Beyond the Arm’s-Length Price: U.S. Transfer Pricing, Cost Sharing and the Coca-Cola Case

Professor David Rosenbloom alongside participants of the Masterclass.

When the arm’s-length principle meets the realities of multinational business

The arm’s-length principle is central to modern transfer pricing: related-party transactions are assessed by reference to the conditions that independent parties might have agreed under comparable circumstances. Yet applying that principle can become considerably more difficult when the transaction involves unique intangibles, integrated multinational operations or facts for which reliable comparables simply do not exist.

These practical difficulties formed the focus of Session 3: The Arm’s–Length Method and Peculiarities of U.S. Transfer Pricing, delivered by Professor David Rosenbloom of New York University School of Law on 1 September 2026 as part of the Masterclass with Professor David Rosenbloom and Professor Dr René Matteotti ↗ (opens in new tab), held from 31 August to 2 September 2026. The programme brings together international tax perspectives across policy, transfer pricing, treaties, dispute resolution and current international tax issues.

Professor Rosenbloom, who retired in May 2025 as James S. Eustice Visiting Professor of Taxation and Director of the International Tax Program at NYU School of Law, approached the subject through the lens of U.S. transfer-pricing practice, litigation and the practical problems that arise when theoretical rules meet complex business arrangements.

A central theme of the session was that the arm’s-length principle may offer an appealing conceptual framework, but applying it does not necessarily produce one precise answer.

The arm’s-length principle does not always produce a single price

Professor Rosenbloom began by examining a fundamental tension within transfer pricing.

In an uncontrolled transaction, the market provides an important mechanism for determining price. Independent buyers and sellers negotiate with their own interests in mind. In a controlled transaction, however, the parties belong to the same economic group. The market friction that would ordinarily constrain their behaviour is therefore absent, creating the need for tax rules to determine an appropriate allocation of income.

The OECD describes the arm’s-length principle as the international consensus for valuing cross-border transactions between associated enterprises. Its Transfer Pricing Guidelines place significant emphasis on comparability and the reliability of the information used to determine an arm’s-length outcome. (See: OECD: Transfer Pricing Guidelines ↗ (opens in new tab))

The difficulty, as highlighted in the session, is that the closer one gets to highly differentiated businesses and unique intangibles, the harder it becomes to identify genuinely comparable uncontrolled transactions.

Professor Rosenbloom illustrated this with situations involving branded products and unique intellectual property. Where there is no sufficiently comparable uncontrolled transaction, practitioners and courts may have to broaden the search or consider other methods.

This led to a broader observation: an arm’s-length analysis may ultimately produce a range of plausible outcomes, rather than a single objectively discoverable price. Differences in bargaining power, information, commercial circumstances and the contributions of the respective parties can all affect what independent parties might have agreed.

The practical challenge is therefore not simply to find a number, but to establish why a particular result falls within a defensible arm’s-length range.

Comparable profits can be powerful precisely because perfect comparables are difficult to find

The discussion then turned to the evolution of U.S. transfer-pricing methods and the emergence of the Comparable Profits Method (CPM).

Professor Rosenbloom described CPM as particularly useful where transaction-level comparables are difficult to identify. Rather than attempting to establish a precise price for a unique transaction, the method examines whether the profitability of a tested party falls within a range that can be supported by comparable businesses.

The U.S. regulations require taxpayers and the IRS to apply the best method rule, under which the method that provides the most reliable measure of an arm’s-length result is selected after considering the relevant facts and comparability.

This is also where the U.S. approach differs in important ways from the OECD framework. Professor Rosenbloom noted that the U.S. regulations are generally more prescriptive, while the OECD Guidelines provide a framework that leaves more room for judgment.

He also discussed the relationship between the U.S. CPM and the OECD’s Transactional Net Margin Method (TNMM). Although the two approaches share important characteristics, the discussion highlighted that their development reflects different legal and institutional histories.

For practitioners, the more important lesson may be methodological: when reliable transaction-level evidence is unavailable, transfer pricing increasingly requires careful consideration of what evidence can reliably demonstrate whether a return is commercially reasonable.

The session also touched on the growing potential role of artificial intelligence in this area. Professor Rosenbloom pointed to the large volumes of corporate financial information available in the United States through securities filings and observed that AI could assist in identifying and analysing large populations of potential comparables.

The underlying data quality, comparability analysis and professional judgment, however, remain critical. Technology can make a search broader and faster; it does not by itself establish that two businesses are economically comparable.

U.S. cost sharing reflects a distinctive approach to intangible development

Cost sharing was another major focus of the session.

Professor Rosenbloom described the U.S. cost-sharing regime as a particularly distinctive feature of U.S. transfer pricing, especially in the context of developing intangibles.

Under the U.S. regulations, a cost-sharing arrangement involves controlled participants sharing the costs and risks of developing cost-shared intangibles in proportion to their reasonably anticipated benefits. The rules also address platform contributions and other transactions associated with the development and exploitation of those intangibles.

