Tax Notes logo

A Major Simplification of the OECD’s Pillar 1 Proposal

Posted on Jan. 11, 2021
Michael J. Graetz
Michael J. Graetz

Michael J. Graetz is a professor of law at Columbia Law School and professor emeritus at Yale Law School. He formerly was assistant to the secretary and special counsel at Treasury. His most recent book (with Ian Shapiro) is The Wolf at the Door: The Menace of Economic Insecurity and How to Fight It. He thanks Walter Hellerstein, Peter Merrill, William Morris, Paul Oosterhuis, and Alvin Warren for comments. They, of course, are not responsible for anything said here.

In this report, Graetz suggests major modifications to the OECD’s pillar 1 blueprint proposal to create a new taxing right for multinational digital income and some product sales that would greatly simplify the proposal. The modifications rely on readily available existing financial information and would achieve certainty in the application of pillar 1, while adhering to its fundamental structure and policies.

Copyright 2021 Michael J. Graetz.
All rights reserved.

In October 2020 the OECD released nearly 500 pages of blueprints on its pillars 1 and 2 projects on the tax challenges arising from digitalization. Pillar 2 represents an effort to create a widely accepted global minimum tax. I address only pillar 1 here.

The blueprint for pillar 1 is the result of two years of admirable, herculean efforts by the OECD, with the support of the G-20 and the 137 countries of the inclusive framework, to create nexus and profit allocation rules more appropriate to the 21st-century economy. This endeavor, which carries on from the OECD’s base erosion and profit-shifting work, attempts to prevent the proliferation of widespread and divergent laws — often based on gross receipts rather than profits — to collect taxes where sales or data from resident users of digital services play an important role in generating profits for a multinational enterprise without those profits being subject to income tax in that jurisdiction.1

Unfortunately, the basic principles undergirding pillar 1 are often buried in the OECD’s complex 230-page blueprint. Almost all of the many pages of comments on pillar 1 submitted in December 2020 address technical aspects of the blueprint. Many complain — appropriately in my view — that the novel and complex concepts and rules in the pillar 1 blueprint will require new and contestable calculations not now being produced either for tax or financial accounting purposes.2 Others complain about the arbitrariness of outcomes under pillar 1 and express concerns about the potential for double taxation. Many express distress that, in its current form, pillar 1 will inevitably produce high compliance costs, uncertain outcomes, and multiple and conflicting national claims for revenues. The capacity of many developing countries to administer the pillar 1 blueprint is also uncertain.

Importantly, the multitude and complexity of ambiguities and potential conflicts undermine the OECD’s insistence that it is essential to ensure certainty in results (to be facilitated by new dispute resolution mechanisms).3 They threaten the viability of pillar 1 (and perhaps also pillar 2 because some countries, such as the United Kingdom, take the view that if pillar 1 fails, so should pillar 2). These problems with the pillar 1 blueprint cannot be fixed through mere tinkering.

I will describe here the foundational principles and basic policy decisions underlying the pillar 1 blueprint and offer a much simpler and more certain alternative that relies on information currently being produced for other reasons. In doing so, I generally accept the principles and many of the policy decisions reflected in the blueprint. Much of the blueprint’s structure, however, contributes substantially to its astounding complexities without being required by the principles and policies that motivate the fundamental changes pillar 1 proposes. I begin with the fundamental principles of the blueprint.

I. The Fundamental Principles of Pillar 1

A. Taxable Profits Without a Physical Presence

Nearly four decades ago, the U.S. Supreme Court said that “allocating income among various taxing jurisdictions bears some resemblance . . . to slicing a shadow.”4 Two decades later, I pointed out that MNEs’ methods of doing business had undermined the basic rules for determining the source of various categories of income.5 More recently, in making the obvious point that current source rules were developed long ago for a very different world economy and need updating to limit tax avoidance techniques involving intellectual property, Rachael Doud and I urged an allocation of some revenues to countries of sales to limit profit shifting and achieve greater alignment between a nation’s sales and its taxable income.6 Many others agree.7

Pillar 1 is based on a judgment by the inclusive framework that a country that produces for an MNE a substantial amount of sales or revenue derived from the monetization of data supplied by domestic users of digital services is entitled to an allocation of some profits, whether or not the MNE has any physical presence within that country’s borders.

Formulary appointment of some share of profits to the jurisdiction where sales occur has long been a feature of the subnational divisions of business income taxes in Canada, Switzerland, and the United States, and has been suggested as a mechanism for the allocation of revenues under the proposed common consolidated corporate tax base proposal in Europe.8 In 2018 the U.S. Supreme Court properly concluded that limiting income taxation based on sales to businesses that have a physical presence is now unsound and each year “becomes further removed from economic reality.”9

Under pillar 1 some portion of profits — labeled amount A — would be allocated to the market/user country. This new taxing right would not fully replace the current allocation of profits to countries where income is produced under existing source rules, arm’s-length pricing allocations, and income tax treaties. For the market country to receive an allocation of a portion of profits under pillar 1, a threshold amount of sales or other revenues, such as advertising revenues, must occur in (or be attributable to) that country.10

It is important that the OECD succeed in its effort to achieve widespread multilateral agreement to allocate revenues from some portion of MNE profits to countries with substantial amounts of sales or revenues derived from the monetization of data derived from digital services. Businesses or countries that view the status quo as an alternative to pillar 1 are engaging in an unrealistic and futile comparison. By my count, there are now at least 36 countries, plus the European Union, that have announced or enacted digital services taxes or withholding taxes on gross revenues from sales or specified offshore digital services or an equalization levy based on domestic sales or residents’ use of digital services within their borders. Over time, given growing frustrations with the existing allocations of revenues and the increasing role of digital products and services, the variety of unilateral measures will undoubtedly expand and existing ones will mutate, increasing their scope and reach.

