage manufacturers and fast food restaurants are not “in the same general industry.” Beverage manufacturers are classified in SIC code 20, Food and Kindred Products, whereas fast food restaur- ants are classified in SIC code 58, Eating and Drinking Places. Thus, the Com- pany and McDonald’s are not even in the same two-digit “Major Group” SIC code, much less in the same four-digit SIC code that researchers commonly em- ploy when searching for comparables. Finally, manufacturing and selling concen- trate to bottlers at wholesale does not involve “the same general market” as pre- paring and selling meals to consumers at retail. The degree of comparability between controlled and uncontrolled transac- tions also “requires a comparison of the significant contractual terms that could affect the results of the two transactions.” Sec. 1.482-1(d)(3)(ii), Income Tax Regs.; see id. sec. 1.482-4(c)(2)(iii)(B)(2); see also Medtronic, Inc. & Consol. - 196 - Subs. v. Commissioner, 900 F.3d 610, 614-615 (8th Cir. 2018) (emphasizing the necessity of analyzing comparability of contract terms), vacating and remanding T.C. Memo. 2016-112. At no point in his analysis did Dr. Unni compare the respective contractual terms under which the supply points and his master franchisees operated. Indeed, he did not have actual master franchise agreements for four of the five transactions he analyzed for the years at issue. Even if Dr. Unni’s CUT theory were supportable in theory, we found it defi- cient in implementation. Many of his master franchisees operated restaurants themselves as well as subfranchising to local restaurant operators. The activities of both commonly included managing real estate as well as serving food. In an effort to isolate a royalty stream payable to the franchisor solely for use of its intangibles, Dr. Unni had to make dozens of assumptions, estimates, adjustments, and cost reallocations involving (for example) G&A expenses, headquarters expense, rental income and expense, real estate expenses, “commissary profits” (derived from supplying key food ingredients), imputation of income from adver- tising, and collateral transactions (e.g., upfront payments to acquire franchise rights). His estimates were aggressive, his assumptions almost invariably favored petitioner, and many of his adjustments were shown by respondent’s experts to be mathematically and economically unsound. Respondent not unreasonably de- - 197 - scribes Dr. Unni’s model as a “Rube Goldberg machine” seemingly programmed to generate a result that petitioner would like. See Whitehouse Hotel Ltd. P’ship v. Commissioner, 139 T.C. 304, 322-323 (2012) (discounting testimony of expert witness who made “hundreds of assumptions” where “relatively minor changes in only a few of his assumptions would have large bottom-line effects”), aff’d in part, vacated in part on other grounds, and remanded, 755 F.3d 236 (5th Cir. 2014). 2. Proposed “Residual Profit Split Method” Dr. Cragg proposes that the arm’s-length royalty payable to TCCC be deter- mined using the “residual profit split method” (RPSM), a version of the “profit split” method authorized by the regulations. See sec. 1.482-6(a), (c)(3), Income Tax Regs. He concludes that the supply points, at arm’s length, would pay TCCC a weighted average royalty rate of 5.4% for 2007, 4.9% for 2008, and 4.6% for 2009. “In comparison,” he notes, “the royalty rate paid by the [supply points] to TCCC was between 10.2% and 11.2%” during 2007-2009. On his theory, there- fore, the appropriate transfer pricing outcome here would be a reallocation of bil- lions of dollars of income to the supply points from petitioner. Like Dr. Unni, Dr. Cragg views his task as allocating income between TCCC’s headquarters in Atlanta and “the Field.” As he explains: “I refer to the Foreign Licensees [i.e., supply points] and the related-party entities (Service - 198 - Companies and Confidential Ingredient Plants) collectively as the ‘Field.’” “From an economic perspective,” he says, “these entities together encompass the counter- party to TCCC in the intercompany licensing transactions that I analyze.” Dr. Cragg’s “Field” construct seems designed to solve several problems that he faced. First, the supply points--the relevant “controlled parties” for section 482 purposes--did not actually perform the economic functions that Dr. Cragg regards as valuable, e.g., implementing consumer advertising and engaging in “franchise leadership” with bottlers. In his view, “these consumer-facing aspects of the Field’s activities [are what] create valuable IP.” But all of these functions were performed by the ServCos, not by the supply points. As the premise for his RPSM, therefore, Dr. Cragg needed to invent a counterparty that would incorpor- ate all the valuable inputs that the ServCos contributed. Second, Dr. Cragg’s methodology posits that residual profit should be split between TCCC and its foreign affiliates on the basis of their respective expendi- tures for consumer marketing, over a period of 70-plus years, in the foreign mar- kets where the Company’s products were sold. But Dr. Cragg did not have relia- ble historical data for all of these markets. More problematically, no supply point had a consistent market of its own, because CPS repeatedly shifted concentrate - 199 - supply for particular markets from one supply point to another. By creating “the Field” as the counterparty to TCCC, Dr. Cragg elided these problems.59 To calculate the residual profit to be split between TCCC and “the Field,” Dr. Cragg first determined the “routine profit” to be assigned to the supply points for their manufacturing activity. Likening their activity to that of contract manu- facturers, he concluded that the supply points were entitled to “a constant 8.5% return * * * for the concentrate manufacturing function.” He subtracted this rou- tine profit from their total operating profit to yield the “residual profit” to be split with petitioner. Dr. Cragg opined that this residual profit should be split between TCCC and “the Field” on the basis of historical spending for consumer advertising. He ac- knowledged that TCCC owned the secret formulas, the proprietary manufacturing processes, and virtually all the foreign trademarks for its branded products. He 59 Dr. Cragg’s “Field” construct also papered over a third problem, which was created by “split invoicing.” Where this occurred, a portion of the ServCos’ consumer marketing expenses were paid directly by the bottlers and not charged by Export to the supply points. During 2007-2009 the bottlers directly paid more than $1 billion of ServCo expenses, see supra p. 65, and it seems likely that simi- larly large amounts of ServCo marketing expenses were paid by bottlers in prior years where split invoicing was used. To the extent this happened, the supply points did not incur the consumer marketing expenditures that form the center- piece for Dr. Cragg’s analysis. By amalgamating the ServCos and the supply points into “the Field,” Dr. Cragg also elided this problem. - 200 - agreed that TCCC had been selling Coca-Cola, Fanta, and Sprite in foreign mar- kets for many years before the supply points were incorporated. And he agreed that TCCC, directly and through its foreign branches, had spent billions of dollars on foreign consumer advertising before such costs began to be assigned to the supply points. But he opined that “the stock of consumer awareness in each country created by TCCC depreciated and was replaced by new investments by the Foreign Licensees [i.e., supply points] in existing and new products.” Dr. Cragg describes these supposed investments by the supply points as “marketing-related IP,” “IP associated with trademarks,” or “intangible develop- ment costs” (IDCs). But in each case he is simply referring to the dollar amount of consumer advertising expenses (plus certain other costs) that were incurred by the ServCos. He defines IDCs to include DME, other marketing expenses, certain reclassified deductions, and a portion of headquarters expense. Dr. Cragg opines that the IDCs incurred in foreign markets by TCCC and its affiliates over the years should be amortized at a 10% rate, giving those invest- ments a half-life of 6.6 years. After applying amortization, he finds that petition- er’s “decades-old legacy IDCs in these markets are small,” so that “TCCC’s in- vestment is unlikely to earn a large share of the profits between 2007 and 2009.” - 201 - And indeed that is how his math