Menu
Tax Notes logo

Progressive Formulary Apportionment: The Case for ‘Amount D’

Posted on June 14, 2021
[Editor's Note:

This article originally appeared in the June 14, 2021, issue of Tax Notes Federal.

]

Mitchell Kane is the Gerald L. Wallace Professor of Taxation at New York University School of Law, and Adam Kern is an associate with Covington & Burling LLP in Washington. They thank Wei Cui, Tsilly Dagan, Steven Dean, Anthony Francis, Michael Keen, Ruth Mason, and Eric Zolt for their comments on earlier drafts of this report.

In this report, Kane and Kern demonstrate the advantages of progressive formulary apportionment over other possible international tax reforms, and they show how it could be implemented seamlessly within the architecture of pillar 1.

The views expressed in this report are solely the authors’ and are not attributable to Covington & Burling or its clients.

Copyright 2021 Mitchell Kane and Adam Kern.
All rights reserved.

Led by negotiations at the OECD, the world is moving quickly toward a major revision of the international tax rules. While the details of reform remain up in the air, one aspect seems clear enough: The revised international tax architecture will, to some extent, consolidate the profits earned by the constituent entities of multinational enterprises and apportion some rights to tax those unitary profits by means of a formula. Formulary apportionment has been a centerpiece of the work under the OECD’s pillar 1 from its inception,1 and even the United States — historically the main holdout to pillar 1’s reforms — appears to be broadly sympathetic to the formulary aspect of the proposal.2

While the members of the OECD are mostly rich countries, the OECD has pursued international tax reform through the inclusive framework on base erosion and profit shifting. The inclusive framework is a governance mechanism that gives countries far beyond the OECD’s membership rolls a voice in the redesign of the international tax regime.3 Because the inclusive framework acts through consensus, developing nations will have to be on board with pillar 1. There are signs that they have made their voices heard.4 But whether their interests have been adequately accounted for is up for debate.

In this report, we argue that pillar 1 has cleared the way for a novel and promising kind of international tax reform — a reform that is designed to benefit developing countries but could have substantial benefits for developed countries as well. We call it “progressive formulary apportionment” or “PFA.” Under traditional formulary apportionment, rights to tax businesses are allocated based on taxpayers’ transactional behavior, such as their payroll, assets, or sales. Progressive formulary apportionment allocates at least some taxing rights based on states’ economic well-being, with worse-off countries receiving more valuable rights, all else being equal.

Progressive formulary apportionment has several advantages over other international tax reforms that have been considered as concessions to developing countries. First, it can be highly targeted. Holding fixed the value of the taxing rights that are to be allocated, progressive formulary apportionment can assign those rights to the countries that need them most. Contrast traditional formulary apportionment. It may be possible to select apportionment factors with a distributional impact that would favor developing countries. Single-factor apportionment based on employees, for example, would have such an effect.5 But traditional approaches necessarily remain tied to company-level factors, which may not align with country-level need. Progressive formulary apportionment, by contrast, assesses country-level need directly.

Second, progressive formulary apportionment can be relatively efficient. Because companies cannot easily avoid taxes levied under a progressive formulary scheme by relocating their activities, those taxes would not distort companies’ locational decisions much, if at all. Finally, although progressive formulary apportionment would raise some administrative challenges, it would also avoid some of the administrative costs created by alternative international tax reforms.

Progressive formulary apportionment may appear to be a radical reform — and in some respects, it is. But it could be implemented seamlessly within the framework of pillar 1. That framework envisions apportioning a fixed percentage of nonroutine profit to market jurisdictions under so-called amount A and an amount of routine profit to jurisdictions physically hosting marketing and distribution functions under so-called amount B. To implement progressive formulary apportionment, we explore and defend what we label “amount D,” a taxing right that would apportion a fixed amount of nonroutine profit to developing countries.6 We use that label to reflect the support given to developing countries and to distinguish the right from an “amount C,” which figured in earlier iterations of pillar 1 work but was dropped from the October 2020 report on the pillar 1 blueprint.7 Amount D would apply in addition to amount A and amount B, not in place of them. Further, the set of countries that would benefit from amount D would not be limited to countries that otherwise benefit from amount A or amount B. Rather, amount D would be apportioned based on a set of agreed development indicators and thus would be divorced from any historic concept of nexus between the company and the taxing state.

Our two goals are to demonstrate the advantages of making formulary apportionment progressive and to show how progressive formulary apportionment can be implemented within pillar 1. Thus, we leave many important issues to other work. We do not discuss the merits of formulary apportionment in general or which scheme of progressive formulary apportionment is optimal. We do not endorse a particular set of allocation factors, a particular division of potential tax revenue between amount A and amount D, and so on. Instead, we hope this report will encourage further discussion of those issues.

I. From Pillar 1 to PFA

A. Types of Formulary Apportionment

Formulary apportionment is a method of allocating taxing rights in which the tax attributes of related companies are consolidated, and rights to tax the group’s profit are allocated among jurisdictions by means of a formula.8 Typically, allocation formulas use factors that track the transactional attributes of the company, such as its assets, employees, and sales.9 The term “formulary apportionment” has been around for decades in debates about international tax policy. Increasingly, however, the term is used in conjunction with mechanisms that are dissimilar. Thus, it is useful to have in hand a typology of approaches to formulary apportionment. We distinguish here three approaches.

One-step formulary apportionment, as practiced within the United States in the context of multistate taxation, channels all the company’s profit into a single pool and apportions that profit through a formula.10 Two-step formulary apportionment, exemplified in some academic calls for “residual profit splits,” carves out a portion of the company’s profit, dubbed “routine” profit, from formulary apportionment.11 All the profit that remains, called “residual profit,” is apportioned by means of a formula. Three-step formulary apportionment subdivides the residual profit even further, using more than one mechanism to divide it across countries.

A stylized example of three-step formulary apportionment might look something like this:

Step 1:

  • allocate routine profit through some non-formulary measure.

Step 2:

  • calculate an amount of nonroutine profit as total profit - routine profit;

  • determine some amount X = x percent * nonroutine profit;

  • determine some amount Y = y percent * nonroutine profit; and

  • determine residual nonroutine profit as nonroutine profit - X - Y.

Step 3:

  • apportion X to jurisdictions under some formula a;

  • apportion Y to jurisdictions by some formula b; and

  • allocate any residual nonroutine profit through some non-formulary measure.

