Menu
Tax Notes logo

What’s It to (the) U.S.? An Impact Analysis of Pillar 1 for U.S. Multinationals

Posted on Apr. 12, 2021
Kartikeya Singh
Kartikeya Singh

Kartikeya Singh is a principal in the transfer pricing practice of PwC’s Washington National Tax Services group. The author is sincerely grateful to Peter Merrill, Jeremiah Coder, and Will Morris for their comments and suggestions over the course of this analysis.

In this article, the author analyzes the effect of amount A, a key part of the OECD’s pillar 1 proposal, using firm-level and aggregated data to estimate how income from in-scope U.S. multinationals would be reallocated under the proposal and how the reallocation would affect groups of jurisdictions and MNEs.

The views expressed herein are solely those of the author and do not necessarily reflect those of PwC. All errors and views are those of the author and should not be ascribed to PwC or any other person. This content is for general information purposes only, and should not be used as a substitute for consultation with professional advisers.

Copyright 2021 PwC. All rights reserved.

In October 2020 the OECD inclusive framework on base erosion and profit shifting published the most extensive set of documents that have been released to date as part of its project to address the tax challenges from the digitalization of the economy. The project, which started in 2018, has the potential to significantly reshape the international rules that govern how multinational businesses are taxed.

In January 2019 the OECD grouped its work on the project into two pillars. Pillar 1 deals with the allocation of taxing rights between jurisdictions with the stated goal of assigning market jurisdictions a greater share of the residual profits earned by multinational enterprises. Thus, the work on pillar 1 involves revising the rules and nexus and income allocation for large MNEs that meet specified criteria (for example, those that engage in specific types of business activity and have high profitability). Pillar 2 deals with the development of rules that would subject large MNEs to a minimum tax on their global income, regardless of where that income may be reported for tax purposes under the applicable nexus and income allocation rules, including any changes brought about by pillar 1.

The documents issued last October include two blueprints that detail the proposals’ features and highlight various open issues and questions related to each pillar.1 The blueprints still leave numerous technical questions unanswered, and their release was followed by a public consultation process. However, at more than 250 pages each, the documents provide more extensive details regarding various design features and related considerations than anything else in the project’s history. Along with the blueprints on the two pillars, the OECD released a third document: an economic impact assessment of the two pillars.2 The OECD impact assessment provides estimates on the two pillars’ effects on the tax base and tax revenues of jurisdictions as well as on the level of investment by MNEs.

This article deals only with pillar 1 and focuses exclusively on amount A. For pillar 1, the OECD impact assessment presents estimates of the tax base that would be reallocated between jurisdictions as a result of the proposed rules on nexus and income allocation as well as the change in tax revenues resulting from the reallocation of the tax base between countries. The OECD impact assessment’s analysis of pillar 1 also focuses exclusively on amount A. The OECD’s analysis does not include estimates of pillar 1’s impact on individual countries (for example, tax revenue gains or losses). Instead, these estimates are presented at the level of country groupings. One set of groupings involves the following four categories: high-income jurisdictions, middle-income jurisdictions, low-income jurisdictions, and investment hubs. Further, the OECD impact assessment is presented at an aggregate level that does not break out the results by industry or based on where the MNEs are headquartered.

This article specifically analyzes the impact of pillar 1 on U.S.-headquartered MNEs. The analysis estimates how much of the pretax income — that is, the corporate income tax base — of U.S. MNEs would be reallocated between jurisdictions, relative to the status quo, as a result of the newly proposed nexus and income allocation rules related to amount A. In addition to presenting the reallocation impact in the aggregate — that is, across all U.S. MNEs — the impact is shown separately for five broad industry groups. In keeping with the OECD’s impact analysis, the income reallocation is also summarized across different country groupings (that is, high-income, middle-income, low-income, and investment hub countries).

The analysis follows the general analytical framework outlined in a previous article by this author, W. Joe Murphy, and Gregory J. Ossi,3 which remains applicable to the design features detailed in the blueprint on pillar 1. However, several updates and refinements are incorporated into the present analysis. First, unlike the prior article, which relied on 2016 data, the present analysis uses 2017 financial data for U.S. MNEs. Second, the earlier analysis relied solely on aggregated and anonymized data from country-by-country reports filed by U.S. MNEs with the IRS for 2016 as published by the IRS Statistics of Income program. In contrast, the primary data in the present analysis comes from two sources: the SOI’s aggregated 2017 CbC reporting data for U.S. MNEs4 and firm-level data for 2017 from Bureau Van Dijk’s (BvD) Orbis database.5 Other sources used in this analysis include the World Bank, the United Nations Conference on Trade and Development (UNCTAD), and the OECD. These additional data sources allow the present analysis to incorporate aspects and design features (for example, scope and revenue sourcing) that have only now been detailed in the pillar 1 blueprint and were not previously known with the same specificity.