works out: He assigns to the supply points 94.6% of the residual profit in 2007, 95.1% in 2008, and 95.4% in 2009. As these figures show, the logic of Dr. Cragg’s model is that TCCC’s share of the residual profit drops with every passing year, as new advertising expenses are defrayed by the supply points and TCCC’s “legacy investments” are amortized out of existence. Conversely, the longer a supply point has been defraying adver- tising expenses, the lower its royalty rate. Indeed, because the Mexican supply point (the oldest of the group) started incurring advertising expenses in 1950, Dr. Cragg’s machine generates a result under which (a) the royalty obligation of the Mexican supply point is zero during 2007-2009, and (b) petitioner must pay the Mexican supply point an annual royalty of 0.2% to 0.3%. We perceive many deficiencies in Dr. Cragg’s analysis. First, like Dr. Unni, he errs in hypothesizing “the Field” as the counterparty to TCCC for purposes of transfer pricing analysis. In so doing he ignores the legal and taxable entities actually involved as well as the requirements of the section 482 regulations. The regulations require that income be properly allocated among “controlled taxpay- er[s].” Sec. 1.482-1(b)(1), Income Tax Regs. The “controlled taxpayer” relevant here are the supply points, not an amalgamation of the supply points and the Serv- Cos. See supra p. 99. - 202 - Second, the premise for Dr. Cragg’s RPSM is that the supply points owned intangible assets that he variously describes as “marketing-related IP,” “IP associ- ated with trademarks,” or “intangible development costs.” But the supply points were neither “[t]he legal owner[s] * * * of [intangible property] pursuant to the intellectual property law of the relevant jurisdiction” nor “holder[s] of rights con- stituting * * * [intangible property] pursuant to contractual terms (such as the terms of a license) or other legal provision.” Sec. 1.482-4T(f)(3)(i)(A), Temporary Income Tax Regs., supra. Dr. Cragg has ginned up supposed IP by capitalizing ordinary advertising costs. But while he may believe that doing so comports with “economic substance,” petitioner may not use this theory to disavow the terms of its own contracts. See supra pp. 159-166. Third, even if the supply points were deemed to own intangible property, Dr. Cragg did not determine “the relative value of nonroutine intangible property contributed” by TCCC and the supply points. See sec. 1.482-6T(c)(3)(i)(B)(2), Temporary Income Tax Regs., 71 Fed. Reg. 44487 (Aug. 4, 2006).60 He bases his 60 This temporary regulation applied for taxable years beginning after December 31, 2006. See sec. 1.482-6T(d)(1), Temporary Income Tax Regs., 71 Fed. Reg. 44487 (Aug. 4, 2006). The temporary regulation was replaced by a final regulation with nearly identical text. See sec. 1.482-6(c)(3)(i)(B)(2), Income Tax Regs. The final regulation applies for taxable years beginning after July 31, 2009. See id. para. (d)(1). - 203 - RPSM solely on the parties’ relative spending over the years on consumer adver- tising. Wholly apart from past advertising expenses, however, TCCC obviously brought to the table many other valuable intangibles--its brands, trademarks, trade- names, patents, logos, secret formulas, and proprietary manufacturing processes. Consumer advertising is worth little unless the seller has a product that people wish to buy. Over the course of 120 years, TCCC had invested billions of dollars, in its laboratories and elsewhere, developing and modifying secret formulas, flavorings, ingredients, sweeteners, manufacturing protocols, quality assurance techniques, and everything else that makes TCCC’s beverages the most popular brands in the world. By focusing solely on historic advertising expenditures, Dr. Cragg’s RPSM puts nothing in TCCC’s side of the ledger to reflect these im- mensely valuable intangible assets.61 Fourth, Dr. Cragg’s RPSM would have a low level of reliability under the regulations even if the foregoing defects did not exist. The RPSM regulations provide that the value of nonroutine intangibles may be determined in two ways. 61 The costs that Dr. Cragg included when calculating IDCs consist almost exclusively of past spending for consumer marketing by TCCC and the supply points in foreign markets. Apart from ignoring TCCC’s other intangible contribu- tions, therefore, he also ignores the massive amounts spent by TCCC in Atlanta on global marketing campaigns, corporate sponsorships, and proprietary marketing tools and databases designed to help ServCos and bottlers maximize sales of the Company’s products abroad. See supra pp. 30-37. - 204 - Their value may be “measured by external market benchmarks that reflect the fair market value of such intangible property.” Sec. 1.482-6T(c)(3)(i)(B)(2), Tempor- ary Income Tax Regs., supra. Alternatively, their value “may be estimated by the capitalized cost of developing the intangible property * * * less an appropriate amount of amortization based on the useful life of each intangible [property].” Ibid. Dr. Cragg employed the second technique--an estimate based on capitalized advertising costs less amortization--and his results are therefore less reliable, es- pecially on the facts here. The soundness of an RPSM depends on the “reliability of the data used and the assumptions made in valuing the intangible property con- tributed by the participants.” Sec. 1.482-6(c)(3)(ii)(C)(3), Income Tax Regs. “In particular, if capitalized costs of development are used to estimate the value of in- tangible property, the reliability of the results is reduced relative to the reliability of other methods that do not require such an estimate.” Ibid.; see id. subpara. (3)(ii)(D) (“[T]o 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.”). That is true here for at least two reasons. Dr. Cragg’s capitalization of IDCs has decreased reliability because it “require[s] assumptions regarding the useful life of the intangible property.” Id. subpara. (3)(ii)(C)(3). As respondent’s experts - 205 - explained, there is no consensus among economists that ordinary advertising costs can properly be capitalized as an intangible asset, much less about what the useful life of such an asset would be.62 More importantly, capitalization of IDCs may yield unreliable results be- cause “the costs of developing the intangible may not be related to its market value.” Ibid. That is plainly the case here. The “intangible” posited by Dr. Cragg is a capitalization of historical costs of advertising Coca-Cola products. No unre- lated 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. Dr. Cragg himself describes these intangibles as “IP associated with trademarks,” and this IP could have no value except when used in conjunction with the trademarks that TCCC owns. Because the intangibles that Dr. Cragg imagines the supply points’ bringing to the table have no discernible market value, his RPSM results are wholly unreliable. 62 There is also no consensus about whether such expenditures should be amortized or “decayed” at a uniform rate over time. By adopting a constant 10% rate for all advertising expenditures, Dr. Cragg afforded TCCC no premium for IDCs incurred during the earlier and riskier stages of foreign market development. - 206 - Finally, we regard Dr. Cragg’s methodology as unreliable because it pro- duces absurd results. His RPSM splits profits between TCCC and the supply points solely on the basis of consumer marketing spending in foreign markets, with TCCC’s older expenditures being gradually amortized out of existence. So long as petitioner continues to charge all foreign consumer advertising costs to the supply points, their share of the “marketing-related intangibles” will eventually approach 100% and TCCC’s share will approach 0%. Indeed, his RPSM produces that outcome for the Mexican supply point during 2007-2009, requiring that peti- tioner pay it a royalty. Under Dr. Cragg’s model, the supply points would eventu- ally be entitled to use all of TCCC’s intangibles--its brands, trademarks, logos, patents, secret formulas, and proprietary manufacturing processes--for free. That outcome cannot possibly be right, because it is obvious that these intangibles would be worth tens of billions of dollars in the open market.63 3. Proposed “Unspecified Method” Mr. Reams proposes that the arm’s-length royalty payable to TCCC by the supply points should be calculated using an “asset management model” often 63 The Company’s market