There are several important features to highlight about this example. First, there is no limit to the number of pools of nonroutine profit generated in step 2. In our stylized example, we illustrate with two pools of income that will be apportioned by formula and a residual pool that will be allocated by a non-formulary mechanism. But all that is required to generate a model of three-step formulary apportionment is that step 2 creates at least two pools of income. There could be more. Second, the pools of income generated in step 2 need not be subject to the same formula or subject to a formula at all. In our example, formula a is different from formula b.12 That difference could take the form of using different factors or using the same factors but weighing them differently. Third, a pool could be allocated through a non-formulary measure in step 3, but the overall procedure would still be considered formulary apportionment because at least some pools of income would be apportioned in a formulary manner.

B. The Architecture of Pillar 1

The pillar 1 blueprint contains much complexity, which has been summarized elsewhere.13 To develop the argument for progressive formulary apportionment and amount D, we highlight five key elements of the pillar 1 architecture.

First, the pillar 1 blueprint defines the scope of business activities and companies that are potentially subject to tax under it. Amount A would apply only to companies whose global revenue exceeds €750 million and whose profit margins (before tax) exceed a threshold — likely 10 percent.14 It would also apply only to companies that operate in one of two sectors, which the pillar 1 blueprint calls automated digital services (ADSs) and consumer-facing businesses (CFBs). ADSs are services that require minimal human input and are provided over the internet or another electronic network.15 CFBs are businesses that generate revenue from products or services generally sold to individual consumers.16

Second, the pillar 1 blueprint also defines a common tax base for its formulary scheme. The common tax base for amount A is the MNE group’s book income, modified by a few adjustments.17

Third, the pillar 1 blueprint defines nexus rules. For an ADS, a nexus would be established by deriving a large amount of revenue (perhaps €1 million) from a given jurisdiction.18 For a CFB, a nexus would be established in one of two ways: A business might exceed a large revenue threshold (perhaps between €15 million and €20 million); or it might exceed a smaller revenue threshold (perhaps €3 million) and have a “plus” factor, such as physical presence in a jurisdiction.19

Fourth, the pillar 1 blueprint demarcates routine profits tied to sales and distribution and allocates the rights to tax them. The blueprint calls these rights “amount B.” Entities within the scope of amount B are those that perform functions that are relatively impervious to profit shifting, such as marketing and distribution located physically within a jurisdiction. Profit attributable to those activities is apportioned, in accordance with amount B, by applying traditional transfer pricing methods.

Finally, the pillar 1 blueprint defines two pools of residual profit and allocates rights to tax them. The first pool is amount A.20 Amount A is distributed among jurisdictions in proportion to the amount of the company’s revenue that is sourced to each jurisdiction.21 The second pool of residual profit consists of whatever is left after amount A has been subtracted from nonroutine, in-scope profit. The taxation of this pool of profit is left unspecified in pillar 1.22

Pillar 1 is a form of three-step formulary apportionment. First, it carves out a slice of companies’ profit that is to be allocated with traditional transfer pricing methods (amount B) and calculates an amount of residual profit.23 Second, it allocates a portion of the residual profit to market jurisdictions (amount A) — but not all of it. Third, it provides a formulary approach to apportion amount A among market jurisdictions. For the portion of nonroutine profit that exceeds amount A, pillar 1 leaves the allocation of that profit to other mechanisms. Those mechanisms are left unspecified within pillar 1 itself, but because pillar 1 is built on top of existing taxing rights, that profit broadly should be subject to residence-jurisdiction taxation and thus would face a global minimum rate of tax under pillar 2.24

The United States’ recent proposal to modify pillar 1, presented in April and referred to as “comprehensive scoping,” preserves most of pillar 1’s architecture. The main difference between the pillar 1 blueprint and the U.S. proposal concerns the definition of pillar 1’s scope. The U.S. proposal does not restrict pillar 1 to companies that provide ADSs and those that operate CFBs. Instead, it extends pillar 1 to all the largest and most profitable businesses in the world, regardless of their sectors, defined by reference to their revenues and their profit margins.25

C. The Liberalization of Nexus

Our stylized presentation of three-step formulary apportionment leaves most of the difficult questions about international tax base allocation unanswered. Which amounts should be subject to formulary apportionment? And once these amounts are identified, what should the apportionment formulas be? To answer those questions, one must try to state some underlying normative theory of allocating taxing rights.

Pillar 1 has made that foundational issue urgent, and it has strained traditional answers. The pillar 1 blueprint remains grounded in the theory that profit apportionment under the proposed formula must track “value creation.” To square its reforms with that theory, however, the blueprint expands the notion of value creation in unexplained ways. It raises the question: If consumers create value by visiting websites or purchasing goods, why don’t countless other people create value — perhaps by voting for or otherwise sustaining the policies that give rise to global commerce? The U.S. counterproposal raises this challenge even more starkly by seemingly severing any meaningful link between value creation and the allocation criteria, as we explain below. These moves open the door to more creative approaches to the allocation of taxing rights, approaches that are not tied to the transactional behavior of companies.

The foundational principle of pillar 1 (and the entire BEPS project to date) is that profit should be taxed in accordance with the geographical or spatial determinants of value creation. Thus, the first paragraph of the pillar 1 blueprint states:

The integration of national economies and markets has increased substantially in recent years, putting a strain on the international tax rules, which were designed more than a century ago. Weaknesses in the current rules create opportunities for base erosion and profit shifting (BEPS), requiring bold moves by policy makers to restore confidence in the system and ensure that profits are taxed where economic activities take place and value is created.26

In the pillar 1 blueprint, the core implication of the value creation principle is amount A — a new taxing right over nonroutine profit for market jurisdictions. As described earlier, amount A does not apply universally across business sectors. Instead, it applies with sectoral scope restrictions so that it covers profit from only ADSs or CFBs. These sectoral scope limitations are described as “activity tests” to distinguish them from the revenue-threshold-based scoping rules.27

Although the pillar 1 blueprint spends far more time describing the tests than explaining their rationale, it suggests that they are justified because customers create value for companies operating in the ADS and CFB sectors.28 Regarding the ADS sector, the pillar 1 blueprint states:

They [ADSs] often exploit powerful customer or user network effects and generate substantial value from interaction with users and customers. They often benefit from data and content contributions made by users and from the intensive monitoring of users’ activities and the exploitation of corresponding data.29 [Emphasis added.]

And it states the following about the CFB sector:

This significant and sustained engagement is able to take place and create value for consumer-facing MNEs because of the broader digitalization of the economy, as technology facilitates more targeted marketing and branding, and the collection and exploitation of individual consumer data — all of which can be achieved with greater efficiency and remotely. Consumer relationships, interactions with users and consumers and wider consumer-facing intangibles drive value.30 [Emphasis added.]