Typically, these laws layer additional taxes on top of existing income taxes and VAT. Some are based on gross revenues (which operate like special additional VAT rates or ad valorem excises) and others on a share of profits. Efforts by MNEs to obtain credits for such taxes from their headquarters or parent country will be of little avail. And if the U.S. response to France’s digital services tax is a harbinger, retaliatory tariffs and trade disputes may follow the imposition of such taxes. Chaotic unilateral measures and unpredictable counter measures are therefore the baseline against which the OECD’s pillar 1 effort must be evaluated.

B. Avoid Double Taxation

For over a century, avoiding double taxation of income has been a goal of taxpayers, countries, and the OECD in multilateral and bilateral income tax treaty arrangements.11 But double taxation is omnipresent: Classical corporate income taxation taxes business income twice; wages are subject to payroll and income taxes when earned and then to sales taxes, VAT, or excise taxes when spent. The OECD policy decision to avoid double taxation should be understood to mean that — unlike the unilateral digital taxes — an allocation of profits to market/user countries under pillar 1 should be in lieu of income taxes allocated to other source or residence countries rather than in addition to those taxes. Accordingly, the goal of pillar 1 should be viewed as updating and modifying the source rules to reallocate some revenue, rather than creating a new tax.

This means that the allocation of a portion of profits to a market/user country will increase an MNE’s taxes only when the tax rate applied in the market jurisdiction is higher than in the jurisdiction with its current allocation of revenues reduced as an offset. This seems appropriate even though it requires answering the difficult question of which countries’ revenues will be reduced to offset amount A’s allocation of profits to market/user countries. The blueprint’s effort to identify the entities whose profits will be reduced to offset reallocations of amounts A is contained in a complex and controversial chapter.12 I address that issue in Section III.

C. Formulary Allocation

The OECD properly describes the allocation of amount A to the market/user countries as based on a formula rather than on arm’s-length pricing methods. Amount A is calculated by allocating to market countries a fixed percentage (described as 10 to 30 percent, but presumably 20 percent) of residual profits calculated alternatively based on a profit amount or a profit margin approach. In either case, the result is equivalent to a formulary apportionment of a portion of profits based on the ratio of local sales revenue to worldwide sales revenue — a familiar method analogous to the sales-based formulary apportionments used by U.S. states.13

Combining arm’s-length pricing and a formulary allocation is inherently complex and may be difficult to justify analytically. But as a prudent political matter it is hardly surprising that the OECD decided to move only incrementally away from current practices — and to limit this move to large MNEs with business activities sufficient to make them vulnerable to the unilateral market-based and digital taxes springing up around the world.

II. Establishing the Scope of Allocation

The pillar 1 blueprint circumscribes allocations to market/user countries along several dimensions. First, the allocation applies only to consumer-facing businesses or automated digital services (ADS) above specified thresholds of worldwide sales or revenues and with a substantial amount of business activity in the user jurisdiction.14 Second, the portion of worldwide profits allocated to market/user countries is based on an MNE’s worldwide book (rather than taxable) income. Third, amount A to be allocated to the market country is a fixed percentage of residual profits.

A. Determining In-Scope MNEs

The OECD limits the allocation of amount A profits to consumer-facing businesses and ADS. A 14-page chapter of the blueprint describes businesses it would include and exclude.

Including MNEs with consumer-facing businesses in the scope of the proposal is justified or the ground that these are the kinds of companies for which the existence of a market is essential to the production of income. The inclusion of ADS businesses that monetize data provided by online services is necessary if pillar 1 is to serve as an effective substitute for the unilateral digital taxes now proliferating.15 A formulary allocation of some portion of profits to user countries based on revenues from advertising would capture some of the profits from data digitization, but this is not the only way in which users’ data is monetized. Indeed, ongoing expansions of the internet of things means that the ways that users’ data can now be monetized are proliferating across the economy in contexts and methods not captured by the blueprint’s determinations of in-scope business activities.

Distinguishing business activities in scope from those out of scope necessarily requires difficult and contestable line drawing. Unfortunately, this is only the first step. The blueprint also requires many additional distinctions and segmentations for in-scope businesses in determining allocations of amounts A and in computing offsets of paying entities in other jurisdictions to prevent double taxation.

Subnational jurisdictions in the United States do not limit the application of their sales-based formulary apportionment to consumer-facing or ADS businesses as the blueprint does. Expanding formulary apportionment of a portion of profits to business activities that are currently out of scope of the OECD blueprint — perhaps with limited exceptions, such as for natural resources where profits are location-specific — would eliminate distinctions among lines of business that cause complexity in the application of pillar 1.16 However, extending amount A substantially beyond the current scope of the blueprint seems unrealistic.17 I therefore do not reevaluate the blueprint’s choices here, but instead endeavor to eliminate the additional distinctions and segmentations the blueprint requires.

The amount A allocation is also limited to large MNEs — defined by global revenue — that have more than a threshold amount of in-scope revenue in the market/user country.18 The blueprint estimates that a global revenue threshold of €750 million to €5 billion would limit application of the pillar 1 formulary allocation to 2,300 to 620 MNEs, respectively. A threshold based on gross revenues to exclude small and medium-size enterprises is obviously sensible to limit compliance and administrative costs for taxpayers and governments and is taken up in Section II.D.

B. Using Worldwide Book Measures of Income

In nations such as the United States that use formulary apportionment of some portion of profits to subnational jurisdictions, the apportionment of income to sales is based, at least initially, on a uniform definition of taxable income at the national level. In contrast, the OECD effort does not start with the advantage of any uniform worldwide calculation of taxable income. Instead, the blueprint uses consolidated financial accounts based on international financial reporting standards (or in some cases generally accepted accounting principles) with a limited number of prescribed adjustments as its starting point in determining amount A.19

I do not question here the decision to use book profits.20 I do, however, urge using existing financial reporting information without requiring companies to create new and contestable segmentations of that information. Financial reports should also be used to exclude companies with less than a specified amount of in-scope revenues.