capitalization in November 2020 exceeded $230 billion. Coca-Cola Company Market Capitalization, KOYFIN (price indicated as of Nov. 12, 2020), https://www.koyfin.com/charts/g/KO?view=chart (last visited Nov. 12, 2020). - 207 - employed “to compensate asset managers in the financial services sector.” Like Drs. Unni and Cragg, he views “the Field” as the counterparty to “TCCC HQ,” by which he means the Company’s headquarters in Atlanta. He posits that the Field “drive[s] the success and profitability of the foreign business,” whereas HQ “pri- marily focuses on governance, sharing of best practices, and high-level strategy.” Mr. Reams analogizes HQ’s activities “to those of a skilled asset manager manag- ing the assets of the Field.” Mr. Reams observes that hedge fund managers are often compensated for their services under a two-tiered fee structure: a base fee computed on “assets under management” and a profit fee computed on annual “net asset appreciation.” He concludes that a 2% base fee and a 20% profit fee would be appropriate for TCCC. Making numerous assumptions and an extremely complex series of calculations, he derives estimates for TCCC’s “assets under management” and annual “net asset appreciation” for 2007, 2008, and 2009. Converting these fees for services into royalties payable by the supply points, Mr. Reams comes up with a weighted average annual royalty rate of 9.3%. Mr. Reams acknowledges that “a services-pricing model * * * would not normally be used to price an intangibles license,” and that is surely an understate- ment. Mr. Reams’ asset management model does not remotely resemble any of the - 208 - “specified methods” for valuing intangibles under the section 482 regulations. In- deed, he describes his assignment as developing a pricing method “without the constraint of specific transfer pricing regulations.” Mr. Reams proposes to compensate TCCC only for asset management, e.g., for services involving “governance, sharing of best practices, and high-level stra- tegy.” His methodology thus bypasses the contributions made by TCCC that are relevant here--the brands, trademarks, tradenames, logos, patents, secret formulas, and proprietary manufacturing processes needed to produce the Company’s bever- ages abroad. Hedge fund managers typically do not supply such intangibles to the portfolio companies they manage. By compensating TCCC only for services, Mr. Reams’ model ignores the intangibles that are central to this case. VII. Collateral Adjustments Having upheld respondent’s primary allocations, we turn to the question of “collateral adjustments.” See sec. 1.482-1(g), Income Tax Regs. Collateral adjustments with respect to allocations of income under section 482 “may include correlative allocations, conforming adjustments, and setoffs.” Id. subpara. (1). We address two subjects under this heading: (1) respondent’s recomputation of petitioner’s section 987 loss as a consequence of reallocating income from the Mexican supply point to petitioner and (2) petitioner’s request that respondent’s - 209 - primary allocations be reduced to reflect dividends paid by the supply points, to the extent those amounts were repatriated to satisfy their royalty obligations. A. Recomputation of Section 987 Loss The Code requires that the determination of taxable income “shall be made in the taxpayer’s functional currency.” Sec. 985(a). Petitioner’s functional currency is the U.S. dollar. See sec. 985(b)(1). To satisfy section 985, petitioner must, for each taxable year, translate into dollars the income it has earned in a foreign currency. This process can be complicated by exchange rate volatility be- tween the time the foreign income is earned and the time it is remitted to the U.S. home office. Section 987, titled “Branch transactions,” provides rules for determining the taxable income of a taxpayer that has one or more “qualified business units” (QBUs) with a functional currency other than the U.S. dollar. It provides that the taxpayer’s income in that event shall be determined: (1) “by computing the taxa- ble income or loss separately for each such unit in its functional currency,” (2) “by translating the income or loss [as thus] separately computed * * * at the appro- priate exchange rate,” and (3) “by making proper adjustments (as prescribed by the Secretary) for transfers of property” between the QBU and the taxpayer. The tax- payer determines its foreign exchange (ForEx) gain or loss when the QBU remits - 210 - cash or property to it, e.g., by paying dividends or royalties. See sec. 1.987-5(d), Income Tax Regs.64 Petitioner’s Mexican supply point was a branch of Export, a domestic corp- oration, and its functional currency was the Mexican peso. Petitioner does not dis- pute that its Mexican branch was a QBU under section 987. The Mexican branch remitted cash or property to petitioner during 2007-2009, a period of financial crisis during which the peso was volatile against the dollar. For each year peti- tioner reported a section 987 loss with respect to its Mexican branch. In the notice of deficiency the IRS reallocated income to petitioner from the Mexican supply point as follows: Year Reallocation 2007 $155,205,260 2008 180,189,734 2009 160,335,364 64 The U.S. owner of a QBU must track (in sec. 987 “pools”) the U.S. dollar equivalent of the profit and loss earned by the QBU for each year at the historical exchange rates at which the items accrued. See sec. 1.987-5(d), Income Tax Regs. The unrecognized ForEx gain or loss carried in the sec. 987 pools is recognized under sec. 987 when the QBU “makes a remittance” to its U.S. home office. Id. para. (b). The ForEx gain or loss recognized under sec. 987 varies with the dif- ference between the exchange rate from the year of the remittance and the histor- ical exchange rates of the undistributed earnings. See generally Stanley Langbein, Federal Income Taxation of Banks and Financial Institutions, para. 12.15 (WG&L 2019). - 211 - The regulations provide that, when the IRS “makes an allocation under section 482 (referred to * * * as the primary allocation), appropriate correlative allocations will also be made with respect to any other member of the group affected by the allocation.” Sec. 1.482-1(g)(2)(i), Income Tax Regs. Thus, when the IRS makes a primary allocation, it “will not only increase the income of one member of the group, but correspondingly decrease the income of the other member.” Ibid. Consistently with this regulation, the IRS in the notice of deficiency made correlative allocations that reduced the income of the Mexican branch in the amounts shown above. These reductions to the branch’s income, originally recorded in pesos, caused the section 987 losses reported on petitioner’s 2007- 2009 returns to be incorrect. The IRS therefore recomputed petitioner’s section 987 losses to reflect the collateral allocations to the Mexican branch. The notice of deficiency accordingly determined that petitioner’s reported section 987 losses should be increased for 2007 and 2008, and should be decreased for 2009, in the following amounts: Adjustment to Year sec. 987 loss 2007 ($1,638,165) 2008 (1,756,566) 2009 5,402,536 - 212 - Petitioner does not dispute that the section 987 losses reported on its returns would be incorrect if the Mexican branch had actually recorded, during 2007- 2009, the reduced amounts of income that the IRS determined in its correlative allocations. And petitioner does not disagree with respondent’s exchange rate figures or otherwise dispute its math. But petitioner advances several arguments to support its view that respondent’s section 987 recomputations were impermis- sible. Petitioner’s first argument keys off the fact that the Mexican branch’s in- come was reported on petitioner’s U.S. consolidated return, so that respondent’s primary allocation from the Mexican supply point does not increase the group’s taxable income. Citing section 6214(a), relating to this Court’s jurisdiction, peti- tioner urges that there “never can be a final determination (i.e., a primary alloca- tion) with respect to respondent’s proposed section 482 adjustments for the * * * [Mexican supply point] that can result in a deficiency.” Because “the final deter- mination (i.e., primary allocation reflected in a Tax Court decision) will be zero,” petitioner asserts that we should not (or cannot) decide the section 987 issue. To the extent petitioner is making a jurisdictional argument, it is clearly off base. The notice of deficiency