The important common thread here seems to be that customers in the market are participating in value creation, either by providing some data or helping to create some company-level marketing intangible. Why should market jurisdictions uniquely benefit from amount A? The answer provided by the pillar 1 blueprint appears to be that for ADS and CFB, (1) these jurisdictions are where the corporate taxpayer’s customers are located, and (2) those customers are generating company-level value that ultimately drives the profit base that is being taxed.

Many questions could be asked about the normative theory implicit in the pillar 1 blueprint. Do consumers also create value captured by companies operating outside the ADS and CFB sectors? If merely streaming a movie counts as creating value, what else might? And if consumers are not special, why should their countries receive all the tax base to be apportioned by formula under pillar 1?

As noted earlier, under comprehensive scoping, the United States has proposed that the ADS and CFB activity tests be removed and that instead, pillar 1 apply to sufficiently large companies, based on revenue and profit measures. Under comprehensive scoping, value creation is all but dead as a justificatory principle.31 Revenue or profit thresholds tell us nothing about how particular profit is created. A dollar of revenue or profit is treated identically with any other dollar of revenue or profit, regardless of the mechanisms underlying the generation of the revenue or profit. Under comprehensive scoping, the operative principle seems to be to tax the large companies because that’s where the bulk of the tax base is and those are the companies most able and willing to comply. If that is the operative principle, though, why should market jurisdictions receive all the tax base to be apportioned by formula?

Pillar 1 represents an important moment in the history of the international tax regime. It substantially relaxes nexus rules but fails to provide a justification for why market countries alone should enjoy expanded taxing rights under the new regime. It thus opens the door to thinking in creative ways about the international allocation of taxing rights.

II. The Case for PFA

Progressive formulary apportionment would allocate rights to tax unitary company profit with country-level macroeconomic or development indicators that transcend the transactional behavior of the company. Countries that are worse off would receive a greater share of rights to tax the company, all else being equal. Progressive formulary apportionment has several advantages over other international tax reforms, and it could be implemented within the architecture of pillar 1.

A. Amount D

Before evaluating progressive formulary apportionment, we offer a more concrete idea of what it might come to. To do this, we demonstrate how pillar 1 could be made progressive through the inclusion of amount D. Although we have been critical of the connection between the activity tests articulated in the pillar 1 blueprint and the concept of value creation, our proposal is not that amount D should displace amount A. Even if one were inclined to that position on distributive grounds, for pragmatic reasons, we begin with the assumption that any global consensus under pillar 1 will involve a substantial allocation of taxing rights to market jurisdictions. The discussion below therefore shows the incorporation of amount D into pillar 1 alongside amount A.

It is possible to implement a form of progressive formulary apportionment within the architecture of pillar 1. One could do this by introducing an amount D, in favor of developing countries, into the formulary split of unitary company profit envisioned by pillar 1. We show this below using the framework set out in Annex B of the pillar 1 blueprint. Annex B describes two approaches: the profit-based approach and the profit-margin-based approach. These are shown to be mathematically equivalent in the pillar 1 blueprint.32 We demonstrate the incorporation of amount D under the profit-based approach, though it could equally be implemented through the profit-margin-based approach.

Annex B describes the following three-step process, reflecting pillar 1’s proposal for three-step formulary apportionment:

  • Step 1. First determine: W = P - (R * z), where P captures all in-scope profit (under the pillar 1 blueprint, this would incorporate the activity tests, but under U.S.-proposed comprehensive scoping, it would not); and the quantity (R * z) is routine profit (calculated as R revenue multiplied by some z profit indicator). The quantity W is thus the aggregate of in-scope nonroutine profit. One can also set W = A + X. In this expression, A refers to pillar 1 amount A, and X captures any residual in-scope nonroutine profit that is not ascribed to amount A.

  • Step 2. Apply a fixed percentage to W to determine amount A: A = y percent * W.

  • Step 3. Determine the portion of amount A that is apportioned to a market jurisdiction, M, as follows: M = S/R * A. In this expression, R is all in-scope revenue, and S is the in-scope revenue sourced to M. This expression is equivalent to S/R * y percent * W.

That process could be modified to incorporate amount D as follows:

  • Step 1. W = A + D + X. In this expression, D reflects that some portion of nonroutine residual profit is ascribed to amount D.

  • Step 2. D = u percent * W. In this expression, u percent reflects the actual portion of nonroutine profit that is ascribed to amount D.

  • Step 3. DevJ = DevIndexJ * D, or, equivalently, DevJ = DevIndexJ * u percent * W. In these expressions, DevJ reflects the portion of amount D that is apportioned to a developing country, J. The term DevIndexJ refers to a composite of development indicators, as applied to J. We can further specify: Equation where Ji = the normalized value of indicator i of n indicators in developing country J of m states eligible for amount D, and w reflects the weight to be given to indicator i.

For an illustration, we apply this framework to an example assuming a composite of particular indicators. We stress, however, that our example adopts arbitrary assumptions simply to demonstrate the method. An important feature of our proposal is that it is generalizable and leaves the following crucial factors unspecified and to be determined under future negotiations: (1) the value of u (that is, the percentage of nonroutine profit ascribed to amount D); (2) the identity of m developing countries; (3) the list of n development indicators; and (4) the weight w to be given to each development indicator i.

We envision that the set of m countries would be demarcated by reference to some cutoff point, likely based on the n development indicators. The indicators would be state-level metrics that are not tied to taxpayer behavior and thus do not implicate or involve historic concepts of nexus. This “country scoping” would be consistent with the overall development-focused goals of progressive formulary apportionment, as we defend it here. Like any tax policy with a sharp cutoff, there could be discontinuities and cliff effects at the margin, although these could be softened by designing the formula so that a relatively small amount of the tax base is apportioned to states right above the cutoff.33

Imagine a company with $1,000x of in-scope profit calculated from consolidated financial statements in accordance with the principles of pillar 1 and $1,500x of in-scope revenue.34 To preserve generality, we make no assumptions about the character of the profit and revenue that are in-scope. The example would work the same way if some activity tests were preserved or if there was comprehensive scoping as proposed by the United States. We assume an 8 percent profitability threshold to determine routine profit under pillar 1’s formulary approach to profit apportionment. We assume a 10 percent allocation factor to determine amount A and a 10 percent allocation factor to determine amount D. In apportioning amount D, we assume that DevIndex will be calculated by giving 50 percent weight to the reciprocal of per capita GDP and 50 percent weight to population. We illustrate in the table above with a set of five hypothetical countries with sample values for per capita GDP and population. Application of pillar 1’s three-step formulary approach would proceed as follows:

  • Step 1.