A high-level summary of the results from the analysis is as follows.6

The estimated share of the overall tax base (that is, pretax income) that is expected to fall within the scope of amount A varies widely by industry7 — from approximately 25 to 95 percent.8 Amount A is estimated for in-scope companies based on consolidated financial statement data and calculated as 20 percent of earnings before tax that exceed 10 percent of revenues. Taking all industries together,9 the estimated aggregate amount A across U.S. MNEs totals $65.4 billion, which represents 8.2 percent of the in-scope pretax income.10 For the five industries that comprise firms with a non-trivial amount of in-scope pretax income, estimates for amount A range from under 3 to almost 25 percent of the total pretax income for the industry across all in-scope firms.11 The portion of the tax base that gets reallocated between jurisdictions because of the new nexus and income allocation rules related to amount A is smaller still and amounts to 3.7 percent of in-scope pretax income in the aggregate across all industries.12 Industry-specific estimates for the interjurisdiction reallocated income range from just over 0.5 percent to almost 6 percent of total in-scope pretax income for the given industry.

Focusing on countries, investment hub countries as a category cede the greatest share (85.6 percent) of the total tax base that gets reallocated because of amount A. Among the beneficiaries of the income reallocation, high-income countries as a category account for the greatest share (55.6 percent) while low-income countries receive the lowest share (6.5 percent). Finally, the incremental tax associated with the tax base reallocation between jurisdictions was estimated to equal 0.7 percent of in-scope pretax income across all industries. The industry-level effects ranged from less than 0.1 to 0.9 percent.

When viewed as a percentage of the pre-change tax base associated with in-scope U.S. MNEs, the income reallocation received by low-income countries represents the highest increase (22 percent) followed by middle-income countries (19.4 percent). The income reallocation received by high-income countries constitutes the smallest percentage of their respective pre-change tax base associated with the in-scope U.S. MNEs (3.1 percent). The investment hubs are estimated to cede 11.6 percent of their pre-change tax bases associated with in-scope U.S. MNEs because of amount A.

Section I of this article describes the different data sources and how these data are used in the analysis. Section II describes the different steps in the method and presents the results from each step. Section III presents the main outputs from the application of the method that were not covered in the previous section and discusses limitations and caveats that apply to the analysis. Section IV offers a conclusion.

I. Data Sources and Uses

A. SOI Summary of CbC Reporting Data

One of the data sets used in this analysis is the aggregated and anonymized 2017 CbC reporting information published by the SOI. This information is a summary of Form 8975 CbC reports and Form 8975 Schedules A, “Tax Jurisdiction and Constituent Entity Filings of U.S. MNEs,” for tax year 2017. The data cover the 1,575 reporting MNE groups that filed Form 8975 with the IRS for tax year 2017. The summary data published by the SOI shows aggregates of profit (or loss) before tax (PBT) for those MNE groups as well as how that PBT is distributed across specific tax jurisdictions where the MNEs reported that PBT. In addition to the PBT, aggregated data on other items that are required to be reported in the CbC reporting forms is contained in the SOI data. These include (related- and unrelated-party) revenues, income taxes paid and accrued in the current year, number of employees, and information on tangible assets. Again, these data are reported in the aggregate as well as by jurisdiction.

A useful feature of the SOI data is that the CbC reporting information is organized across seven major industry groups. Of these seven, the analysis focuses on five industry groups: manufacturing; wholesale and retail trade, transportation, and warehousing, referred to herein as “wholesale, retail, and related”; information; professional, scientific, and technical services, referred to as “professional services”; and management of companies and enterprises, and all other services (except public administration), referred to as “other services.” The remaining two — agriculture, forestry, fishing, hunting, mining, quarrying, oil and gas extraction, utilities, and construction, referred to as “agriculture, mining, and others”; and finance and insurance, real estate, and rental and leasing, referred to as “finance and related” — were assumed to be out of scope of amount A based on the carveouts and other criteria detailed in chapter 2 of the pillar 1 blueprint. This assumption was later validated using Orbis firm-level data described later. The breakdown of the SOI CbC reporting data by major industry groupings is shown in Table 1.