decreased, by about $2 million in the aggregate, the section 987 losses petitioner had reported for 2007-2009. Petitioner in its petition - 213 - assigned error to that determination, and we plainly have jurisdiction to review it. See sec. 6213(a). Once this Court acquires jurisdiction, “that jurisdiction extends to the entire subject matter of the correct tax for the taxable year.” Naftel v. Com- missioner, 85 T.C. 527, 533 (1985). The fact that no portion of the overall defi- ciency will be directly traceable to a primary allocation from the Mexican supply point is irrelevant in determining our jurisdiction. In a related vein, petitioner suggests that respondent is improperly using section 482 in this instance for the sole purpose of generating a section 987 adjust- ment. Again, petitioner’s premise is that it “reported 100% of the income at issue on its U.S. returns in a manner that did not evade taxes or distort income.” For that reason, petitioner says, the Commissioner’s “offsetting section 482 adjus- tments * * * fall beyond section 482’s scope.” Assuming petitioner’s premise to be correct, its conclusion does not follow from its premise. Respondent determined, for the Mexican supply point, primary section 482 allocations parallel to those it determined for the Irish, Brazilian, Chilean, Costa Rican, and Swazi supply points. We have sustained those alloca- tions. The regulations provide that, when the IRS makes a primary allocation, “appropriate correlative allocations will also be made with respect to any other member of the group affected by the allocation,” including allocations that “cor- - 214 - respondingly decrease the income of the other member.” Sec. 1.482-1(g)(2)(i), Income Tax Regs. Petitioner cannot plausibly dispute that a correlative allocation decreasing the Mexican branch’s income is a logical corollary of the primary allocation increasing petitioner’s income.65 Having made the correlative allocation authorized by the regulations, the Commissioner determined that the section 987 losses petitioner had reported with respect to the Mexican branch were no longer correct. He accordingly recomputed those losses, increasing the reported losses by approximately $3.4 million for 2007-2008 and decreasing the reported losses by approximately $5.4 million for 2009. In principle, these recomputations resemble those that commonly occur when the IRS adjusts a taxpayer’s income in other respects, e.g., by recomputation of a medical expense deduction or deductible rental real estate loss to reflect changes to adjusted gross income. See secs. 213(a), 469(i). Alternatively, they 65 Whenever the IRS makes a primary allocation to one member of a consoli- dated group, there may be a correlative allocation that decreases the income of an- other member of the group, resulting in no net tax effect. But the statute and the regulations clearly permit sec. 482 allocations within a consolidated group. See sec. 1.482-1(f)(1)(iv), Income Tax Regs. (“Section 482 and the regulations there- under apply to all controlled taxpayers, whether the controlled taxpayer files a separate or consolidated U.S. income tax return.”); see also Guidant LLC v. Com- missioner, 146 T.C. 60, 78 (2016). - 215 - can be viewed as “further correlative allocations * * * required by the initial correlative allocation.” Sec. 1.482-1(g)(2)(i), Income Tax Regs. We have determined that allocations between TCCC and the Mexican sup- ply point are necessary to reflect income clearly. These allocations by themselves do not change the consolidated group’s overall income, but they do reduce the in- come of the Mexican branch, which did business in pesos. The allocations thus change the taxable income of the Mexican branch for purposes of computing peti- tioner’s foreign currency gain or loss under section 987. At the end of the day, the primary allocations, correlative allocations, and section 987 recomputations--taken together--produce the same result that would have occurred if TCCC and its Mexi- can branch had reported income consistently with the arm’s-length standard from the outset. Petitioner next contends that we should reject respondent’s section 987 re- computations for reasons analogous to those that led us to reject his position in Coca-Cola Co. & Subs. v. Commissioner, 149 T.C. 446 (2017). That Opinion addressed respondent’s disallowance of foreign tax credits (FTCs) that petitioner claimed for Mexican income taxes paid by the Mexican branch for the tax years at issue. Respondent contended that the Mexican branch had reported insufficient royalty expenses for use of petitioner’s intangible property, thus artificially inflat- - 216 - ing the branch’s income and the Mexican corporate tax paid thereon. Respondent contended that the Mexican taxes were to that extent “noncompulsory” payments ineligible for the FTC. See sec. 901; sec. 1.901-2(a)(2)(i), Income Tax Regs. Petitioner filed a motion partial summary judgment on this question, which we resolved in petitioner’s favor. We held that petitioner had calculated its Mexi- can tax liabilities “‘in a manner that [wa]s consistent with a reasonable interpreta- tion and application’ of Mexican law” so as to minimize petitioner’s reasonably expected liability for Mexican tax. See Coca-Cola Co., 149 T.C. at 459 (quoting section 1.901-2(e)(5)(i), Income Tax Regs.). We further held that petitioner had “exhausted all ‘effective and practical remedies’ to reduce its liability for Mexican tax.” Id. at 465 (quoting section 1.901-2(e)(5)(i), Income Tax Regs.). We accord- ingly held that the Mexican taxes were compulsory levies eligible for the FTC. Ibid. We find no analogy between our analysis in that Opinion and the analysis that is appropriate here. Petitioner’s eligibility for the FTC depended on the character of its payments to the Mexican Government; this required an investiga- tion of Mexican law and an inquiry into whether petitioner had reasonably inter- preted that law. See id. at 455-459. We rejected respondent’s FTC argument not on the theory that it was an improper collateral adjustment--the theory petitioner - 217 - urges here--but because respondent erred in characterizing petitioner’s Mexican tax payments as “noncompulsory.” We find no comparable error by the IRS here: Computations of section 987 gain or loss are mechanical (albeit complex), and they follow more or less automatically once correlative allocations have been made to the income of the Mexican branch.66 Finally, petitioner suggests that respondent’s correlative allocations (and the recomputations of section 987 loss they generate) may be premature. The regula- tions provide that, for purposes of making collateral adjustments, “a primary allo- cation will not be considered to have been made (and therefore, correlative alloca- tions are not required to be made) until the date of a final determination with res- pect to the allocation under section 482.” Sec. 1.482-1(g)(2)(iii), Income Tax Regs. In this case, the “final determination with respect to the allocation under section 482” will not occur until this Court’s decision has become final within the meaning of section 7481. See id. subdiv. (iii)(E). 66 Petitioner attempts a rhetorical link to our earlier Opinion by insisting that it calculated its transfer pricing with the Mexican supply point “pursuant to a rea- sonable interpretation and application of Mexican law.” But while that fact was quite relevant in assessing whether the Mexican tax payments were compulsory, see sec. 1.901-2(e)(5)(i), Income Tax Regs., it has no relevance in determining whether petitioner’s reported section 987 losses needed to be recomputed. - 218 - We think petitioner is trying to convert a shield into a sword. In providing that correlative allocations “are not required to be made” until a judicial decision has become final, the regulation affords the Commissioner more time to make cor- relative allocations. Nothing in this regulation prevents the Commissioner from making such adjustments earlier, e.g., in the notice of deficiency, as he did here. See Inverworld, Inc. v. Commissioner, T.C. Memo. 1997-226, 73 T.C.M. (CCH) 2777, 2786-2788 (evaluating whether correlative allocations were appropriate concurrently with examining the Commissioner’s primary section 482 allocation), supplementing T.C. Memo. 1996-301. And nothing in the regulation affects our jurisdiction. Indeed, it is unclear how we would ever be able to review respond- ent’s section 987 adjustments if we do not