    • W = P - (R * z)

    • W = $1,000x - ($1,500x * 0.08) = $880x

    • $880x = A + D + X

  • Step 2.

    • A = 0.1 * $880x = $88x

    • D = 0.1 * $880x = $88x

    • X = $704x

  • Step 3.

    • Apportion $88x of amount A among market jurisdictions in proportion to revenue from in-scope sales. We do not further illustrate this here, but note that this apportionment is accomplished under a formula that is separate from the formula used to apportion amount D.

    • Apportion $88x of amount D among five developing jurisdictions in proportion to the chosen DevIndex, calculated for each jurisdiction as illustrated in the table.

Apportionment of Amount D With 50-50 Weighting of Two Indicators

Jurisdiction

GDP (millions)

Population (1,000s)

Population (normalized)

GDP Per Capita

1/(GDP) (per capita dollars) (normalized)

DevIndex

Nominal Amount D

1

$2,000

5,000

2.15

$400

2.46

46.10%

$40.56x

2

$3,000

5,000

2.15

$600

1.64

37.90%

$33.35x

3

$1,500

1,000

0.43

$1,500

0.66

10.86%

$9.56x

4

$4,000

500

0.22

$8,000

0.12

3.38%

$2.97x

5

$1,000

125

0.05

$8,000

0.12

1.77%

$1.56x

We stress again that the choice of DevIndex and the selection of five jurisdictions here is purely for illustration. The illustrated procedure could be used with any set of development indicators and any set of chosen weights. We opted to illustrate with two indicators and a small set of countries so that it would be readily apparent how indicators can interact with one another in such a formula.35 A different formula, potentially with different factors and different weights, might be better for an actual amount D.

We do not further specify in this report any particular formula as optimal. We would note, however, that the selection of development indicators should heed the potential adverse effects of particular indicators on country-level behavior. For example, although the apportionment of amount D should generate revenue for jurisdictions that are most in need of fiscal capacity, it might not be wise to include tax/GDP ratios in the formula because doing so might simply encourage countries to collect less revenue through existing channels.36 Similarly, particularly granular indicators (related, for example, to health or education outcomes) could distort choices about which social services to provide.

B. Advantages of PFA

The allocation of taxing rights to developing nations is worthy of attention for several reasons. The most straightforward reason is ethical. Many people in developing nations could be better off if their states had greater fiscal capacity, and the international tax regime should account for their interests. Also, developing countries are not merely passive bystanders to international tax diplomacy. Several members of the inclusive framework have enacted, or are considering enacting, unilateral digital services taxes, and they are looking to receive concessions in exchange for abandoning unilateral approaches in favor of multilateral ones. Even a hard-bitten realist who would like to make those concessions as small as possible should be interested in how to structure them in the best possible way.

Progressive formulary apportionment has several advantages over alternative mechanisms of allocating fiscal capacity to developing countries: It could allocate fiscal capacity in a more targeted fashion; it could allocate resources efficiently; and it could save some administrative costs.

1. Distribution.

From a distributive perspective, progressive formulary apportionment has two valuable features. First, it is immensely flexible. The formula can be fine-tuned to achieve a wide range of possible distributions. Second, progressive formulary apportionment tracks developing countries’ unmet needs for fiscal capacity. If those unmet needs are, in general, inversely correlated with countries’ degrees of development, progressive formulary apportionment ensures that countries receive more fiscal capacity insofar as they need more of it.37 Thus, progressive formulary apportionment can be highly targeted. It can ensure that each dollar’s worth of taxing rights earmarked for developing countries does a relatively large amount of good.

This is a notable advantage of progressive formulary apportionment over other international tax reforms, such as traditional formulary apportionment. For traditional formulary apportionment to deliver fiscal capacity to the countries that need it most, there must be a coincidence between the aggregate transactional choices made by companies (such as where to locate their assets, employees, and sales) and countries’ unmet fiscal needs. This coincidence is unlikely to obtain. It might be true that more populous countries tend to host greater volumes of economic activity, and it is plausible that more populous countries tend to have greater needs for fiscal capacity. Even so, many other determinants of countries’ unmet fiscal needs do not vary along with companies’ aggregate transactional choices. The greater the volume of economic activities hosted by a country, the more likely it is that that country is highly developed, has access to a more extensive tax base and capacity to administer taxes, and therefore has a greater ability to meet its revenue needs. Thus, trying to allocate taxing rights to developing countries by tinkering with traditional formulary apportionment — perhaps by giving greater weight to the payroll factor or by lowering the revenue or profit margin thresholds required for companies to be within the scope of amount A — is likely to be less targeted than progressive formulary apportionment.38

2. Efficiency.

The efficiency of an international tax regime turns, in large part, on how easy it is for taxpayers to relocate their profits so that they fall under the jurisdiction of a more lenient sovereign. Taxes are efficient insofar as they do not distort decisions made by private actors. Taxes tend not to distort behavior when they fall on behavior that is hard for taxpayers to change. And in the international context, one way to avoid tax is to change the location of one’s profit so that it no longer falls within the scope of a given tax. Thus, one central ambition of recent academic work on international tax reform has been to design a tax base that companies cannot easily move.39

In this respect, progressive formulary apportionment is hard to beat. It is virtually impossible for a company to manipulate the economic features of a state to such an extent that would seriously affect its tax liabilities. Moreover, if a particular implementation of progressive formulary apportionment, such as amount D, does not include nexus rules, companies cannot remove themselves from a state’s jurisdiction to tax by relocating any of their transactional behavior.

Progressive formulary apportionment’s efficiency is another reason to prefer it to other possible international tax reforms. Consider, again, efforts to tweak traditional formulary apportionment so that it is more favorable to developing nations. Perhaps the most prominent academic argument for destination-based formulary apportionment is that destination-based taxes are relatively efficient.40 It is hard for businesses to relocate their consumers or their sales — relative, at least, to other traditional formulary factors, such as assets or payroll. But it is harder still for businesses to manipulate the macroeconomic features of states.

The efficiency of progressive formulary apportionment can be advantageous to developing and developed countries alike. Because progressive formulary apportionment is more efficient than alternative international tax reforms, embracing it would increase the total stock of resources to be shared between developing and developed countries. Thus, for a given value of taxing rights enjoyed by developing countries, more valuable taxing rights can be held by developed countries, and vice versa.