Table 1. SOI CbC Reporting Data: Number of MNE Groups by Major Industry Group

 

Number of MNEs

Agriculture, mining, and others

74

Manufacturing

565

Wholesale, retail, and related

312

Information

123

Finance and related

202

Professional services

127

Other services

172

Total

1,575

The role of the SOI CbC reporting data in the analysis was to identify the geographical distribution of U.S. MNEs’ reported profits, reported revenues, tangible assets, and employees within each of the five industry groupings covered by the analysis.

B. BvD Orbis

BvD Orbis is a global database containing financial statement and other firm-level data. The database includes listed and private firms, although many large unlisted U.S. firms are not covered by the database. The Orbis database was used to construct a sample of U.S. companies and to obtain 2017 financial data for companies within this sample for use in conjunction with the SOI CbC reporting data for purposes of the analysis. The Orbis sample was constructed by selecting U.S. companies with total worldwide third-party revenues for 2016 over $850 million — the same criterion used to determine whether a U.S. company would have been required to file Form 8975 with the IRS for tax year 2017. Based on that criterion, 1,269 companies were initially selected in the Orbis sample.

The primary purpose of the Orbis sample was to calculate amount A for each firm in the sample that would be considered in-scope for purposes of amount A. Unlike the previous article, which relied on aggregated CbC reporting data to calculate amount A for each industry grouping, the present analysis performs this calculation at the level of each firm and then aggregates this to determine industry-level estimates.

Importantly, the determination of whether and how much of a given industry’s PBT would be within the scope of amount A rules is based on a firm-level analysis. This was done by categorizing each firm as either in-scope — that is, as an automated digital services (ADS) company or a consumer-facing business (CFB) — or out of scope (that is, a company that is subject to a carveout, neither an ADS company nor CFB, and so forth). As discussed later, each firm in the Orbis sample was mapped to one of the seven major industry groupings in the SOI CbC reporting data so that data from those two sources could be used together in the analysis (and the firm-level scope categorization could be translated into an industry-level measure of in-scope revenues and so forth).

C. Other Data Sources and Uses

Other data used in the analysis were sourced from the World Bank; UNCTAD; the Bureau of Economic Analysis; the OECD; and a paper by Thomas R. Torslov, Ludvig S. Wier, and Gabriel Zucman.13 As described further in the following section, country-level data for 2017 from the World Bank on GDP per capita, population, and internet users (as a percentage of the population) were used to construct revenue-sourcing factors (RSFs) for the ADS businesses.14 Similarly, country-level data for 2017 on total consumption expenditure and consumption expenditure incurred by the government were used to derive country-level figures for private consumption expenditure to construct RSFs for companies in the CFB category.15

UNCTAD data on the ratio of inward foreign direct-investment stock to GDP were used to categorize countries in the analysis as investment hubs when the ratio exceeded 150 percent. Countries not categorized as investment hubs in the analysis were then categorized as high-income, middle-income, or low-income based on the World Bank’s income-level country classification.

Data on total employees and total employee compensation from the Bureau of Economic Analysis’s annual survey of U.S. MNEs for 2017 were used to construct estimates of cost-per-employee for the different industry groupings and used together with the information on the number of employees by jurisdiction from the SOI CbC reporting data.

Finally, data on the statutory corporate income tax (CIT) rate for different countries in the OECD Tax Statistics database were used to estimate the tax impact.16 The OECD data were supplemented by estimates on effective tax rates for select countries that Torslov, Wier, and Zucman provided in their paper.

II. Method and Analysis

A. Mapping Orbis Sample Into Industry Groups

As a first step in the analysis, each company in the Orbis sample was mapped into one of the seven major industry groupings used in the SOI CbC reporting data. This enabled the use of data from both sources: that is, firm-level data from the Orbis sample along with the aggregated industry-level SOI CbC reporting data. Each firm in the Orbis sample has a four-digit standard industrial classification (SIC) code assigned to it to designate its primary business activity. The mapping of each firm in the Orbis sample into one of the seven SOI industry groupings was undertaken using the four-digit SIC code as a bridge. While the SOI does not report the CbC reporting data at a level that is more disaggregated than the seven major industry groupings, the IRS does use this classification for other data (for example, other corporate tax statistics). Within these classifications, the IRS reports subsector industries that comprise each major industry grouping — for example, the manufacturing industry grouping is composed of 20 subsector industries.