C. Allocating a Portion of Residual Profits

In reallocating amount A to market/user countries, the OECD allocates a fixed percentage, presumably 20 percent, of the excess of profits over a specified threshold of profits — defined as the excess of profits over some fixed percentage of sales or service revenues (which the blueprint frequently describes as 10 percent). It labels this amount “residual profits.” The two-step process is necessary because pillar 1 allocates only a portion of residual profits to market/user countries.

When an MNE has in-scope consumer-facing business and/or ADS revenues above the thresholds and also has substantial profits from businesses that are out of scope, a division of profits is necessary. The necessary separations between in- and out-of-scope activities should be based on disclosures routinely made in financial statements.21 However, the method for allocating only residual profits to market/user countries requires numerous additional segmentations when profit margins differ significantly across different lines of business.

Consider, for example, Amazon, whose activities are either consumer-facing or ADS businesses, and is obviously intended to be in-scope. In some recent years, Amazon has earned no residual profits before taxes. Martin A. Sullivan has estimated that Amazon’s pretax profit in 2017 was 2 percent.22 But Amazon’s low overall profit margin reflects a mix of low-profit business (from physical stores and in some cases online sales) and higher-profit businesses (from cloud computing, some third-party seller services, and various subscription services).

To take just one other example clearly intended to be in scope, consider Apple, which has large residual profits (28 percent pretax profits in 2017).23 Apple’s growth in profits and sales has historically been driven by products: the iPhone (which accounted for half or more of total sales revenue); the Mac computer (which accounted for just under 10 percent); the iPad (which accounted for about 8 percent); wearables, like AirPods and the Apple Watch, and home accessories (which together accounted for about 9 percent). In recent years, however, the company’s revenue and especially its profitability growth have slowed from its products and instead have come mostly from digital services, including subscriptions to Apple Music and apps, AppleCare warranties, and iCloud storage services.24 It is uncertain under the OECD’s segmentation requirements to allocate residual profits whether Apple has two lines of business — products and services — or at least 11, including iPhones, Mac computers, iPads, AirPods, Apple Watches, home accessories, Apple Music, Apple TV, the App Store, AppleCare warranties, and iCloud services. The answer may turn — but not necessarily — on how Apple divides its lines of business for financial reporting businesses.25 Profitability — and therefore residual profits — in various jurisdictions also differs among the 11 lines of business.

Segmenting lines of business to determine which have residual profits in excess of the threshold (presumably 10 percent) requires allocating both costs and revenues to different lines of business. Some companies already do this for financial reporting or transfer pricing purposes, but others do not. The OECD is apparently concerned that allocating only residual profits without segmenting profit margins by in-scope lines of business would exclude companies that do not meet the residual-profit threshold for all their in-scope lines of business. The Amazon example suggests (for 2017 at least) that the company does not have profits in excess of the 10 percent routine profit threshold and therefore might be excluded despite its large residual profits — for example, from cloud computing.26

As the examples illustrate, there are numerous ways to segment businesses in determining the amount A residual profit allocations. The blueprint describes a hierarchy that applies to each type of in-source revenue.27 It also says that further segmentation might be required to account for different degrees of digitization among in-scope business activities or for variations in profitability among different market segments. It calls these “digital differentiation” and “jurisdictional segmentation.”28

The blueprint also discusses various modifications of its formulas for determining residual profits and even profitability thresholds, called “differentiation mechanisms,” which are intended to take into account different degrees of digitalization among various in-scope business activities and variations in determining residual profits of in-scope MNEs in different market/user jurisdictions.29

If each jurisdiction decides what profitability threshold and reallocation percentage to use and then whether to apply a differentiation mechanism, chaos will reign. If these kinds of differences among nations are permitted, pillar 1 will not ameliorate the kinds of unilateral measures to tax digital services and market-based profits that have stimulated the OECD’s efforts. A high degree of uniformity seems essential. Moreover, the potential variations seem rather arbitrary and undermine the rationale for using complex determinations of residual profits with their inevitably controversial allocations of costs based on segmentations that serve only to create illusions of precision where there is none.30

The blueprint recognizes that large compliance costs and administrative burdens will occur as a result of the way it intends to measure residual profits for the segmentations it would require beyond the necessary distinctions between in- and out-of-scope businesses. Disputes and controversies over segmentations and the allocations of revenues and costs to them are inevitable. To limit these costs and controversies, the pillar 1 blueprint indicates its willingness to rely on multinational group financial statements — as long as the financial statements meet “agreed hallmarks.”

When countries see their results, based on an MNE’s financial statement segmentations, they might want to demand changes, but there is no reason to expect all countries to want the same changes. And when MNEs see the results under their existing financial accounting practices, they might try to aggregate or disaggregate products or services into different segmentations. After all, the prices of their securities turn on their overall profitability, not necessarily how they segment their lines of business in their financial statements.

The OECD’s decision to reallocate amount A to market/user jurisdictions based only on a specified percentage of residual profits, rather than reallocating a smaller fixed percentage of all of the profits of a MNE’s in-scope activities, introduces serious complexities and uncertainties into the pillar 1 blueprint contrary to the OECD’s efforts to achieve certainty and reduce complexity.31 It makes the blueprint’s claim that “securing tax certainty is an essential element of Pillar 1” implausible.32

The OECD’s decision to allocate a portion of only residual profits may have evolved from its desire to maintain and move only incrementally from existing profit splits that sometimes entitle market countries to a share of revenues from residual profits under the arm’s-length method.33 It also limits the blueprint’s departure from existing permanent establishment rules to situations in which residual profits are present. So the limitation to residual profits may simply be one way in which the OECD appears to be hewing closely to existing practices. But the costs of doing so in the manner that the blueprint requires are large and the justifications small or nonexistent.

Limiting amount A to residual profits reflects a decision that market/user countries are entitled only to a share of entrepreneurial and other synergies that produce profits for an MNE in excess of those that could be produced by the MNE’s related parties operating independently. Under arm’s-length pricing, the allocation of entrepreneurial and synergistic profits requires an intense factual case-by-case inquiry for MNEs with difficult-to-value intangible assets (including both production and marketing intangibles) that contribute to residual profits.