3. Administrability.

As we discuss later, progressive formulary apportionment raises some administrative difficulties. But it would save some administrative costs as well, compared with some alternative international tax reforms that are under serious consideration. For example, in negotiations on pillar 1, developing countries have pushed for lowering the revenue thresholds required for companies to be within pillar 1’s scope.41 This proposal has been resisted, in part, because lowering the revenue thresholds would increase the aggregate costs of tax compliance.42 The lower the revenue threshold, the more companies are within the scope of pillar 1, and the smaller their size, the more likely it is that they face high marginal costs of complying with pillar 1. Implementing amount D rather than lowering the revenue threshold would save those costs.

C. Answering Objections to PFA

We anticipate two classes of objections to progressive formulary apportionment: one related to the fairness of the base allocation and the other related to the specter of perverse effects on public-good production and governance.

According to the first type of objection, the signature feature of progressive formulary apportionment — its allocation of some taxing rights based on indicators of countries’ economic well-being — is unfair because the rights so allocated do not properly reflect countries’ relative contributions to companies’ profits. When the people of a country make a company’s profit possible, the objection goes, it is only fair that they receive a return on their efforts. This objection rests on what one of us has called the “contributory principle,” according to which the value of each country’s set of taxing rights ought to be proportionate to its contributions to the global surplus.43 This objection’s conclusion does not follow from its premises.44 Because the world as we know it is the product of centuries of international economic interaction, income that arises anywhere may be the product of contributions from elsewhere. Thus, even if it is true that countries ought to receive proportionate returns on their contributions to companies’ profits, it does not follow that all apportionment factors must track the transactional behavior of companies.45

Even if we were to concede that countries that host a company contribute to its profit in a way that other countries do not, it is far from clear that allocating amount A to market jurisdictions — without provision for something like amount D — would help ensure that the value of each state’s set of taxing rights is more proportionate to its contributions to the global surplus. That conclusion would follow only if we make several further assumptions about how well the rest of the international tax regime tracks countries’ contributions to companies’ profits and about the relationship between the values of market jurisdictions’ contributions and their shares of companies’ residual profits. No one — including the authors of the report on pillar 1 — has defended these assumptions, and we are skeptical that any attempted defense would succeed.

According to the second type of objection, progressive formulary apportionment inhibits the efficient production of public goods. There is a long line of work in public economics that connects benefits-based taxation to producing the optimal basket of public goods.46 The basic idea behind this work is that benefits-based taxation acts as a price signal for governments. If governments can see the exact quantity of benefits produced by bundles of public inputs to production, they will spend no more and no less than is required to produce the benefits. If governments are denied access to this signal, they will produce worse baskets of public goods. They might overproduce public goods, they might underproduce them, or they might produce the wrong kinds of public goods.

There are two responses to that objection. First, benefits-based taxation does not always lead to the efficient production of public goods in realistic settings. Indeed, progressive formulary apportionment might improve the production of public goods. Some recent academic work suggests that under conditions of tax competition, governments will choose suboptimal baskets of public goods, overproducing public inputs to production and underproducing goods to be consumed (such as parks).47 The intuition behind this result is that under tax competition, governments face excessive pressure to produce public goods that will attract investment. Sharing potential tax revenue across jurisdictions can relieve that pressure. Second, given the poor fit between benefits conferred by jurisdictions and taxing rights allocated to them under pillar 1 — as specified in both the blueprint and comprehensive scoping — it is highly unlikely that pillar 1 as currently envisioned would send a strong price signal to governments.

According to a related objection, progressive formulary apportionment could cause global taxing rights to be misused in other ways. Perhaps it removes a selection effect, causing taxing rights to be allocated to governments that do not serve their people well. The rough idea behind this thought is that the quantity of amount A — the overall revenue derived from large businesses operating within a government’s territory — might be a proxy for good governance. Or perhaps progressive formulary apportionment would remove an incentive for governments to govern well. If governments’ funding shortfalls are cushioned to some extent by receiving greater quantities of amount D, they would have less incentive to promote development within their countries.

We have three responses. First, we doubt that the quantity of amount A would be strongly correlated with governmental performance. Second, even if states would receive less of amount D as they develop, they have ample other incentives to promote development. Finally, amount D’s apportionment formula could incorporate indexes of good governance. This would strike a balance between progressivity and performance.

III. Implementation

We discuss here several potential implementation issues. In keeping with our overall approach to retain as much generality in our proposal as possible, we do not reach firm conclusions in this part but rather elaborate on what we take to be relevant considerations and potential solutions.

A. Tax Rate

In our discussion above, we described amount D in terms of tax base — not an amount of tax revenue. Thus, in the example in Section II.A, we determined an aggregate amount D of $88x to be apportioned among five countries. We have said nothing about the tax rate that might apply to this tax base to determine associated tax revenue. This lack of specificity is consistent with the pillar 1 blueprint, in which we are given a method to determine amount A in the aggregate and a method to apportion amount A among market jurisdictions, but the tax rate to apply to those amounts is left unspecified. In terms of the tax rate to apply to amount D, there are two possible approaches.

First, amount D could be treated similarly to amount A, and the jurisdictions to which amount D is apportioned could set the relevant tax rate. Jurisdictions might be expected to favor this decentralized rate approach based on some notion of tax sovereignty — that once companies’ profits are apportioned, it should be in the power of the jurisdiction to determine how to tax them. This tracks the historic approach to international tax base allocation, in which the only way we have achieved any sort of coordination of tax rates by nonresidence countries is through treaty mechanisms, under which, for example, numerous jurisdictions may settle on identical or similar withholding rates on outflowing payments. But even treaty coordination takes place against the backdrop of the jurisdiction’s ability to set some higher domestic rate, which is then negotiated away in the treaty process.

The alternative approach would be to set a single, uniform tax rate across all jurisdictions for amount D as part of pillar 1. This would bring substantial administrative gains. Further, we believe that the agreed global minimum tax rate under pillar 2 would be an obvious candidate for a single tax rate. In principle, if pillar 1 and pillar 2 work together as envisioned, setting the tax rate on amount D at any level lower than the global minimum tax rate would simply have the effect of companies paying the same aggregate amount of tax but shifting tax receipts from developing countries to non-developing residence countries. This conflicts with the main reason for establishing an amount D in the first place.