For purposes of the mapping, the additional detail underlying the SOI’s seven industry groupings was used to map all four-digit SIC codes to one of the seven major industry groupings. Next, the primary four-digit SIC code reported in the Orbis database for each company in the Orbis sample was used to assign one of the seven major industry groupings to the specific company.

Table 2 summarizes the results of mapping the Orbis sample into the SOI major industry groupings. Table 2 compares the composition of the Orbis sample across the five industry groupings assumed to be in scope of amount A against that of the SOI CbC reporting data across the same industries.17 The composition of the Orbis sample in terms of industry groupings is fairly similar to that of the SOI CbC reporting data. The starkest exception is the professional services industry grouping, which is significantly underrepresented (in terms of the number of firms) in the Orbis sample relative to the SOI CbC reporting data.

Table 2. U.S. MNEs by Industry Grouping (2017): SOI CbC Reporting Data and Orbis Sample

 

Number of MNEs

 

SOI CbC Reporting Data

Orbis Sample

 

Number

Percentage of Total

Number

Percentage of Total

Manufacturing

565

43%

429

44%

Wholesale, retail, and related

312

24%

286

29%

Information

123

9%

98

10%

Professional services

127

10%

37

4%

Other services

172

13%

129

13%

Total (of the above)

1,299

100%

979

100%

Table 3 compares industry aggregates for (unrelated-party) revenues from the SOI CbC reporting data and the Orbis sample.18 As shown in Table 2, the Orbis sample covers fewer firms than those represented in the SOI CbC reporting data.19 A “coverage ratio” for the Orbis sample was calculated for each industry grouping — that is, the ratio of the aggregate third-party revenues for the industry grouping per the Orbis sample to the third-party revenues for the same industry per the SOI CbC reporting data.20 For example, the coverage ratio for the information industry grouping is 86 percent — that is, the total revenues for the subset of firms that comprise the Orbis sample within the information industry represents 86 percent of the total across firms within the same industry in the SOI data. The coverage ratio is the highest for the manufacturing industry grouping and lowest for other services. Even though the professional services industry is represented by very few firms in the Orbis sample relative to the SOI CbC reporting data, the coverage ratio is 78 percent for this industry.21

Table 3. Revenues and Coverage Ratio (2017): SOI CbC Reporting Data and Orbis Sample

 

Unrelated-Party Revenue (millions USD)

Coverage Ratio

 

SOI CbC Reporting Data

Orbis Sample

Orbis/SOI

Manufacturing

$5,084,838

$4,581,555

90%

Wholesale, retail, and related

$4,496,888

$3,655,172

81%

Information

$1,247,526

$1,073,183

86%

Professional services

$326,981

$255,150

78%

Other services

$1,183,770

$678,982

57%

Overall

$12,340,004

$10,244,041

83%

B. Scope

In addition to mapping each firm in the Orbis sample into one of the seven SOI industry groupings as described above, each firm in the sample was categorized as either in-scope or out-of-scope for amount A. A firm was categorized as in-scope if it met the definition of either an ADS business or a CFB under the activities test part of the scope determination detailed in the pillar 1 blueprint and if the exclusions and carveouts in the blueprint were not viewed as being applicable to the firm.22 As a first step, the primary SIC code was used to determine whether the firm would be within or outside the scope of amount A based on the description of ADS and CFBs. Second, exclusions and carveouts from the scope of amount A, which are detailed in the pillar 1 blueprint, were considered. Third, a manual review of the scope categorization from step 1 was done to further analyze the results for specific companies. In this review, the first-step categorization (based on the SIC codes) was reversed if the initial outcome was not found to be accurate based on a more specific review of the company’s operations using publicly available information (that is, beyond the primary four-digit SIC code assigned to it).

The firm-level scope categorization confirmed the initial assumption that firms in the agriculture, mining, and others, and the finance and related industry groupings would fall outside the scope of amount A. No firm in the Orbis sample that was mapped (in the prior step) into the agriculture, mining, and others industry group was categorized as being in-scope based on the three-step process described above. Of the 221 firms that were mapped into the finance and related industry grouping, only five were categorized as falling within the scope of amount A. These five firms accounted for approximately 0.2 percent of the aggregate PBT across all 221 firms in the Orbis sample firms within this industry. Furthermore, none of these of these five firms had a PBT-to-revenue ratio higher than 10 percent.