The OECD similarly recognises cost contribution arrangements (CCAs) as contractual arrangements under which associated enterprises share contributions and risks relating to the joint development, production or acquisition of intangibles, tangible assets or services. (See: OECD: Transfer Pricing Guidelines ↗ (opens in new tab))

Professor Rosenbloom’s discussion went beyond the mechanics of cost sharing and examined the policy and practical tensions surrounding it.

A recurring issue was whether arrangements labelled as cost sharing genuinely reflect the economic contributions and risks of the participants. He questioned whether independent parties would necessarily enter into arrangements to share the development of highly valuable intangibles in the same way as related companies.

The discussion also considered the role of buy-in or platform contribution transactions, particularly where an existing intangible or capability is brought into a cost-sharing arrangement.

These questions illustrate why cost-sharing arrangements cannot be evaluated solely by looking at the contractual allocation of costs. The anticipated benefits, the rights obtained, the contributions made and the economic substance of the arrangement all matter.

For Singapore-based multinational groups, this is particularly relevant because IRAS’ current guidance similarly emphasises that cost contribution arrangements must satisfy the arm’s-length principle. Among other requirements, participants must share the risks associated with achieving the anticipated outcomes and their contributions must be consistent with what independent parties would have agreed under comparable circumstances. (See IRAS: Trasfer Pricing ↗ (opens in new tab))

Secondary adjustments show that transfer pricing does not end with the primary adjustment

Another part of the session examined secondary adjustments.

A primary transfer-pricing adjustment changes the taxable income recognised by a taxpayer. But the economic consequences do not necessarily stop there. If the tax authority determines that a controlled transaction should have produced a different result, the question arises as to what happened to the corresponding amount already recorded elsewhere in the group.

Professor Rosenbloom explained three forms of secondary adjustment encountered under U.S. rules: correlative adjustments, set-offs and conforming adjustments.

The distinction is important because secondary adjustments can create consequences beyond the original transfer-pricing calculation, including potential withholding-tax or dividend consequences depending on how the deemed movement of funds is characterised.

The OECD has long recognised that secondary adjustments can create additional tax consequences and, in some circumstances, additional double-taxation risks. The OECD Guidelines also distinguish secondary adjustments from corresponding adjustments and note that countries differ in whether and how they apply them.

This makes the interaction between domestic transfer-pricing rules and international dispute-resolution mechanisms particularly important. A primary adjustment in one jurisdiction does not automatically require another jurisdiction to accept the corresponding economic consequence.

In Singapore, IRAS provides mechanisms including Advance Pricing Arrangements (APAs) and the Mutual Agreement Procedure (MAP) to help prevent and resolve cross-border transfer-pricing disputes. (See IRAS: Transfer Pricing ↗ (opens in new tab))

Coca-Cola brings the methodological and procedural issues together

The centrepiece of the latter part of the session was The Coca-Cola Company and Subsidiaries v. Commissioner, one of the most closely watched U.S. transfer-pricing disputes.

The case concerns Coca-Cola's foreign manufacturing affiliates, or “supply points”, which licensed intellectual property from the U.S. parent to manufacture concentrate for Coca-Cola beverages in overseas markets. For the 2007–2009 tax years, the IRS made transfer-pricing adjustments that increased Coca-Cola's aggregate taxable income by more than US$9 billion and resulted in tax deficiencies exceeding US$3.3 billion before subsequent adjustments. The U.S. Tax Court sustained the core IRS transfer-pricing position in its 2020 decision. (See: CiteLaw: The Coca-Cola Company and Subsidiaries ↗ (opens in new tab))

The dispute is particularly instructive because it involves both the selection of a transfer-pricing method and the reliability of the evidence used to support the resulting allocation of income.

The Tax Court upheld the IRS's use of CPM, with the foreign supply points treated as the tested parties and independent bottlers used as comparables. The court concluded that the supply points' returns under Coca-Cola's historical methodology were substantially higher than could be supported by the relevant comparable evidence.

Professor Rosenbloom also highlighted the importance of the factual record. Among the issues discussed were the distinction between Coca-Cola's supply points and service companies, the allocation of marketing-related expenses, the economic relationship between Coca-Cola and its bottlers, and the treatment of foreign intangibles.

The case also raises a separate procedural question: to what extent can a taxpayer rely on a transfer-pricing methodology that had previously been accepted or agreed with the IRS for earlier years?

The 1996 closing agreement concerning the earlier tax years did not, according to the Tax Court, prevent the IRS from making subsequent transfer-pricing adjustments for 2007–2009. The case therefore illustrates an important distinction between historical administrative treatment and a legally binding determination of the appropriate transfer-pricing outcome for later years.

As of 1 September 2026, the dispute remains pending before the U.S. Court of Appeals for the Eleventh Circuit. The court's official case page records oral argument on 25 June 2026 in The Coca-Cola Company and Subsidiaries v. Commissioner of Internal Revenue, No. 24-13470. No appellate judgment was located in the court's publicly available case information reviewed for this article.