B. Collection Agent

A closely related issue is the identity of the collection agent for any tax on amount D. Under traditional concepts of international taxation, one would presume that companies would pay tax on amount D to the relevant country to which a portion of amount D had been apportioned. This is how tax on amount A is intended to apply.48

There are, however, significant drawbacks to that approach in terms of administrative complexity. Although the identity of the countries to be apportioned tax base under amount D remains to be determined, we imagine that this could be a fairly large number of countries. Further, as we argued earlier, the apportionment of amount D would not be limited to countries that have some nexus to the company under historic nexus concepts or even to countries that touch the company’s supply chains. Thus, to take the standard approach would seemingly require the company to file tax returns with many jurisdictions with which it has not had historic contact. We thus think there are strong reasons to administer amount D tax through a central collection agent. The obvious option would be a fund established by one of the existing international organizations, such as the U.N., the World Bank, or the IMF.

We predict that some critics will reject this out of hand as an illicit first step toward a global taxing authority. We believe those criticisms are likely to be overblown. We see three basic hurdles to establishing a central collection agent to administer tax on MNEs in a way that severs the link between the collection function and the individual states that will benefit from the collected funds. Each can be overcome.

The first two hurdles relate to the determination of the tax rate and the tax base. If the company has to separately determine a tax rate and tax base for each of the many jurisdictions to which its tax will ultimately flow, we will have sacrificed nearly all the administrative gains that might have followed from centralization in the first place. But that variability in base and rate does not seem to be where we are headed.

As for the tax base, a central aspect of pillar 1 is to determine a common measure of unitary company profit. The calculation of this base would not be affected by the domestic laws of the various states that might benefit from amount D. Further, the quantity of aggregate amount D would not be affected by domestic law but rather would be an output of the variable u, as specified earlier, which is meant to be a multilaterally agreed consensus amount.

As for the tax rate, we have already mentioned that there are sound reasons within the context of a joint adoption of pillar 1 and pillar 2 to set a uniform tax rate on amount D that is the same as the globally agreed minimum tax rate under pillar 2. This would solve administrative issues that would otherwise follow from companies having to determine a composite of rates. Under a centralized collection agent, the administrative burden on companies is almost laughably simple. The company would have already calculated nonresidual in-scope profit in step 1 of the pillar 1 framework. At that point, the company need only multiply this amount by u percent and then apply the global minimum tax rate to determine the amount to be remitted to the centralized collection agent. The apportionment of that amount and the delivery of proceeds would be the function of the collection agent, not the company.

Note further that effective collection and delivery of amount D tax likely would not require endowing the collection agent with independent enforcement powers. Under the pillar 1 blueprint, it is envisioned that amount A (and, by implication, amount D under our proposal) would apply only to fairly large multinational companies. Under the U.S. proposal for comprehensive scoping, the size of affected companies would, on average, be even larger; for a constant amount of revenue, comprehensive scoping would apply to some large companies that flunk the pillar 1 blueprint activity tests. We imagine that the identity of affected companies will, in general, be a matter of public record and readily available to the collection agent. If a large multinational were to refuse to remit tax on amount D, this fact could be made publicly known by the collection agent, presumably with substantial reputational costs to the company. Because a properly structured tax on amount D should be creditable against the pillar 2 global minimum tax, it is unlikely that companies would choose to incur those costs. The company that chooses to ignore its amount D remittance obligation would risk incurring substantial reputational costs, with no offsetting financial benefit.

The third hurdle will initially appear to be a larger one, and it links directly to concerns about a global taxing authority. Critics of the idea of paying tax to a centralized collection agent of an international organization can be expected to raise concerns about democratic deficits. We think those arguments are unpersuasive. Concerns about democratic deficits must relate to either decisions made by an authority to set remittance obligations or decisions made by an authority about how to spend those remittances. The collection agent we describe here would have no such powers. The remittance obligation and the identity of the beneficiaries of remittances (and the extent to which they benefit) would be outside the control of the collection agent. Rather, those features would be decided at the outset as part of the globally negotiated consensus framework consisting of pillar 1 and pillar 2. Terminologically, it would not really be correct even to describe the company as paying a “tax” to the collection agent (although from the company’s standpoint, it should be considered a “tax” for tax credit and minimum tax purposes, as mentioned earlier). We have chosen the term “collection agent” on purpose. That agent would simply act as a ministerial intermediary — calculating and delivering the appropriate amount of revenue to the principals in this arrangement — that is, the democratically responsive sovereigns that themselves have the power to levy tax and are parties to the global consensus.

Because tax liability for amount D would ultimately be a legal obligation under domestic law, it would be necessary for the relevant developing countries to enact legislation matching the contours of whatever might eventually be agreed through the inclusive framework process. Generally, this will raise the same sorts of issues that have already been detailed in the pillar 1 blueprint regarding domestic-law implementation of amount A.49 If amount D tax were to be remitted to a centralized collection agent, however, it would also be necessary to explore constitutional restrictions on the delegation of sovereign authority to outside actors.

C. Dispute Resolution

Any implementation of amount D would have to resolve potential disputes arising from its application. This might appear particularly troublesome under any model with a centralized non-state collection agent. We do not believe, however, that complications regarding dispute resolution would pose a serious hurdle to implementing amount D.

Comparison with amount A is instructive. The pillar 1 blueprint enumerates four sorts of disputes that might arise under pillar 1. The blueprint says these disputes might concern “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 eliminating double taxation.”50

Two of these disputes will not arise with amount D. Nexus is not an operative concept in the calculation of amount D, and concerns about double taxation should not arise because amount D does not overlap with taxing rights that currently exist. The other two kinds of disputes — regarding the correct delineation of business lines and the allocation of central costs and tax losses to business lines — could arise under amount D.51 But because these disputes would mirror disputes arising under amount A, amount D could piggyback on whatever dispute resolution mechanisms are adopted for amount A.

D. The Apportionment Formula

As with all exercises in multilateral formulary apportionment, the final determination of the apportionment factors and their weights — and the related question of who are the winners and who are the losers — can be expected to be difficult and contested. We do not suggest any concrete answers here, and we leave the analysis of possible instantiations of our DevIndex to experts in economic development and development finance. We would only note here that much work has already been done in this direction on development indicators, which could be incorporated without difficulty into our framework. This extends to the inclusion of noneconomic development factors and preexisting composites. For example, consider the human development index developed by the U.N.52 This is itself a composite of indicators of development relating to noneconomic factors. The U.N. human development index could be incorporated directly into our DevIndex as one of the i indicators to be considered.