The firm-level scope categorization together with the industry mapping was used to generate industry-level figures for the percentage of total industry revenues as well as total industry PBT estimated to be within the scope of amount A. Thus, the “in-scope PBT percentage” for the manufacturing industry grouping represents the total PBT for in-scope firms in the Orbis sample within the manufacturing industry grouping as a percentage of the total PBT of all manufacturing firms within the Orbis sample. The “in-scope revenue” percentage has an analogous meaning regarding revenues. The in-scope percentages for the five major industry groupings relevant to the analysis are shown in Table 4. Unsurprisingly, the in-scope percentage (for both revenues and PBT) is the highest for the information industry grouping, reflecting the presence of firms that meet the scope criteria for ADS businesses within this industry.

Table 4. In-Scope Revenue and PBT by Industry Grouping (2017): Orbis Sample

 

In-Scope Percentage — Revenues

In-Scope Percentage — PBT

Manufacturing

47%

59%

Wholesale, retail, and related

46%

41%

Information

93%

95%

Professional services

47%

75%

Other services

44%

25%

Total (for the above)

51%

61%

C. Amount A

Using the Orbis data for 2017, a firm-level calculation was performed to determine amount A for each in-scope firm. The calculation relied on each in-scope firm’s 2017 PBT and (third-party) revenues as reported in the Orbis database. The profitability threshold, which was used to identify routine profits and isolate the residual profits subject to reallocation, was assumed to be 10 percent for purposes of this calculation.23 The reallocation percentage — the portion of the deemed residual profit allocable to market jurisdictions — was assumed to be 20 percent for purposes of this calculation. The profitability threshold and the reallocation percentage together determine what the pillar 1 blueprint calls the “allocable tax base” under amount A. This figure for each firm (indexed by i) is zero for any firm outside the scope of the new taxing right and is given by the equation below for an in-scope firm, with “PBT margin” referring to the ratio of the firm’s PBT to third-party revenues (denoted by Revenuei).

Equation 1

Equation 1

The aggregate amount A for each of the five in-scope industry groupings (with each industry grouping indexed by k) was calculated based on the firm-level calculation described above across all firms in the Orbis sample. This is shown in the equation below, with Ik denoting the total number of firms mapped to a given industry grouping k within the Orbis sample.

Equation 2

Equation 2

As shown in Table 2, the SOI CbC reporting data comprises more firms than are covered by the Orbis sample (that is, the latter is at best a subset of the former). Consequently, the industry-level estimates for amount A derived from the Orbis sample were rescaled (denoted by the “hatted” version of the same variable) using the coverage ratio for each industry shown in Table 3 (and denoted by CRk in Equation 3).

Equation 3

Equation 3

Table 5 shows the aggregate industry-level amount A estimates derived from the Orbis sample and rescaled based on the coverage ratio for each industry. Unless specifically noted otherwise, these rescaled amount A estimates are used in the rest of the analysis.

Table 5. Aggregate (Rescaled) Amount A by Industry Grouping (2017)

 

Aggregate Amount A: Orbis Sample (millions USD)

Coverage Ratio

Rescaled Aggregate Amount A (millions USD)

Manufacturing

$29,670

90%

$32,929

Wholesale, retail, and related

$3,004

81%

$3,695

Information

$19,016

86%

$22,106

Professional services

$3,879

78%

$4,971

Other services

$994

57%

$1,732

Total (of the above)

$56,562

 

$65,433

D. Reallocation Amount

The final steps in the analysis were to estimate the jurisdiction-level impact that would result from the allocation of the amount A estimated for each industry grouping. This required estimating which jurisdictions would receive a share of the allocable tax base (that is, amount A) and how much. The quantification of the overall impact also required estimating which jurisdictions would have to cede a part of their tax base (and how much) to make up the total amount A received by the first group of jurisdictions.

To the extent that a jurisdiction is the location of the final consumers of an in-scope company’s product or services (for CFBs) or is the location of users of online platforms, viewers of online advertising, recipients of digital services, and so forth (for ADS businesses), the jurisdiction will be characterized as a market jurisdiction. As a market jurisdiction, a given country will have the relevant company’s revenues sourced to it under the revenue-sourcing principles applicable to the relevant business (that is, ADS or CFB).24 Other things being equal, a country will gain taxable income in its capacity as a market jurisdiction via a share of the amount A calculated for each company and each industry. The share of amount A received by a country in its capacity as a market jurisdiction will be proportionate to the revenues sourced to the jurisdiction under the principles for ADS businesses or CFBs, as applicable.