review them now. B. Dividend Offset The 1996 closing agreement specified that the compensation due petitioner from the supply points, for use of petitioner’s intangible property during 1987- 1995, would be determined using the 10-50-50 method. The amounts the supply points were obligated to pay petitioner were thus in the nature of royalties. How- ever, the closing agreement permitted the supply points to satisfy their royalty obligation by paying actual royalties or by repatriating funds to petitioner in other ways, e.g., by paying dividends. - 219 - For the tax years at issue, as for all years after 1995, petitioner calculated the supply points’ royalty obligation under the 10-50-50 method. As had been permitted by the closing agreement, the supply points discharged about $1.8 bil- lion of their royalty obligation to petitioner during 2007-2009 by remitting divi- dends rather than royalties. The Brazilian and Chilean supply points remitted, in satisfaction of their royalty obligations, aggregate dividends of about $887 million and $233 million, respectively. Atlantic, which operated the Costa Rican, Irish, Egyptian, and Swazi supply points, remitted aggregate dividends of about $682 million in satisfaction of those supply points’ royalty obligations. Petitioner contends that these dividends should be offset against, i.e., should reduce, the royalty obligations of the supply points as determined in this Opinion. Failure to allow such an offset, petitioner urges, would in substance require the supply points to repatriate income twice, subjecting it to tax each time--first when remitted as surrogates for royalties, and again when reallocated as actual royalties. In a very real sense, failure to allow a dividend offset would punish petitioner for adhering to the terms of its tax obligations as it understood them at the time. Peti- tioner urges that this result would be inequitable and irrational, and we agree.67 67 Petitioner concedes that allowing dividend offsets would require it to for- feit the “deemed paid” FTCs that it claimed on those dividends during 2007-2009, (continued...) - 220 - We need not rely solely on equitable grounds, however, because we find a strong technical basis for reaching the same result under “collateral adjustment” principles. The regulations provide that “[a]ppropriate adjustments must be made to conform a taxpayer’s accounts to reflect allocations made under section 482.” Sec. 1.482-1(g)(3)(i), Income Tax Regs. Where (as here) income is reallocated from a subsidiary to a parent, the subsidiary ends up with excess income (cash) on its books. To account for how that cash got there, the parent is normally deemed to have made a capital contribution to the subsidiary in the amount of the primary section 482 reallocation. Ibid. Alternatively, the regulations permit “repayment of the allocated amount [by the subsidiary to the parent] without further income tax consequences.” Ibid. The regulations state that the latter type of conforming adjustment may be implemented “pursuant to such applicable revenue procedures as may be provided by the Commissioner.” Ibid. When the parties executed the closing agreement in 1996, the applicable revenue procedure was Rev. Proc. 65-17, 1965-1 C.B. 833. In the closing agree- 67 (...continued) as the closing agreement had permitted for dividends paid with respect to tax years 1987-1995. Respondent has accordingly amended his answer to assert that FTCs of $40,717,804 for 2007, $65,941,179 for 2008, and $49,977,463 for 2009 should be disallowed if we permit dividend offsets. Petitioner has not disputed those numbers. - 221 - ment the parties agreed that the “product royalties” due petitioner from specified supply points would be calculated using the 10-50-50 method and that, under that method, petitioner would be allocated $337,896,485 of additional “product royal- ties” during 1987-1995. They also agreed that petitioner would be granted “the treatment provided by section 4 of Rev. Proc. 65-17 with respect to certain of the section 482 allocations.” Section 4 of Rev. Proc. 65-17, 1965-1 C.B. at 834, afforded a qualifying taxpayer two forms of relief with regard to adjustment of accounts. First, the tax- payer was permitted to exclude from gross income, for the year for which the sec- tion 482 allocation was made, any taxable dividends it had received during that year from the subsidiary with which it had engaged in the transaction giving rise to the section 482 allocation. Id. sec. 4.01, 1965-1 C.B. at 834. In effect, the divi- dends were offset against the primary section 482 allocation. Second, with respect to the balance of the allocation, the taxpayer was per- mitted “to establish an account receivable” from the subsidiary with which it had engaged in the transaction giving rise to the section 482 allocation. Id. sec. 4.02, 1965-1 C.B. at 835. If the subsidiary repaid the account receivable--by remitting a dividend or otherwise--within a specified 90-day period, the payment could be received by the parent “without tax consequences.” Ibid. - 222 - The 1996 closing agreement afforded petitioner both forms of relief. The agreement permitted petitioner to offset against the reallocation of royalty income $77,789,895 of dividends that it had received from the supply points during 1987- 1995. And it permitted petitioner to establish with the supply points “accounts receivable” in the aggregate amount of $260,106,590, which could be received by petitioner “without further Federal income tax consequences” if the accounts were repaid within a specified 90-day period. Beginning with its return for 1996, petitioner claimed the same “dividend offset” treatment that the closing agreement had allowed for dividends paid during 1987-1995. Petitioner did so by making what is called a “taxpayer-initiated” pri- mary allocation under section 1.482-1(a)(3), Income Tax Regs. That provision, titled “Taxpayer’s use of section 482,” provides: “If necessary to reflect an arm’s length result, a controlled taxpayer may report on a timely filed U.S. income tax return (including extensions) the results of its controlled transactions based upon prices different from those actually charged.” For example, when preparing its return for 1996, petitioner used the 10-50- 50 method to calculate the royalties due from its supply points for 1996. It com- pared that amount to the royalties it had actually received from those supply points during 1996--viz., the “prices actually charged”--which were less than the 10-50- - 223 - 50 required amounts. Petitioner then initiated a section 482 adjustment against itself for the balance of the supply points’ royalty obligation. And it offset against that balance the taxable dividends it had received from the supply points during 1996, as the closing agreement had permitted for dividends paid during 1987- 1995. Petitioner claimed the same “dividend offset” treatment when preparing its returns for every year after 1995, including the tax years at issue. In 1999 the IRS issued a new revenue procedure that superseded Rev. Proc. 65-17. See Rev. Proc. 99-32, 1999-2 C.B. 296. This new revenue procedure was effective for taxable years beginning after August 23, 1999. Id. sec. 6.01, 1999-2 C.B. at 301. It provides essentially the same two forms of relief as had its prede- cessor. Of particular relevance here, Rev. Proc. 99-32 permits the offset, against a primary section 482 allocation, of otherwise-taxable dividends received from the other controlled taxpayer “during the taxable year for which the section 482 allo- cation is made.” Id. sec. 4.02, 1999-2 C.B. at 299-300. And Rev. Proc. 99-32 specifically provides that dividend offset treatment is available in the case of tax- payer-initiated primary adjustments, as well as for section 482 adjustments initi- ated by the Commissioner. Id. sec. 2, 1999-2 C.B. at 298 (citing section 1.482- 1(a)(3), Income Tax Regs.). - 224 - However, Rev. Proc. 99-32 added a new procedural requirement that did not appear in Rev. Proc. 65-17. A taxpayer seeking to avail itself of dividend offset treatment for a taxpayer-initiated section 482 adjustment was directed to “file a statement with its Federal income tax return reporting the primary adjustment.” Rev. Proc. 99-32, sec. 5.02, 1999-2 C.B. at 300. Petitioner did not file such state- ments with its 2007-2009 returns, and respondent contends that this omission is fatal to petitioner’s claim to dividend offsets. Petitioner urges that it substantially