In transfer pricing parlance, routine profits are the profits that a third party would earn in performing activities conducted by a related party that does not share in the overall risks and synergies of the MNE.34 Those routine profits are assumed to arise in the country where the activities take place. But the distinction between routine and residual profits of related enterprises and the allocation to particular entities and countries under arm’s-length principles is hardly straightforward, as transfer pricing controversies amply demonstrate.35

Some authors have argued for an allocation of all residual profits to market jurisdictions based on each country’s proportion of worldwide sales or, alternatively, after allocating all routine profits to the jurisdictions where employees or physical capital are deployed.36 But — even if residual profits were defined in a manner to reflect profits attributable to synergies of an MNE that do not have any particular location — allocating all profits in excess of a routine return to market/user countries would shortchange countries that produce residual profits through production intangibles, headquarters activities, and research and development. If it were practical to separate out the residual profits attributable to marketing intangibles and allocate those — or a substantial portion of those — to market countries, that could be defended conceptually.37 But recognizing this is impractical, the OECD resorted to a formulary method for determining the allocable amount A, along with an effort to ensure the allocation of some routine profits to marketing and distributional activities (which it calls amount B).38

Limiting amount A reallocations to residual profits may reflect a judgment by the OECD that allocating residual profits to the market/user countries will reduce tax planning opportunities that now enable MNEs to shift income from IP (or financial capital) to low- or zero-tax countries.39 In those terms, pillar 1 could be viewed as an extension of the OECD’s BEPS efforts, but, as Section III shows, limiting amount A to residual profits is not necessary for that.40

The OECD’s pillar 1 blueprint endeavors: (1) not to duplicate residual profits currently allocated to market countries under profit splits; (2) to determine when market countries already receive sufficient revenues under existing profit-split methods; (3) to determine residual profits on a segmented line-of-business basis often not required for either tax or financial reporting purposes; and (4) to develop (inevitably controversial) rules regarding the identification of paying entities and countries that will give up the revenue allocated to market/user countries.41

If providing a market and users who supply data that produce profits for an MNE entitles a country to a share of revenues from those profits, why is it sound to limit such claims to complex arbitrary computations of residual profits? The answer must be grounded in existing practices and the OECD’s desire for only incremental change. But, as I have described, the status quo is not stable. Market/user countries are taxing, and will continue to tax, revenues beyond those to which they are entitled under current international income tax nexus and transfer pricing rules — and even when there are no profits. In limiting amount A to its calculations of residual profits by lines of business, the OECD has produced a convoluted and controversial structure that in the absence of universal agreement on the application of its segmentation decisions — which certainly will not happen — guarantees the kinds of uncertainties, complexities, and controversies that the OECD insists it wants to avoid.

It would be far simpler, and at least as defensible in principle, to allocate a portion of all profits (and losses) of in-scope companies to market/user countries.42 This, of course, is what happens in formulary apportionments of business income to U.S. states. And it is what the G-24 working group of developing countries has requested.43

The OECD estimates that its reallocation of amount A residual profits to market jurisdictions by in-scope MNEs above a €750 million threshold would shift less than 1 percent of global corporate income tax revenues from low- to higher tax-rate jurisdictions.44 The estimates imply that a percentage allocation to market/user jurisdictions of all such profits of about 10 percent would produce a revenue reallocation similar to that estimated by the OECD with a 20 percent allocation of residual profits and a 10 percent profitability threshold.45

Formulary allocation of some percentage of all profits — with the percentage set to produce revenues in market/user countries roughly comparable to those estimated to be allocated under amount A of the pillar 1 blueprint — would effectuate a dramatic simplification of pillar 1. If a fixed percentage of all profits were allocated to such countries, segmentation would be limited to the distinction between in- and out-of-scope activities. In essence, this rule would require a formulary allocation based on sales or revenues of consumer-facing or ADS businesses of a specified percentage of the net profits (and losses) of MNEs within the scope of pillar 1.46 The remainder of the MNE’s profits would be allocated under existing arm’s-length principles.

D. Selecting Thresholds and Percentages

Determining what portion of profit to allocate to market/user countries is necessarily a political exercise. As I have described, under the blueprint, the residual profit of an MNE is based on the excess of the MNE’s worldwide profit over a profitability threshold.47 The potential profitability thresholds in the blueprint range from 8 to 25 percent. The blueprint also describes a range of reallocation percentages of residual profits to be allocated as amount A that range from 10 to 30 percent. The OECD estimates of the total global residual profits allocable to market jurisdictions vary based on 15 different combinations of reallocation percentages and profitability thresholds. These produce estimated total revenues allocable to market jurisdictions under amount A that range from $10 billion to $147 billion.48 Obviously, many more possible variations exist. As one specific example, the blueprint observes that a 10 percent profitability threshold and a 20 percent reallocation percentage would allocate just under $100 billion to market/user jurisdictions.49 As I have indicated, if all profits (rather than residual profits) were allocated to user jurisdictions, the allocation percentage could be reduced by about half.

Even though it is based on 2016-2017 data and has other data limitations, the OECD’s 2020 analysis of the economic impact of pillar 1 sheds additional light on the potential for political acceptance of a fixed percentage allocation of profits to market/user jurisdictions.50 The report estimates a total of $415 billion to $493 billion in residual profits for in-scope MNEs (as defined in the pillar 1 blueprint) with gross revenue thresholds ranging from €750 million to €5 billion.51 With a profitability threshold of 10 percent, the difference between a threshold of €750 million (which the OECD blueprint emphasizes) and €2 billion is only €27 billion, about 5.5 percent of the residual profits at the €750 million level.52 This suggests that a higher threshold of up to €2 billion might produce substantial simplification by eliminating many MNEs from the scope of pillar 1 at a relatively small cost in revenues to market jurisdictions.