The continuing appeal makes Coca-Cola particularly relevant to the session's wider discussion. It brings together the choice of transfer-pricing method, comparability, intangible value, evidentiary questions, administrative consistency and the limits of taxpayer reliance.

The U.S. statutory framework also matters

One of the session's recurring themes was that U.S. transfer pricing cannot be understood solely through the phrase “arm's length”.

The principal U.S. statutory provision is Internal Revenue Code section 482, which authorises the Secretary of the Treasury to allocate income, deductions, credits and allowances among commonly controlled taxpayers where necessary to prevent tax evasion or clearly reflect income. The statute also contains specific rules concerning transfers and licences of intangible property, including the commensurate-with-income principle and realistic-alternatives valuation. (See: Legal Information Institute: U.S. Code 482 ↗ (opens in new tab))

The arm's-length standard is elaborated through the Treasury regulations under section 482. The regulations provide the general framework for determining taxable income from controlled transactions and include the best method rule, comparability considerations and specific transfer-pricing methods.

This statutory-regulatory distinction was important to the session's analysis because it helps explain why U.S. transfer pricing can look both familiar and different from the OECD framework.

It also reinforces a broader lesson from the discussion: transfer pricing outcomes are shaped not only by economic concepts but by the institutional and legal architecture through which those concepts are administered.

A Singapore perspective: certainty depends on both rules and administration

For Singapore, the session's U.S.-focused discussion provides a useful comparative perspective rather than a prescription for Singapore's own transfer-pricing regime.

IRAS expressly endorses the arm's-length principle and recommends a three-step approach: conduct a comparability analysis, identify the most appropriate transfer-pricing method and tested party, and determine the arm's-length result. Taxpayers are also required, where applicable, to prepare and maintain contemporaneous transfer-pricing documentation. (See: IRAS: Transfer Pricing ↗ (opens in new tab))

The comparison is particularly relevant for multinational businesses operating through Singapore. Transfer pricing requires practitioners to understand not only the transaction and the appropriate methodology, but also the administrative environment in which that methodology will ultimately be tested.

The session also ended with Professor Rosenbloom raising a possible “synthetic APA” concept for jurisdictions that have an exchange-of-information arrangement with the United States but do not have a comprehensive income-tax treaty with it. He described this as an idea under which a taxpayer could propose a transfer-pricing methodology to the two administrations, with information exchange potentially facilitating communication between the authorities.

This was presented by Professor Rosenbloom as his own proposal for consideration, rather than as an established mechanism or Tax Academy position. Singapore and the United States do have a Tax Information Exchange Agreement, which entered into force on 5 March 2020, and Singapore's agreement permits the exchange of information relevant to the administration and enforcement of covered taxes.

Whether the concept could operate within the existing legal and administrative framework is a separate question requiring consideration by the relevant authorities. It nevertheless illustrates the broader point raised throughout the session: transfer-pricing certainty depends not only on methodology, but also on the mechanisms available to prevent and resolve disputes across jurisdictions.

What the discussion reveals about transfer pricing practice

Session 3 of the Masterclass with Professor David Rosenbloom and Professor Dr René Matteotti ↗ (opens in new tab) ultimately presented U.S. transfer pricing as a field where economic theory, statutory authority, administrative practice and litigation interact continuously.

The arm's-length principle remains the organising concept. But its practical application requires judgment where perfect comparables are unavailable. CPM illustrates one way of dealing with that difficulty. Cost-sharing arrangements demonstrate the challenges of allocating the costs, risks and benefits associated with intangibles. Secondary adjustments show that a primary transfer-pricing adjustment can generate consequences beyond the original price. And Coca-Cola demonstrates how methodology, evidence, administrative practice and litigation strategy can converge in a dispute involving billions of dollars.

For tax professionals, the wider lesson is not that one transfer-pricing method is universally preferable. Rather, it is that a defensible outcome depends on the coherence of the entire analytical chain: the facts, the transaction, the functions and risks, the comparables, the methodology, the evidence and the legal framework.

That perspective is consistent with Tax Academy of Singapore's broader role in developing tax capability and facilitating knowledge exchange between government, practice, business and academia. Tax Academy describes its mission as elevating tax literacy and building a technically strong, internationally connected and future-ready tax community.

In that sense, the value of examining U.S. transfer pricing from Singapore is not simply to understand U.S. rules. It is to sharpen the ability to recognise how different legal systems approach the same fundamental problem: how should income be allocated when related parties operate across borders?

Professor David Rosenbloom conducting Session 3 of the Masterclass

Continue Learning on TaxWatch

The Masterclass with Professor David Rosenbloom and Professor Dr René Matteotti ↗ (opens in new tab) continues through 2 September 2026. Registration remains available through TaxWatch ↗ (opens in new tab) until 5 September 2026.

Readers who register during the eligible period can join applicable remaining sessions live, catch up on eligible completed sessions through available recordings, and retain access to eligible recordings and notes until 31 December 2026, subject to the applicable access arrangements.