A country will cede taxable income, other things being equal, if it is a “relieving jurisdiction” as described in the pillar 1 blueprint. In turn, a relieving jurisdiction is the location of a paying entity — that is, an entity within the MNE group that is an intended claimant of residual profits (as the term is used in a transfer pricing context) under the arm’s-length principle. An entity’s residual profit potential under the arm’s-length principle depends on the functions performed, assets owned and used, and risks borne by the entity. Under the application of the arm’s-length principle, high residual profits — also referred to as non-routine profits — are generally claimed by entities that own or have been assigned rights to intangible property, perform related functions, and bear associated risks. Thus, the paying entities and the associated relieving jurisdictions will generally be those that have above-normal profits (that is, in excess of routine returns).

The overall effect on a given jurisdiction from the introduction of amount A will be the net impact of the two effects described above. A given quantum of amount A for an MNE, by itself, is not sufficient to assess the reallocation effect of the new nexus and income allocation concepts. If the relieving jurisdictions are the same jurisdictions that have a claim on amount A in their capacity as market jurisdictions, the reallocation effect will be minimal. This will be the case for MNEs with intercompany pricing arrangements that already assign significant residual profits to market jurisdictions. Said differently, whether a jurisdiction gains tax base under amount A depends on whether its share of the market for in-scope companies is greater than its share of non-routine profits under the arm’s-length principle reported for those in-scope companies.

If, on the other hand, the MNE’s non-routine profits are allocated under the arm’s-length principle to jurisdictions that are not major market jurisdictions, the reallocation effect of the new rules will be more significant. In the extreme, when the relieving jurisdictions have no claim on amount A — because they are not market jurisdictions based on the revenue-sourcing and nexus rules specifically relevant for amount A — the overall income reallocation will be equal to the amount A calculated for the MNE group (or segment, as applicable). Thus, the income reallocation that an MNE will experience as a result of the rules governing amount A will lie between zero and the full quantum of the amount A calculated for the group. The greater the extent to which an MNE’s intercompany pricing arrangements allocate income to market jurisdictions under the status quo, the less pronounced the income reallocation effect of the proposed new rules will be and vice versa.

The estimation of the income reallocation effect resulting from the amount A aggregates calculated for each of the five industry groupings shown in Table 5 is described below. First, the jurisdictions that are estimated to receive a share of the aggregate amount A for each industry in their capacity as market jurisdictions are analyzed. Next, the jurisdictions that are estimated to have to cede income to make up the amount A allocations to the market jurisdictions (to avoid double taxation) are identified, along with estimates of the quantum of income each relieving jurisdiction will need to cede. Finally, the net effect for each jurisdiction based on the two effects described above is calculated.

1. Market Jurisdictions

A jurisdiction’s claim to a share of the overall amount A as calculated for an MNE (or a segment of that MNE) will depend on two things — first, whether the MNE is deemed to have nexus with the jurisdiction under the new concept of nexus specifically applicable for purposes of amount A; and second, assuming the nexus threshold is met, the share of the overall amount A allocable to an eligible market jurisdiction will be based on the proportion of the business’s revenues sourced to the jurisdiction using the applicable revenue-sourcing criteria in the blueprint.

For purposes of this analysis, the nexus and revenue-sourcing steps to determine eligible market jurisdictions are combined into one — that is, the attribution of revenues to jurisdictions based on an approximation of the revenue-sourcing criteria described in the pillar 1 blueprint. For the present analysis, which deals with the aggregate amount A estimated for each industry grouping, it was assumed that jurisdictions that have significant revenues sourced to them under the relevant criteria for amount A will also meet the other applicable criteria for nexus.

The SOI CbC reporting data for each industry grouping shows the aggregated unrelated-party revenues for that industry grouping as well as the revenue amounts across jurisdictions as recorded for tax or (more often) for statutory or management reporting purposes. The distribution of revenue by jurisdiction as reported in a company’s CbC report — or in any other tax or statutory financial document — does not reflect the destination-based sourcing principles of amount A, which require tracing revenues to the location of users, viewers, consumers, and so forth.