IV. Conclusion

In this report, we have introduced progressive formulary apportionment and urged that it be implemented within pillar 1 by establishing a new taxing right, amount D, in favor of developing countries. As we have acknowledged, our proposal is radical in important ways. Despite decades of debates about formulary apportionment, we are aware of no prior proposal in the tax literature or in the policy space that urges the adoption of apportionment factors that would use state-level macroeconomic or development indicators in addition to company-level factors. But in other ways, our proposal is simply a natural extension of the trajectory that we are observing in pillar 1.

Pillar 1 and amount A are, of course, about demarcating the rights of states to tax nonresident companies. (Although there may be pragmatic reasons, such as inversion risk, to curtail the taxation of resident companies, the absence of the right to tax those taxpayers under international norms is not thought to be a limiting factor.) Historically, the taxation of nonresidents has required at least some transactional connection between the state and the company that the state seeks to tax. Conceptually, one way to understand the shift from one-step to two- or three-step formulary apportionment — and the division between routine profit and some residual profit that this shift entails — is that it is an acknowledgment that our standard transactional tools run out at some point given the hyperconnectivity and integration of the global economy and of the large multinational corporations that play such an important role in it. The standard mechanisms (arm’s-length transfer pricing and separate entity accounting) and the standard logic (search for some nexus) work well enough for cases in which we can still readily determine physical presence and ascribe some ordinary return to the functions physically taking place within a jurisdiction. When the standard tools run out, however, we need a different approach — thus, the need to apportion residuals by formula.

In our view, the pillar 1 blueprint is misguided in its attempt to cling to historic demands for nexus when undertaking this task. The pillar 1 blueprint does so by tying some delimited business sectors to the amorphous concept of value creation. We think this effort is bound to fail because it is at odds with the very indeterminacy of the question about linking profit to jurisdictions that drives the need for reform in the first place. The recent call by the United States to replace the activity tests with comprehensive scoping has amplified this point.

We have not offered anything like a comprehensive normative defense of how best to allocate a portion of the international tax base in the absence of historic concepts of nexus. We have indicated the need for that project and begun its undertaking elsewhere.53 Here, we simply observe that it is difficult to see how the current contours of pillar 1 can be squared with a justifiable approach to this basic problem.

If we accept that any sound justification would be based on some mix of efficiency, distributional concerns, and administrative concerns, it is hard to see why market jurisdictions should have the exclusive rights to tax nonresidents on their residual profit. In terms of distribution, the only argument that would seem to be on the table to justify that position is that the market jurisdictions continue to contribute to company profit (even absent any reflection of historic nexus standards) because they play a crucial role as the ending point of a company’s supply chain. But we have questioned that contributory argument in this report. Further, even if one were wedded to the idea that nexus in the future will just mean participating in the company’s supply chain, many other jurisdictions touch global supply chains, such as jurisdictions in which a company is not resident but sources labor supply, which the pillar 1 blueprint ignores. In terms of efficiency, apportioning residual profit based on final sales may well be efficient as compared with apportionment that tracks more mobile production factors. But, as we have argued, that apportionment is even less efficient than an approach that does not look to company transactional factors at all.

We are rapidly approaching an inflection point in the design of the international tax system in which nexus will not play its historic role in allocating rights to tax large portions of taxable profit of multinational companies to nonresidence jurisdictions. The time is thus ripe to think in new ways about how to allocate taxing rights in that world. In this respect, progressive formulary apportionment, implemented through amount D, has highly desirable distributional characteristics, is efficient, and could easily be administered through the emerging global international tax architecture.

FOOTNOTES

1 See OECD, “Programme of Work to Develop a Consensus Solution to the Tax Challenges Arising From the Digitalisation of the Economy” (May 28, 2019). In the program of work, the OECD first set out the broad guidelines of a two-pillar approach, with pillar 1 concerning profit allocation and nexus rules and with pillar 2 concerning a global minimum tax. The program of work describes possible allocation approaches involving “allocation keys,” thus implicating a formulary method. See id. at paras. 28 and 30.

2 A recent Treasury presentation to the OECD indicated that the work on profit allocation (that is, the aspect of the work with a formulary component) was progressing. See Treasury, “Steering Group of the Inclusive Framework Meeting: Presentation of the United States,” at slide 20 (Apr. 8, 2021).

3 For background on the inclusive framework, see Ruth Mason, “The Transformation of International Tax,” 114 Am. J. Int’l L. 353 (2020); and Allison Christians, “BEPS and the New International Tax Order,” 2016 BYU L. Rev. 1603, 1605 (Dec. 2016).

4 For example, in its recent presentation to the OECD, Treasury stated, “We are prepared to be flexible regarding nexus thresholds to ensure that Pillar 1 benefits developing countries.” Treasury, supra note 2, at slide 17.

5 See Ruud de Mooij, Li Liu, and Dinar Prihardini, “An Assessment of Global Formula Apportionment,” IMF Working Paper No. 19/213, at 23 (Oct. 11, 2019).

6 After submitting this report for publication, we learned that the African Tax Administration Forum (ATAF) had submitted a proposal to the inclusive framework to modify pillar 1 to include an “amount D.” See ATAF statement (May 12, 2021). ATAF’s proposal regarding “amount D” would, like our proposal, involve an allocation of taxing rights that favors developing countries. However, ATAF’s proposal would not implement a form of “progressive formulary apportionment” as we use that term in this report. It also departs from our proposal regarding other crucial features, such as those concerning nexus and the relationship to amount A.

7 Amount C would have allocated returns to market jurisdictions under arm’s-length principles based on in-country functions that would not be properly remunerated under amount B. See OECD, “Public Consultation Document, Secretariat Proposal for a ‘Unified Approach’ Under Pillar One” (2019). The status of pillar 1 work is captured in the pillar 1 blueprint. See OECD, “Tax Challenges Arising From Digitalisation — Report on Pillar One Blueprint: Inclusive Framework on BEPS” (2020) (pillar 1 blueprint).

8 See Urban-Brookings Tax Policy Center, “Briefing Book: How Would Formulary Apportionment Work?” (updated May 2020).

9 Id.

10 Id.

11 See, e.g., Reuven S. Avi-Yonah, Kimberly A. Clausing, and Michael C. Durst, “Allocating Business Profits for Tax Purposes: A Proposal to Adopt a Formulary Profit Split,” 9 Fla. Tax. Rev. 497 (2009); and Michael P. Devereux et al., Taxing Profit in a Global Economy (2021).

12 Indeed, if there were no difference between a and b, there would be no reason to treat them as separate pools of income in the first place.