Including all in-scope profits, rather than just residual profits, in the allocation of amount A to market/user jurisdictions, as suggested here, also allows greater potential simplification with relatively little revenue cost from moving the threshold up from €750 million to €2 billion, or perhaps even higher.

III. Offsetting Amount A

As I have described, the OECD concluded that allocating amount A to market/user countries should not cause a tax on in-scope MNEs’ profits in more than one country and would require additional taxes from MNEs only when profits are allocated from lower- to higher-tax countries. So the blueprint requires that the amount A allocation of some profits to market countries needs to be offset by a reduction of taxes currently allocated elsewhere under arm’s-length transfer pricing. This makes it necessary to specify which entities in what jurisdictions will have their taxes reduced, or as the OECD puts it, to specify the paying entities. The blueprint details its method for such offsets in a complex four-stage process that takes 14 pages to describe and reaches results that are often ambiguous and will inevitably be controversial.

Identifying the paying entities under the pillar 1 blueprint requires several steps. First, each paying entity must have residual profits (as before, a profit margin in excess of a profitability threshold, such as 10 percent of revenues). This would eliminate several MNE’s entities as potential paying entities. The blueprint then requires further inquiries or tests to identify an entity’s contributions to the residual profits allocated under amount A. There is also a priority test based on an entity’s connection to the market jurisdiction, which turns on the activities an entity performs to make “a material and sustained contribution to [the] MNE’s residual profits.”53 Other amounts would be allocated pro rata.

The blueprint also suggests using an exemption method and/or a foreign tax credit method to eliminate double taxation, even though simply exempting an entity’s portion of amount A from its taxable income would be far simpler. To achieve tax certainty, the computation of the exempt (or creditable) amount must be approved by all the relevant jurisdictions. If they fail to agree, binding arbitration by a determination panel follows.54

Under these complex determinations, uncontroversial, predictable, and certain outcomes are a mirage. Offsets for amount A should instead be based on a pro rata reduction of profits taken from readily available financial information. As under the blueprint, the paying entities should be limited to those that earn residual returns, even though a portion of all in-scope profits would be allocated to market/user countries under the regime I have proposed here.

Limiting offsets by paying entities to residual returns is appropriate because mobile residual profits from IP (and in some cases financial transactions) are often shifted to lower- or zero-tax jurisdictions. Related entities in some jurisdictions are allocated only a relatively small cost plus markup on their activities under current transfer pricing rules, and it does not seem proper to offset amount A by reducing profits now taxed in jurisdictions where only such profits are earned. The OECD’s economic assessment of pillar 1 concludes that most of the offsets under its method for determining paying entities come from income currently allocated to affiliates in investment hubs.55

The blueprint’s method of limiting offsets to residual profits of paying entities only to lines of business closely linked to amount A again introduces unnecessary complexities and the potential for controversies. Instead, a pro rata offset to determine the paying entities of an in-scope MNE, excluding only entities that earn a routine return (based on a specified profit threshold) would be simpler and minimize controversies. This would also require offsets from financial hubs in low- or zero-tax jurisdictions that are allocated more than routine profits.

Because the fixed percentage of profits to be allocated to market/user countries would be based on worldwide income on financial reports, it is appropriate to use a financial reporting measure of paying entities’ profits (rather than taxable income) to determine the paying entities’ offsets. Unfortunately, however, the consolidated worldwide reports used for determining the total profits allocated to market countries are not apt: They do not show entity-by-entity profits in particular jurisdictions, which is the information necessary to determine paying entities’ offsets. Using existing sources of data, therefore, requires basing these offsets on country-by-country reports.56 Using a pro rata reduction of entities’ residual profits (profits above a specified threshold) based on an in-scope MNE’s CbC reporting of profits would greatly narrow the scope of the potential controversies. There would be no need to determine whether the paying entities themselves are in scope or to link their activities to those of particular market or user jurisdictions. Allocating costs to particular entities for determining their share of residual profits also would not be necessary as it is under the blueprint. Costs are already taken into account in entities’ CbC submissions. Profits are stated in CbC reports for all countries where there is a legal entity (often a subsidiary but sometimes a partnership) or a branch regarded as a PE within a jurisdiction, as are profits of any constituent entity that is transparent for tax purposes in its local jurisdiction.57 One reason the OECD offers for rejecting the use of consolidated financial statements to determine offsets is their elimination of intercompany transactions, which the OECD regards as inappropriate for this purpose. These transactions are not eliminated in CbC reports. Beginning in December 2020, however, CbC reporting eliminates intercompany dividends, an adjustment that is appropriate here. Stateless income, which is double counted in a parent company’s CbC reports, should also be eliminated here.

CbC reports are not perfect sources of financial information for this purpose. CbC reports are generally based on financial statements for financial reporting or regulatory purposes — which is consistent with the use of book profits, rather than taxable income, to determine the amount of profit allocated to market/user countries — but taxpayers have flexibility to use records developed for internal management or tax reporting purposes for CbC reports. That flexibility may prompt some gamesmanship by MNEs. But because the total amount of the reductions of income in various entities must add up to 100 percent of amounts A allocated to market countries, the incentives for such games should be relatively small. My approach for determining offsets of paying entities might also create an impetus to require more uniformity in CbC reports based on IFRS (or GAAP where applicable) financial reporting.

IV. An Alternative Proposal

In summary, this is how the system proposed here would work. First, isolate all MNEs above a specified threshold of sales or revenues that are engaged in consumer-facing business or ADS and therefore are in scope. Allocate a fixed percentage of the MNE’s worldwide book profits to market/user countries, based on each country’s percentage of sales or revenues (amount A). Next, offset the taxable income of the MNE’s affiliated entities by the total of amounts A allocated to market countries by reducing the entities’ local pro rata shares of CbC positive residual profits (profits above a specified threshold) to the MNE’s total worldwide CbC positive profits (with stateless income reported by parent entities eliminated). All other taxable income would continue to be determined under existing transfer pricing principles.