To approximate the outcome — at an aggregate level for each industry grouping — of the revenue-sourcing principles laid out in the pillar 1 blueprint, the analysis relied on country-level indicators that are expected to correlate with specific revenue-sourcing criteria applicable for ADS businesses and CFBs. For ADS businesses, this assumes that the magnitude of revenues sourced to a jurisdiction for amount A purposes — that is, based on the users of a company’s online platform, viewers of online advertising served on its platform, and so forth — will be positively correlated with the number of internet users in the jurisdiction as well as the average income level of those internet users. For CFBs, the assumption is that the magnitude of revenues sourced to a jurisdiction for amount A purposes — that is, based on final consumers of a company’s goods or services — will be positively correlated with the final household (that is, nongovernmental) consumption expenditure of the given country.25

Based on the above premise, an RSF was derived for each type of in-scope business (that is, ADS or CFB) that was used to allocate the aggregate industry revenues in the SOI CbC reporting data to individual jurisdictions. Given that these data are for U.S. MNEs, the share of aggregate industry revenues as shown for the United States in the CbC reporting information was assumed to be a reasonable approximation for what would be sourced to the jurisdiction under the destination-based sourcing principles of amount A. In other words, the RSFs for ADS and CFBs were used only to reallocate non-U.S. revenues (per the SOI CbC reporting data) among non-U.S. jurisdictions. Thus, the share of aggregate industry revenues sourced to the United States for amount A purposes was taken to be the same amount that is shown for in the SOI’s aggregated CbC reporting data.26

Thus, the RSF for a given jurisdiction (indexed by j), other than the United States, for an industry (indexed by k) that comprises ADS businesses is shown in the equation below (InUsersj denotes the total number of internet users in the country and GDPPCj is GDP per capita for the country). All data are for 2017.

Equation 4

Equation 4

Similarly, the equation below shows the RSF for a given jurisdiction (other than the United States) for an industry that comprises CFBs, with HHCEj denoting the total private household consumption expenditure for the country.27

Equation 5

Equation 5

Finally, the aggregate (non-U.S.) revenues based on the SOI CbC reporting data for a given industry was allocated to each jurisdiction based on the relevant RSF applicable to that industry — that is, ADS or CFB. This is shown in the equation below, with Revenuekj denoting the revenue for industry k sourced to jurisdiction j and Revenue(NonUS)k denoting the aggregate revenues for industry k across all jurisdictions other than the United States.

Equation 6

Equation 6

Thus, the aggregate amount A estimated for each industry in Table 5 was allocated to different jurisdictions based on the total revenues for the industry sourced to each jurisdiction. The revenues sourced to the United States were taken to be the amount shown for the country using the SOI CbC reporting data for the relevant industry. The revenues sourced to all other jurisdictions were calculated using Equation 6 (with the RSFs being determined by Equation 4 for ADS businesses and Equation 5 for CFBs).

The RSFs for ADS businesses were used for two industry groupings — information and professional service. The RSFs for CFBs were used for three industry groupings — manufacturing; wholesale, retail, and related; and other services. The use of the RSFs for ADS or CFBs in relation to a given industry was based on the makeup of the in-scope firms mapped to the industry. Table 6 shows the revenues sourced to the different country groupings for industries comprising ADS and CFBs, respectively. For comparison, the share of revenues for each country grouping as originally reported in the SOI CbC reporting data is also shown in the table.

Table 6. Share of Revenues Sourced to Jurisdictions (ADS and CFB): Summary by Country Grouping

 

Reported Share of Global Unrelated-Party Revenue

Share of Sourced Revenue — ADS

Share of Sourced Revenue — CFB

High Income

85.7%

85.3%

86.3%

Middle Income

5.3%

10.8%

9.6%

Low Income

0.8%

2.6%

2.8%

Investment Hubs

8.2%

1.3%

1.3%

Finally, the amount A for a given industry that a country stands to receive in its capacity as a market jurisdiction was calculated as the product of the industry-level aggregate amount A (shown in Table 5) and the share of the aggregate industry revenues sourced to that jurisdiction based on the above as shown in Equation 7.

Equation 7

Equation 7

2. Relieving Jurisdictions

The last step in the analysis was to estimate — for each industry — the amount of income that each relieving jurisdiction would need to cede to make up the total amount A allocated to the eligible market jurisdictions. This part of the analysis draws from the framework for the elimination of double taxation described in the pillar 1 blueprint.28 The blueprint describes a four-step process to identify paying entities as those entities within the MNE group that would have to cede taxable income to make up the amount A allocated to market jurisdictions under the new taxing right. The jurisdictions where the paying entities are resident — that is, the relieving jurisdictions — will cede a part of their tax base corresponding to the tax base gained by the market jurisdictions in the amount A allocation. The intent of the framework in the pillar 1 blueprint is to identify entities and associated jurisdictions that claim residual profits (in the conventional transfer pricing sense) — that is, profit above routine returns attributable to people functions and tangible capital — under the existing income allocation paradigm (that is, based on the separate-entity accounting concept and the arm’s-length principle).