complied with the requirements of Rev. Proc. 99-32. On the unusual facts of this case, we agree. The IRS prefaced Rev. Proc. 99-32 with “Supplementary Information” that summarized the changes it made. See Rev. Proc. 99-32, 1999-2 C.B. at 296-297. The IRS had originally proposed to eliminate dividend offset treatment altogether, expressing concern that “[d]ividend offset treatment is inconsistent with the cur- rent policy under sections 482 and 6662(e) that taxpayers should strive upfront to price their related party transactions in compliance with the arm’s length stan- dard.” Announcement 99-1, 1999-1 C.B. 302 (cited in Rev. Proc. 99-32). In other words, the IRS feared that the potential availability of dividend offset treatment in- centivized taxpayers to play what has been called the “audit lottery”: - 225 - [T]he existence of the possibility of a dividend offset lessens the incentive built into the section 482 and 6662(e) regulations [for taxpayers] to comply upfront and conform their transfer pricing to the arm’s length standard in the first instance, since they are able to mitigate the tax effect of non-arm’s length pricing by means of the dividend offset. * * * [Ibid.] The IRS received numerous comments in response to Announcement 99-1. Several commenters “expressed the view that elimination of dividend offsets would discourage current repatriation of earnings, prolong transfer pricing dis- putes, and pose problems when payment of a form of income is restricted under foreign law.” See Rev. Proc. 99-32, 1999-2 C.B. at 296. Agreeing with these comments, the Commissioner changed course and decided to continue to “allow[] taxpayers to offset accounts by distributions, including those that would otherwise be dividends, in the same tax year as that to which a taxpayer-initiated primary adjustment relates.” Ibid. However, to reduce the risk of taxpayers’ playing the “audit lottery,” Rev. Proc. 99-32 required that dividend offset treatment must be “claimed on a timely- filed income tax return (including extensions).” Ibid. And taxpayers claiming such treatment were directed to file with the return a statement providing certain information, including “[a] statement that the taxpayer desires * * * [dividend off- - 226 - set treatment] for the years indicated and acknowledges that it is bound by its elec- tion of such treatment.” Id. sec. 5.02, 1999 C.B. at 300. Petitioner made taxpayer-initiated section 482 adjustments and elected divi- dend offset treatment on timely filed returns for 2007, 2008, and 2009, as the regu- lations and the revenue procedure require. See sec. 1.482-1(a)(3), Income Tax Regs.; Rev. Proc. 99-32, sec. 5, 1999-2 C.B. at 300. But it did not attach to those returns an explanatory statement informing the IRS of its election and agreeing to be bound by that election. Under the peculiar circumstances of this case, we find that this was harmless error. In omitting to include an explanatory statement as Rev. Proc. 99-32 direct- ed, petitioner was not playing the “audit lottery.” Quite the contrary: It was reporting its royalty income and claiming dividend offsets exactly as had been agreed in the 1996 closing agreement. The IRS was aware that petitioner was doing this, because it had examined petitioner’s returns for every year between 1996 and 2006. And although petitioner did not explicitly agree to be bound by its election, it was bound in a practical sense, because doing otherwise would have caused it to forfeit the penalty protection that the closing agreement afforded. In a variety of contexts, we have held that directives in regulations and other IRS guidance can be satisfied by substantial rather than strict compliance. “There - 227 - is no defense of substantial compliance for failure to comply with the essential requirements of the governing statute.” Estate of Clause v. Commissioner, 122 T.C. 115, 122 (2004). But where “requirements are procedural or directory in that they are not of the essence of the thing to be done but are given with a view to the orderly conduct of business, they may be fulfilled by substantial * * * compli- ance.” Bond v. Commissioner, 100 T.C. 32, 41 (1993) (quoting Taylor v. Com- missioner, 67 T.C. 1071, 1077-1078 (1977)). Petitioner filed timely returns for 2007-2009. On those returns it made tax- payer-initiated section 482 adjustments and timely elected dividend offset treat- ment. Given the peculiar circumstances of this case, both of these facts were well- known to the IRS, and petitioner’s inclusion of explanatory statements would have added nothing to the IRS’ sum of knowledge. On these facts, we find that peti- tioner has satisfied the essential requirements of the regulations and Rev. Proc. 99- 32. Where a taxpayer has made a valid and timely election on his tax return, regulations imposing procedural directives regarding that election, such as including statements or documents with the return, can be satisfied by substantial rather than literal compliance.68 68 See Columbia Iron & Metal Co. v. Commissioner, 61 T.C. 5, 8-10 (1973) (finding substantial compliance with regulation governing election by corporate (continued...) - 228 - It is true that petitioner could have filed with its returns the explanatory statements that Rev. Proc. 99-32 directed, even if only as a protective matter. But we conclude on the facts here that petitioner omitted to do this “solely through inadvertence.” See Hewitt v. Commissioner, 109 T.C. 258, 265 n.10 (1997), aff’d, 166 F.3d 332 (4th Cir. 1998). Insisting on strict compliance would deny petitioner dividend offsets of $1.8 billion, effectively requiring the supply points to repatri- ate taxable royalties twice. We have no difficulty concluding that this “would constitute a sanction which is not warranted or justified.” Bond, 100 T.C. at 42; see Columbia Iron & Metal Co. v. Commissioner, 61 T.C. 5, 10 (1973) (declining to insist on strict compliance where it would “establish a sanction which is out of 68 (...continued) taxpayer to treat charitable contribution as having been paid during prior taxable year); Hoffman v. Commissioner, 47 T.C. 218, 236-237 (1966) (finding substan- tial compliance with regulation governing revocation of election by corporate tax- payer, despite failure to include specified statements with tax return), aff’d, 391 F. 2d 930 (5th Cir. 1968); Sperapani v. Commissioner, 42 T.C. 308, 330-333 (1964) (finding substantial compliance with regulation governing election by taxpayer to have sole proprietorship taxed as domestic corporation, despite failure to include formal statement of election with tax return); Cary v. Commissioner, 41 T.C. 214, 218-219 (1963) (finding substantial compliance with regulation govern- ing stock redemption, despite failure to include corporate agreement with tax re- turn). By contrast, we have held that the substantial compliance doctrine has no application where a taxpayer has failed to make a timely election at all. See, e.g., Dunavant v. Commissioner, 63 T.C. 316, 320 (1974); Kohli v. Commissioner, T.C. Memo. 2009-287. - 229 - proportion to the shortcoming and not warranted or justified under the circum- stances”). To implement the foregoing, Decision will be entered in due course under Rule 155. - 230 - APPENDIX Petitioner’s Expert Witnesses 1. Susan Athey Dr. Athey is a professor of economics and technology at Stanford Graduate School of Business. She received her B.A. in economics, computer science, and mathematics from Duke University and her Ph.D. in economics from Stanford. She is also a member of the National Academy of Sciences and a research asso- ciate for the National Bureau of Economic Research. She previously served as consulting chief economist for Microsoft Corporation and now serves on numer- ous corporate boards. Her research and publications focus on industrial organiza- tion, the study of markets, and competition. The Court recognized Dr. Athey as an expert in the economics of industrial organization. 2. Anthony Barbera Dr. Barbera is a senior consultant at Charles River Associates. He received his B.A. in mathematics from Loyola College and his Ph.D. in economics from the University of Maryland. He previously served as a consultant at KPMG, Deloitte, and PwC before starting his own professional services firm, which Charles River Associates acquired in 2006. He has advised taxpayers on international tax con- troversies, including transfer pricing disputes, for more than 30 years. The Court - 231 - recognized Dr. Barbera as an expert in economics with a specialization in transfer pricing. 