Assuming an MNE meets the basic revenue threshold and is in scope, its tax base in a particular country would be increased by the following calculation:

(C * f)(S/W - P/T)

where C equals worldwide in-scope profits from financial statements;

f is the fixed percentage of all in-scope profits to be allocated to market/user countries;

S is local sales and/or revenues from ADS;

W is worldwide sales and/or revenue from ADS (the sum of all Ss);

P is positive local profits above a specified threshold of profits of in-scope entities based on CbC reporting; and

T is the total worldwide sum of all Ps.58

This calculation follows the fundamental policies and basic structure of the pillar 1 blueprint’s allocation of a formulary portion of in-scope MNE profits to market/user countries while eliminating double taxation.59 But it is dramatically simpler and will produce clear and certain outcomes, which the pillar 1 blueprint fails to do.

V. Achieving Tax Certainty

The OECD insists in its pillar 1 blueprint that achieving tax certainty is essential.60 It endeavors to ensure certainty by adopting new mandatory dispute prevention procedures and dispute resolution institutions. These are commendable efforts, but the difficulty is that the blueprint’s complex methods of determining amount A and offsets from paying entities open too many potential avenues for controversies. It is virtually impossible to have confidence that any dispute resolution process can achieve predictable and certain results. Here is how the blueprint describes potential controversies:

Such disputes could concern, for example, the correct delineation of business lines, allocation of central costs and tax losses to business lines, the existence of a nexus in a particular jurisdiction, or the identification of the relieving jurisdictions for purposes of elimination double taxation.61

The simplifications described here would substantially narrow the potential controversies. Lines would still need to be drawn to determine which business activities are in scope and which are out of scope. But the amount to be allocated to the market/user countries would simply be a fixed percentage of all in-scope MNE profits based on financial reporting. The amount so allocated would then be offset by a reduction of the MNE’s entities’ taxable income in their local jurisdictions based on the entities’ pro rata allocation of positive profits in excess of a specified threshold, based on data readily available from CbC reporting.

These major simplifications of pillar 1 would eliminate most of the issues and controversies that would arise under the blueprint. Successful implementation of the mandatory dispute resolution procedures of the sort set forth in chapter 9 of the blueprint could then succeed in resolving controversies. Importantly, the prospects of achieving the essential OECD goal of ensuring tax certainty would also become realistic.

VI. Conclusion

No one seems happy with the OECD’s pillar 1 blueprint. Borrowing from Tolstoy, much unhappiness is being expressed in its own way. Some simply want to torpedo the project. The position of the United States, expressed in Treasury Secretary Steven Mnuchin’s insistence that the pillar 1 allocation to market countries not be mandatory, but instead some sort of a safe-harbor regime, seems to fall into that category.62 The secretary’s position, however, did not halt the project, and it allowed Treasury representatives to participate in OECD discussions of the technical issues. But Mnuchin might just as well have said that the United States opposes the pillar 1 effort.

Many other objections to the pillar 1 blueprint, however, are well justified. Businesses that will have to comply with pillar 1 are properly concerned with its complexities and attendant compliance costs; the ambiguities in its application, which threaten multiple and overlapping claims to revenue; and the practical challenges of achieving the level of certainty in its application that the blueprint promises. Many countries rightly worry about their ability to administer the rules described in the blueprint.

The incoming Biden administration will no doubt reevaluate the U.S. position because, as I have said, it is foolish to think that the status quo predating these OECD efforts is an appropriate baseline for comparison. Instead, market/user jurisdictions will continue to develop and impose new and varying taxes based on the use of digital services and sales of goods and services in their jurisdictions. Pillar 1 is intended to substitute a multilateral agreement on the allocation of some profits to those jurisdictions for the many different unilateral regimes that are emerging and will emerge — and for the retaliatory tariffs and other trade measures that have been promised and seem likely to occur in response.

The 230 pages of details in the pillar 1 blueprint obscure its underlying principles. Of course, it is difficult to discern clear principles in a structure that uses arm’s-length transfer pricing to allocate the great majority of profits and a formulary method of apportionment to allocate a limited portion of residual profits to market/user countries. Nor is there an obvious principle that yields clear answers to the question of where the offsets to the reallocation of amount A to eliminate double taxation should come from. The mechanisms in the OECD’s pillar 1 blueprint to allocate some relatively small portion of residual profits to user countries produce great complexity without eliminating transfer pricing disputes or concerns with income shifting of proponents of allocating all residual profits based on sales and digital services.63 Nevertheless, success of the OECD’s effort is important.

In my view, the best path forward lies in a dramatic simplification of the pillar 1 proposal. While working within the OECD’s basic framework, I have demonstrated how substituting formulary apportionment of a smaller percentage of all profits, rather than attempting to allocate an arbitrary percentage of residual profits to market/user jurisdictions, and requiring MNE entities to share in that allocation based on their pro rata shares of residual profits offer the potential for a major simplification of pillar 1 that also avoids double taxation. The proposal here is based on readily available financial information and avoids debatable segmentations, with inevitably controversial revenue and cost allocations, as well as contestable efforts to link an MNE’s particular entities to specific market jurisdictions that would benefit from the reallocations of amount A. All of these inevitably will create controversies that undermine the OECD’s quest for certainty, despite its efforts to institute new mandatory dispute resolution arrangements.

Architects of the OECD pillar 1 blueprint may complain that the pro rata allocations urged here sacrifice the blueprint’s fidelity to greater precision for unscientific outcomes. But, in my view, the blueprint’s veneer of preciseness sows seeds for pillar 1’s downfall, and its promise of certain outcomes is a chimera.


1 OECD, “Tax Challenges Arising From Digitalisation Report on Pillar 1 Blueprint” (Oct. 14, 2020) (hereinafter “pillar 1 blueprint”).

2 A good example is the comments from the coalition Business at OECD (BIAC) (Dec. 14, 2020) (hereinafter “BIAC Comments”). For all public comments on pillar 1 (and pillar 2), see “OECD Comments Received on Reports on Pillar 1, Pillar 2 Blueprints” (Dec. 16, 2020).