13 For a useful summary, see Harmen van Dam et al., “Taxing the Digitalized Economy: Key Takeaways From the OECD Public Consultation on the Pillar One and Pillar Two Blueprints,” 28 Int’l Transfer Pricing J. 3 (2021).

14 Pillar 1 blueprint, supra note 7, at paras. 37 and 507-510.

15 Id. at paras. 24-30.

16 Id. at paras. 31-34.

17 Id. at paras. 407-410.

18 Id. at paras. 186-201.

19 Id. at paras. 202-215.

20 Id. at para. 496.

21 Id.

22 Id. at paras. 511-512.

23 The relationship between amount B and nonroutine profit under pillar 1 is more complex than what is captured by our stylized example. Amount B is accorded to market jurisdictions under traditional arm’s-length principles. In calculating the residual profit for purposes of amount A, however, what is subtracted from total company profit is not amount B itself but a fixed return. Subtracting amount B for these purposes would be incorrect because amount B does not have the scope limitations of amount A. See id. at para. 658.

25 See Treasury, supra note 2.

26 Pillar 1 blueprint, supra note 7, at 3.

27 Id. at para. 23.

28 The pillar 1 blueprint does hint at an alternative rationale, according to which the activity tests align with cases in which there is remote participation in the economy even absent traditional physical nexus. See id. at Box 2.18.

29 Id. at para. 24.

30 Id. at para. 31.

31 Understandably, the United States has not defended comprehensive scoping by appealing to value creation. Instead, it has appealed to administrative concerns and the fact that the activity tests would disproportionately burden companies headquartered in the United States.

32 See pillar 1 blueprint, supra note 7, at Annex B.

33 As a formal matter, nothing in our proposal requires country scoping. The set of m countries could be extended to include all countries, or at least all countries that participate in the negotiations regarding pillar 1. For a constant DevIndex and quantity of amount D, however, country scoping is more progressive than universal scoping.

34 Our example assumes the company is profitable. The pillar 1 blueprint has an extensive discussion of how to handle losses. The basic idea is that losses in a year will be pooled (rather than apportioned to particular market jurisdictions) and then carried forward. Pillar 1 blueprint, supra note 7, at paras. 479-480. We envision that amount D would work in parallel with whatever approach is ultimately taken for losses under amount A. In other words, a company would have a tax base ascribed to amount D only in tax years in which there is also some base ascribed to amount A.

35 For example, note that jurisdictions 1 and 2 have the same population, but Jurisdiction 1 is poorer based on a measure of per capita GDP. With the chosen DevIndex, Jurisdiction 1 thus receives a higher share of amount D. To similar effect, jurisdictions 4 and 5 have the same income levels measured by per capita GDP, but Jurisdiction 4 has a much larger population, and with the chosen DevIndex, it receives a higher share of amount D than Jurisdiction 5.

36 This is sometimes called the “rate tax-back problem.” See, e.g., Leonard S. Wilson, “Macro Formulas for Equalization,” in Intergovernmental Fiscal Transfers 339, 348-349 (2007).

37 This assumption is plausible on a range of normative theories about the fair allocation of international taxing rights, including those that recommend allocating taxing rights to achieve global egalitarian or sufficientarian distributions among individuals, and those that recommend allocating taxing rights so as to fit each country’s contributions and allege that developing countries’ contributions are currently undervalued. For examples of the lattermost theory, see Christians and Laurens van Apeldoorn, “Taxing Income Where Value Is Created,” 22 Fla. Tax Rev. 1 (2018); and Steven Dean, “Neither Rules Nor Standards,” 87 Notre Dame L. Rev. 537 (2011).

38 A similar point holds, albeit to a lesser degree, for Ilan Benshalom’s brief sketch of a scheme of progressive formulary apportionment in “How to Redistribute? A Critical Examination of Mechanisms to Promote Global Wealth Redistribution,” 64 U. Toronto L.J. 317, 355 (2014). Benshalom suggests a scheme of one-step formulary apportionment, in which traditional formulary factors are weighted by country-level factors. Because Benshalom’s scheme remains linked to the companies’ transactional behavior, it is not as targeted as amount D, which solely tracks country-level factors.

39 See, e.g., Devereux et al., supra note 11; Devereux, “How Should Business Profit Be Taxed?” 40 Fisc. Stud. 591, 595 (2019); Wei Cui, “The Digital Services Tax: A Conceptual Defense,” 73 Tax L. Rev. 69 (2019); and Avi-Yonah, Clausing, and Durst, supra note 11, at 509.

40 See, e.g., Devereux et al., supra note 11; and Devereux, supra note 39.

41 See, e.g., “UN Committee of Experts on International Cooperation in Tax Matters, Comments Submitted to OECD Secretariat on the ‘Unified Approach’” (Nov. 12, 2019).

42 Pillar 1 blueprint, supra note 7, at para. 176.

43 See Adam Kern, “Illusions of Justice in International Taxation,” 48 Phil & Pub. Aff. 151, 167 (2020).

44 Id. at 167-178.

45 See id. (developing this argument in detail).

46 Seminal contributions include Richard A. Musgrave, “The Voluntary Exchange Theory of Public Economy,” 52 Q.J. Econ 213 (1939); Paul A. Samuelson, “The Pure Theory of Public Expenditures,” 36 Rev. Econ. & Stat. 387 (1954); and Charles M. Tiebout, “A Pure Theory of Local Expenditures,” 64 J. Pol. Econ. 416 (1956).

47 See Michael Keen and Maurice Marchand, “Fiscal Competition and the Pattern of Public Spending,” 66 J. Pub. Econ. 33 (1997); and Yongzheng Liu, “Does Competition for Capital Discipline Governments? The Role of Fiscal Equalization,” 21 Int’l Tax & Pub. Fin. 345 (2014).

48 See pillar 1 blueprint, supra note 7, at paras. 810-881.

49 Id. at para. 809.

50 Id. at para. 705.

51 The delineation of business lines and the allocation of central costs to them may drop away if pillar 1 evolves in the direction of comprehensive scoping and away from the use of activity tests.

52 See United Nations Development Programme, “Human Development Index.”

53 See, e.g., Mitchell Kane, “Tax and Human Rights: The Moral Valence of Entitlements to Tax, Sovereignty, and Collectives,” in Tax, Inequality, and Human Rights (2019); Kern, supra note 43; Kern, “Optimal Taxation for the World” (May 1, 2021) (unpublished manuscript) (on file with the authors); and Kern, “Principles of International Taxation” (Nov. 21, 2020) (PhD thesis, Princeton University) (on file with the Princeton University Library System).

END FOOTNOTES

Copy RID