3. Randolph Bucklin Dr. Bucklin is a professor of marketing at the UCLA Anderson School of Management, where he holds the Peter W. Mullin Chair in Management. He re- ceived his A.B. in economics from Harvard University, his M.S. in statistics from Stanford University, and his Ph.D. in business from Stanford. He has extensive teaching and research experience in the areas of consumer purchasing decisions, marketing channels, and digital advertising. He has published his research in numerous academic journals. The Court recognized Dr. Bucklin as an expert in marketing and distribution. 4. Michael Cragg Dr. Cragg is the chairman of the Brattle Group, a global economic consult- ing firm. He received his B.S.E. in engineering and architecture from Princeton University, his M.A. in economics from the University of British Columbia, and his Ph.D. in economics from Stanford University. He has also taught economics courses at Columbia University and at UCLA Anderson School of Management. He has extensive teaching and consulting experience in the areas of industrial organization, financial economics, public finance, and valuations. He has also - 232 - written numerous articles on antitrust law, fiscal policy, and transfer pricing. The Court recognized Dr. Cragg as an expert in economics and transfer pricing. 5. Robert Dolan Dr. Dolan is a Baker Foundation Professor at Harvard Business School. He received his B.A. in mathematics from Boston College and an M.B.A. and a Ph.D. in business administration from the University of Rochester. He has taught at several business schools, including the University of Michigan and the University of Chicago. He has extensive teaching and consulting experience in the areas of customer purchasing decisions and building brand awareness. He has published numerous books and articles on marketing, product pricing, and brands. The Court recognized Dr. Dolan as an expert in marketing. 6. David Franklyn Professor Franklyn is a professor of intellectual property law at Golden Gate University School of Law. He received his B.A. in history, philosophy, and reli- gion from Evangel College and his J.D. from the University of Michigan Law School. He has experience working as a consultant, advising companies on issues relating to trademarks and management of trademark portfolios. He is also the co- author and editor-in-chief of McCarthy’s Desk Encyclopedia of Intellectual Prop- erty, and he has written numerous articles on trademark law. The Court recog- - 233 - nized Professor Franklyn as an expert in international trademark law, including Brazilian industrial property law as it relates to trademarks and the empirical evaluation of trademarks. 7. Richard Hall Mr. Hall is the chairman and founder of Zenith Global, a consulting firm that specializes in the food and beverage industry. He received his honors degree in economics and government from the University of Bath. Mr. Hall has provided consulting services related to the food and beverage industry for more than 30 years. He is also the publisher of Beverage Digest and the chairman of FoodBev Media, which he founded in 2000. He previously served as a director of the Dairy Trade Federation and Secretary General of the European Federation of Dairy Retailers. The Court recognized Mr. Hall as an expert in the beverage industry, which includes NARTD beverages. 8. Dominique Hanssens Dr. Hanssens is a research professor of marketing at the UCLA Anderson School of Management. He received his licentiate in applied econometrics from the University of Antwerp and his M.S. and Ph.D. degrees in management from Purdue University. He has written numerous award-winning books and articles on economics, marketing strategies, marketing productivity, and data analytics. He - 234 - also provides consulting services to dozens of multinational enterprises, including Coca-Cola. He previously served as executive director of the Marketing Science Institute. The Court recognized Dr. Hanssens as an expert in marketing and mar- keting science. 9. Kevin Lane Keller Dr. Keller is the E.B. Osborn Professor of Marketing at the Tuck School of Business at Dartmouth. He received his A.B. in mathematics and economics from Cornell University, his M.B.A. from Carnegie Mellon University, and his Ph.D. in marketing from the Fuqua School of Business at Duke University. He previously taught at the University of California, Stanford University, the University of North Carolina, and Duke University. He has extensive teaching and research experi- ence in the areas of advertising, branding, and marketing communications. He has also written two textbooks that are used in graduate-level marketing courses. The Court recognized Dr. Keller as an expert in marketing, advertising, and branding. 10. Michael Lasinski Mr. Lasinski is a senior managing director at Ankura Consulting Group. He holds a B.S. in electrical engineering and an M.B.A. from the University of Michi- gan. He is a certified public accountant and is certified in financial forensics. He has worked in intellectual property for more than 20 years. He was formerly the - 235 - president of the Licensing Executives Society of the United States and Canada. Mr. Lasinski’s work has focused primarily on reviewing and analyzing hundreds of intellectual property license agreements. He has reviewed thousands of license agreements and has negotiated between 50 and 100. The Court recognized Mr. Lasinski as an expert in intellectual property licensing and negotiation. 11. Timothy Luehrman Dr. Luehrman is a retired professor of finance from Harvard Business School. He received his B.A. in economics and English literature from Amherst College and his M.B.A. and Ph.D. in business economics from Harvard Business School. He taught graduate-level finance courses at Harvard, the International Institute for Management Development in Switzerland, the Massachusetts Institute of Technology Sloan School of Management, and the Arizona State University Thunderbird School of Global Management. He has extensive teaching and re- search experience in the areas of corporate finance and financial economics. The Court recognized Dr. Luehrman as an expert in financial economics. 12. Keith Reams Mr. Reams is a principal at Deloitte Tax who focuses his practice on trans- fer pricing. He currently serves as the firm’s U.S. and Global Leader for Clients and Markets. He received his B.S. in chemical engineering from Stanford Univer- - 236 - sity and an M.A. in economics from California State University at Sacramento, and he has completed the course requirements for a Ph.D. in international finance at New York University Graduate School of Business. He has advised clients on transfer pricing matters for more than 30 years. The Court recognized Mr. Reams as an expert in economics and transfer pricing. 13. Sanjay Unni Dr. Unni holds a Ph.D. in economics and is currently the managing director of the Berkeley Research Group, an expert services firm specializing in economics and financial analysis. He received his B.A. in economics from the University of Delhi in India and his master’s degree and Ph.D. in economics from Southern Methodist University. He has taught courses on corporate finance, investment analysis, market structures, and finance at several institutions in the United States and the United Kingdom. As a transfer pricing economist he has drafted more than 30 transfer pricing reports, primarily related to technology firms. He has also published and taught in the field of financial economics. The Court recognized Dr. Unni as an expert in economics, financial economics, and transfer-pricing economics. - 237 - 14. Robert Wentland Mr. Wentland is a senior managing director at Ankura Consulting Group. He received his B.B.A. in accounting from the University of Wisconsin Madison. He is a certified public accountant and is also certified by the AICPA in financial forensics. He previously worked for Arthur Anderson as an accounting and con- sulting partner and for Huron Consulting as a managing director. Mr. Wentland has provided forensic accounting and data analysis services related to international tax controversies, including transfer pricing disputes, for more than 20 years. The Court recognized Mr. Wentland as an expert in financial statement analysis and forensic accounting. 