3 Pillar 1 blueprint, supra note 1, at chapter 9.

4 Container Corporation of America v. Franchise Tax Board, 463 U.S. 1594 (1983). See also David Bradford and Hugh Ault, “Taxing International Income: An Analysis of the U.S. System and Its Economic Premises,” in Taxation in the Global Economy 11, 30-31 (1990): “The source of income is not a well-defined economic idea.”

5 See Michael J. Graetz, Follow the Money: Essays on International Taxation 83-134 (2016). Originally published as “Taxing International Income: Principles, Outdated Concepts and Unsatisfactory Policies,” 54 Tax L. Rev. 261 (2001).

6 Id. at 213-219 (originally published as Graetz and Rachael Doud, “Technological Innovation, International Competition and the Challenges of International Income Taxation,” 113 Colum. L. Rev. 347 (2013)).

7 E.g., Alan J. Auerbach, “A Modern Corporate Tax,” Center for American Progress (Dec. 2010); Auerbach, Michael P. Devereux, and Helen Simpson, “Taxing Corporate Income,” in Mirrlees Review: Dimensions of Tax Design 837, 882-888 (2020); Reuven S. Avi-Yonah, Kimberly Clausing, and Michael Durst, “Allocating Business Profits for Tax Purposes: A Proposal to Adopt a Formulary Profit Split,” 9 Fla. Tax Rev. 497 (2009); and Devereux et al., Taxing Profit in a Global Economy (coming 2021).

8 See Walter Hellerstein, “Reflections on the Cross-Border Tax Challenges of the Digital Economy,” Tax Notes Int’l, Nov. 25, 2019, p. 671; and South Dakota v. Wayfair Inc., 138 S. Ct. 2080, 2095 (2018).

9 Wayfair, 138 S. Ct. 2080.

10 I focus here on amount A of pillar 1. Pillar 1 also contains an amount B intended to provide a routine return for the remuneration of related-party distributors that perform “baseline marketing and distribution activities” in a market jurisdiction. Amount B is apparently controversial, especially with developing countries, and I do not regard it as essential element to the transformation attempted by pillar 1. I limit my comments here to the calculation of amount A, although the revisions to amount A suggested here may also implicate (or even eliminate the need for) amount B.

11 See, e.g., Follow the Money, supra note 5, at 19-24 (originally published as Graetz and Michael M. O’Hear, “The Original Intent of U.S. International Taxation,” 46 Duke L.J. 1021 (1997)).

12 Pillar 1 blueprint, supra note 1, at ch. 7.

13 When a portion of profits is allocated to the market/user country, this is obvious. When the amount is based on profit margins, the equivalence to formulary sales apportionment may be obscure but is demonstrated by the following simple algebra:

(1) Standard formulary apportionment of profits (where P = worldwide profits, L = local sales, W = worldwide sales):

Formulary allocation of Profits by Sales taxed at rate t:

t * P (L/W)

(2) With apportionment based on worldwide profit margins rather than profits (where profit margin is P/W), the tax is exactly the same as the allocated profits tax above:

t * (P/W)L.

14 A high threshold for business activity in the market/user jurisdiction may exclude some developing countries from an allocation of amount A. The OECD may regard this as necessary, given the complexity and inevitable administrative costs and complexities of the pillar 1 blueprint, but the simplification suggested here, relying on formulas and existing information, may allow a lower threshold of such activity.

15 If instead of exchanging free services for user data, an ADS paid for such data, the users would be required to pay personal income taxes on the fees they receive. Those taxes would go to the countries where the users reside. In addition, those countries might also impose VAT on the consumption of those services. Because these transactions are bartered rather than priced, the users’ countries lose both personal income tax and VAT revenues, whether the supplier of the services is domestic or foreign. Rather than taxing the users on the value of their barters, as a practical substitute it should be possible to impose tax on the providers, with an amount of profits allocated to the countries where the users reside.

16 See also Itai Grinberg, “Design of Scope Limitations for OECD Pillar 1 Work,” Tax Notes Federal, June 15, 2020, p. 1843.

17 To take just one example, the exclusion of financial services for individual consumers seems more likely grounded in political than analytical reasoning. For comments questioning the blueprint’s determinations, see, e.g., BIAC comments and those of Johnson and Johnson.

18 Following the U.S. Supreme Court decision in Wayfair concluding that no physical presence is required for a state to tax online sales, most states embraced the thresholds prescribed by the statute at issue in that case and applied the new rules only to sellers of more than $100,000 of goods or services or more than 200 transactions in the state in a given year. A few states adopted thresholds of $500,000. See Hellerstein, supra note 8; and Wayfair, 138 S. Ct. 2080.

19 Pillar 1 blueprint, supra note 1, at ch. 5. IFRS is more widely used and perhaps more uniform in its applications than GAAP.

20 The pillar 2 blueprint makes some adjustments to book income to reflect a limited number of book-tax differences. It is not clear whether the OECD intends to make similar adjustments under pillar 1. I do not evaluate here which adjustments are apt.

21 The divisions required by IFRS 8 for operating segments should suffice for this purpose.

22 Sullivan, “OECD Pillar 1 ‘Amount A’ Shakes Up Worldwide Profit,” Tax Notes Federal, Feb. 24, 2020, p. 1238. Sullivan’s calculations for 2017 suggest that other U.S. MNEs, such as Caterpillar and Broadcom, have profits of 10 percent or less, and many others, such as IBM and Boeing, are only slightly above the 10 percent threshold.

23 Id.

24 Shoshanna Delventhal, “Apple’s 5 Most Profitable Businesses,” Investopedia, Mar. 10, 2020.

25 See pillar 1 blueprint, supra note 1, at ch. 6. Even if so, this may not reflect how Apple’s various entities are organized, which will be relevant for determining the “paying entities” and jurisdictions discussed in Section III.