15. Robert Willig Dr. Willig holds a Ph.D. in economics and is an Emeritus professor of econ- omics and public affairs at Princeton University. He received his A.B. in mathe- matics from Harvard University, his M.S. in operations research from Stanford University, and his Ph.D. in economics from Stanford. While teaching at Prince- ton, he also served as a principal external adviser at the Inter-American Develop- ment Bank and as a Deputy Assistant Attorney General at the U.S. Department of Justice. His research and publications focus on asset decay, how markets work, how markets influence economic outcomes, and how the forces of economics - 238 - affect the marketplace. The Court recognized Dr. Willig as an expert in micro- economics. Respondent’s Expert Witnesses 1. Kusum Ailawadi Dr. Ailawadi is the Charles Jordan 1911 TU’12 Professor of Marketing at the Tuck School of Business at Dartmouth. She received her B.S. in physics from St. Stephen’s College in India, her M.B.A. from the Indian Institute of Manage- ment, and her Ph.D. from the Darden Graduate School of Business at the Universi- ty of Virginia. She has extensive teaching experience and her scholarly articles have won a host of awards from established marketing journals, including the Journal of Marketing Research, Marketing Science, and the Journal of Marketing. She is also the president-elect of the INFORMS Society for Marketing Science. The Court recognized Dr. Ailawadi as an expert in marketing. 2. Brian Becker Dr. Becker is the president and founder of Precision Economics, an expert services firm specializing in transfer pricing, valuation, commercial damages, in- tellectual property, international trade, and antitrust. He received his B.A. as a double major in applied mathematics and economics from Johns Hopkins Univer- sity. He received an M.A. and a Ph.D. in applied economics from the University - 239 - of Pennsylvania Wharton School of Business. He previously served as a senior managing economist at LECG, another expert services firm. He has also taught courses on corporate finance, statistics, and operations at several universities. As a transfer pricing economist he has published dozens of articles and drafted numerous transfer pricing reports. He also serves on the board of the Mathemati- cal Association of America and the Gerald R. Ford Presidential Foundation. The Court recognized Dr. Becker as an expert in economics and transfer pricing. 3. Jorge Luis Contreras Professor Contreras is a professor of intellectual property law at the Uni- versity of Utah S.J. Quinney College of Law and the founder of Contreras Legal Strategy, a boutique legal advisory firm. He received his B.A. in English and B.S. in electrical engineering from Rice University and his J.D. from Harvard Law School. He was previously an associate and partner at Wilmer, Cutler, Pickering, Hale, and Dorr LLP (WilmerHale), where he advised clients on transactions in- volving intellectual property, including licensing, technology development, and product manufacturing, distribution, and sale. He has edited four books and writ- ten more than 60 articles and book chapters in the areas of intellectual property, technology licensing, technical standards, patent litigation, and regulation of - 240 - science. The Court recognized Professor Contreras as an expert in intellectual property law and international intellectual property transactions. 4. Paul Farris Professor Farris is the Landmark Communications Professor Emeritus of Business Administration at the University of Virginia Darden School of Business. He received his B.S. in business economics from the University of Missouri, his M.B.A. from the University of Washington at Seattle, and his D.B.A. from Har- vard University. He previously taught marketing management and advertising at Harvard Business School. He has extensive teaching and consulting experience in the areas of marketing, marketing productivity, and brand management. He has coauthored several books, one of which was selected by Strategy + Business as the marketing book of the year. The Court recognized Professor Farris as an expert in marketing. 5. Stuart Harden Mr. Harden is a partner at Hemming Morse, a consulting firm specializing in forensic and financial accounting. He received his B.A. in business administra- tion from Wichita State University. He is a certified public accountant, certified by the AICPA in financial forensics, and a certified fraud examiner. He has pro- vided forensic accounting services for more than 30 years, and served for 14 years - 241 - as a member of the Emerging Issues Task Force of the Financial Accounting Standards Board (FASB). He has also served as the chief investigator for the State Boards of Accountancy in California and Texas. The Court recognized Mr. Harden as an expert in financial statement analysis and forensic accounting. 6. Andrew Metrick Dr. Metrick is the Janet L. Yellen Professor of Finance and Management at the Yale School of Management, where he has taught since 2008. He received his B.A. in economics and mathematics from Yale and his A.M. and Ph.D. in eco- nomics from Harvard. He previously taught at Harvard and the University of Pennsylvania Wharton School of Business. From 2009 to 2010, Dr. Metrick took a leave of absence from Yale to serve as a senior economist and (later) the Chief Economist at the Council of Economic Advisers in Washington, D.C. His re- search and publications focus on financial stability and the regulation of financial institutions. The Court recognized Dr. Metrick as an expert in finance and economics. 7. T. Scott Newlon Dr. Newlon is a managing director at Horst Frisch, an economic consulting firm specializing in economics and transfer pricing. He received his B.S. in eco- nomics from the University of Delaware and his M.A. and Ph.D. in economics - 242 - from Princeton University. He previously served as a principal at KPMG. He also served as a senior economist at the U.S. Department of the Treasury. Dr. Newlon was a principal author of both the 1994 transfer pricing regulations and the 1995 cost-sharing regulations, and he participated in the drafting of the 1995 OECD Transfer Pricing Guidelines. He has also published in the field of international tax and transfer pricing. The Court recognized Dr. Newlon as an expert in economics and transfer pricing. 8. David J. Reibstein Dr. Reibstein is a professor of marketing at the University of Pennsylvania Wharton School of Business. He received his undergraduate degrees in business administration, statistics, and political science from the University of Kansas and his Ph.D. in industrial administration from Purdue University. He previously taught at Stanford Business School and Harvard Business School and also served as the chairman and a member of the board of directors of the American Marketing Association. He has extensive teaching and consulting experience in the areas of marketing and marketing metrics. His research and publications focus on mar- keting performance management and global branding. The Court recognized Dr. Reibstein as an expert in marketing. - 243 - 9. Jean-Marc Thevenin Mr. Thevenin is the founder of Ramaco Development, a consulting firm specializing in the food and beverage industry. He received his B.S in agricultural science and engineering from the Institut Supérieur Agricole de Beauvais (ISAB), and his master’s in business studies from the Institut d’Administration des Entre- prises de Paris (IAE Paris). He has worked in the food and beverage industry for more than 30 years. He previously worked for SABMiller, where he served as the vice president of business development and strategy for Latin America. The Court recognized Mr. Thevenin as an expert in operational management of branded food and beverages. 10. Juliana L.B. Viegas Dr. Viegas is a licensed Brazilian attorney specializing in Brazilian Intellec- tual property law. She now serves as the managing partner of J L Viegas Comer- cio Ltda., and she previously was of counsel to Trench, Rossi & Watanabe Advo- gados. She received her law degree and Ph.D. in commercial law from the Uni- versity of São Paulo School of Law. Dr. Viegas has more than 50 years of experi- ence advising companies on issues relating to intellectual property and Brazilian industrial property law. She has also helped file more than 2,000 trademark appli- cations and has negotiated more than 500 licensing agreements. The Court recog- - 244 - nized Dr. Viegas as an expert in Brazilian property law and practice, including Brazilian trademark law and practice.