26 This exclusion would not occur if all in-scope profits of a company like Amazon were allocated under amount A.

27 As the next section describes, different segmentations are also necessary to determine the “paying entities” and jurisdictions. See pillar 1 blueprint, supra note 1, at ch. 3, 4, 5, and 7.

28 See, e.g., pillar 1 blueprint, supra note 1, at 211.

29 Id. at 126-128.

30 For some MNEs, distinguishing in-scope from out-of-scope profits may require similar, if less intricate, determinations of profits, but that can be solved only by eliminating the scoping distinctions or basing them exclusively on financial statements.

31 See discussion in ch. 6 of the pillar 1 blueprint, supra note 1.

32 See id. at 168 and ch. 9.

34 For further discussion, see Devereux et al., “Residual Profit Allocation by Income,” Oxford University Centre for Business Taxation WP 19/01 (Mar. 2019).

35 For recent examples, see, e.g., Medtronic Inc. v. Commissioner, 900 F.3d 610 (8th Cir. 2018), rev’g T.C. Memo. 2016-12; and Coca-Cola Co. v. Commissioner, 155 T.C. No. 10 (2020).

36 See, e.g., Avi-Yonah, Clausing, and Durst, supra note 7; and Devereux et al., supra note 7.

37 Allocation of profits to production intangibles is obviously no easier. My Treasury experience dealing with controversies leading up to the enactment of section 197, relating to the amortization of purchased intangibles (making no attempt to distinguish among production and marketing intangibles), makes this clear.

38 I do not discuss amount B here, although allocating all profits of in-scope MNEs, rather than only residual profits, might eliminate the need for a separate amount B.

39 Some economists contend that business income taxation should apply only to “economic rents” and that those rents should be allocated to market countries, typically as destination-based cash flow taxes. See, e.g., Devereux et al., supra note 7.

40 As Section III describes, this can be accomplished by requiring the offsets to the amount A allocations by “paying entities” to come out of residual profits based on country-by-country reports. These calculations also show that when an MNE’s entities total positive residual profits are less than the percentage of profits allocated to market/user countries, amount A will be reduced to that total. See infra note 58.

41 See pillar 1 blueprint, supra note 1, at ch. 6 and 7.

42 Limiting the amount A allocation to situations where an MNE meets an overall residual profit threshold, such as 10 percent, is possible, but, as indicated above, this would eliminate companies like Amazon when overall profit margins do not exceed the threshold.

44 Id. at 61.

45 This estimate is based on estimates of U.S. residual profits of about half of profits and somewhat lower global residual profits. See, e.g., Kartikeya Singh, W. Joe Murphy, and Gregory J. Ossi, “The OECD’s Unified Approach — An Analysis of the Revised Regime for Taxing Rights and Income Allocation,” Tax Notes Int’l, Feb. 3, 2020, p. 549.

46 Allocations of amount A could be further limited to situations where market/user countries are not allocated a substantial portion of an MNE’s in-scope profits, say 25 percent, under current arm’s-length pricing.

47 Pillar 1 blueprint 1, supra note 1, at ch. 6 at 123-124.

48 Id. at 124.

49 Id.

50 OECD, “Tax Challenges Arising From Digitalisation — Economic Impact Assessment” (Oct. 12, 2020) (hereinafter, “economic impact assessment”).

51 Id. at Table 2.3, at 35.

52 The report also estimates that with a profitability threshold of 10 percent and a €750 million threshold the great majority of in-scope residual profits (70 to 85 percent) is concentrated among 60 MNE groups.

53 Pillar 1 blueprint, supra note 1, at ch. 7 and 212.

54 Id. at ch. 7 and 214.

55 Economic impact assessment, supra note 50, at ch. 2.

56 This is an important source of information used by the OECD in estimating the share of residual profits in particular jurisdictions. Id. at 29-30.

57 See, e.g., IRS Form 8975 and its instructions.

58 Where T (the sum of all positive residual profits) is less than (C * f) (the total of amounts A to be allocated based on worldwide profits from financial statements), the formula in the text will not produce appropriate offsets. In such cases, f (the percentage of profits to be allocated to market/user countries) cannot exceed T/C. So f should be interpreted to be the lesser of the fixed percentage or the ratio of T to C. If, for example, f is equal to 10 percent, amount A will be reduced for MNEs that have total positive residual profits less than 10 percent of all profits. In such cases, the amounts A to be allocated will equal T, the sum of all positive residual profits from CbC reports.

59 This can be shown by comparing the equation in the text to a formula that the OECD has used in describing pillar 1. The formula indicates that the tax base increase for a country is the global residual profit allocated times the country’s share of in-scope MNE sales profits reduced by the country’s share of residual profits. This formula is contained in slide 29 of the slides for a February 13 OECD presentation on pillar 1. OECD, “Tax Challenges Arising From the Digitalisation of the Economy — Update on the Economic Analysis & Impact Assessment,” OECD Webcast, Feb. 13, 2020. Another somewhat more detailed and realistic version of the formula can be found in the economic impact assessment, supra note 50, at 29.

In the notation of the text, if G = Global residual profit allocated and Q = share of residual profit, the formula in the slide is G(S/W - Q/T), which is structurally similar to the formula in the text. However, the C * f in the text is much easier to determine than G, which is shown as y[P - (R * z)] in Annex B of the pillar 1 blueprint, supra note 1, at 215. (P in that formula is comparable to C above, R is the revenues of a group or segment, z is equivalent to f above, and y is an agreed portion of residual profit attributable to amount A.) P/T in the text is also much easier to calculate than Q/T for reasons described in the text of this article. The point here is simply to show the fundamental structural similarity of the much simpler mechanism of this article to the OECD’s blueprint.

60 Pillar 1 blueprint, supra note 1, at ch. 9.

61 Id. at 168.

63 See, e.g., Devereux et al., supra note 7; and Avi-Yonah, Clausing, and Durst, supra note 7.


Copy RID