To identify the paying entities (and the corresponding relieving jurisdictions), the framework described in the blueprint consists of four steps, beginning with an activity test (section 7.2.2 of the blueprint) and a profitability test (section 7.2.3 of the blueprint).

The activity test will rely on both positive and negative indicia with the overarching objective of identifying paying entities as, according to paragraph 585 in chapter 7 of the blueprint, “the member or members of an MNE group (or segment) that perform functions, use or own assets and/or assume risks that are economically significant, for which they are allocated residual profits.” The general principle of the activity test “draws on a series of concepts that already form part of transfer pricing today.” The goal of the activity test is to identify the entities within an MNE group that are the intended claimants of residual profits under the arm’s-length principle in their capacity as (beneficial) owners of rights to IP, principal entities, and other risk-bearing entities that satisfy the substance requirements (that is, in terms of having adequate people functions) necessary for their status as residual profit claimants to be respected under the guidance applicable to the arm’s-length principle.

The second step is a profitability test. The intent is to ensure that the entities and associated jurisdictions identified in the first step in fact have the capacity — in terms of sufficient residual profits — to bear the amount A liability or alternatively, to further narrow down the list of entities identified in the first step to those that will be the paying entities. The framework envisions a profitability test that will identify entities that earn income exceeding a fixed return on tangible assets and payroll expenses — presumably under the premise that the fixed return is a reasonable approximation of profits for routine activities (for example, services or manufacturing) that rely on tangible assets or labor (as opposed to residual profits attributable to intangibles, risk-bearing, and so forth).

The next two steps — the market connection priority test (section 7.2.4 of the blueprint) and the pro rata allocation (section 7.2.5 of the blueprint) — can be viewed as refining the outcome from the first two steps. The intent of the market connection priority test is to match the paying entities identified in steps 1 and 2 (for example, the principal company within an MNE group that has responsibility for specific markets) with specific market jurisdictions (for example, in that same geographic region) based on a sufficient connection between a given paying entity and a specific set of market jurisdictions — for example, the IP rights possessed by the paying entity for a particular territory and the regions of the select group of market jurisdictions. If the paying entities that have been identified as connected to a set of market jurisdictions do not have sufficient residual profits (above the routine return on tangible assets and payroll expenses), then the remaining amount A allocation — that is, in excess of the connected paying entities’ ability to pay based on their residual profits — will be borne by other paying entities and associated relieving jurisdictions identified in the first two steps in proportion to their residual profits.

For purposes of this analysis, the outcome of the process described above was approximated using the SOI CbC reporting data to identify the jurisdictions that will serve as relieving jurisdictions within each industry.29 First, for each industry grouping, the amount of non-routine profits in each jurisdiction was calculated as the difference between the total profits reported in the jurisdiction and an estimate of normal profits attributable to the tangible assets and employees in that jurisdiction.30 Jurisdictions accounting for a significant share (assumed to be 3 percent) of worldwide non-routine profits for that specific industry grouping were then identified as relieving jurisdictions. A significance threshold was used to reduce the impact of noise in the estimation and attribution of non-routine profits across jurisdictions based on the observed distribution of PBT. Application of the threshold identifies the jurisdictions that account for a non-trivial share of estimated global non-routine profits. These jurisdictions are more likely to be the intended recipients of income based on the arm’s-length principle and, therefore, the relevant relieving jurisdictions under the principles and criteria in the pillar 1 blueprint.

Second, each relieving jurisdiction within the specific industry grouping is assumed to cede a portion of its non-routine profits to constitute the total amount A for that industry. The amount of relief provided — that is, reduction in the tax base experienced — by each relieving jurisdiction is calculated in proportion to the jurisdiction’s share of total worldwide non-routine profits for the industry.31 This is shown in the equation below for each jurisdiction that is identified as a relieving jurisdiction for industry k. In the equation, the set of relieving jurisdictions for the industry is denoted by  (which is a subset of all relevant jurisdictions Jk for the industry); Relkj denotes the relief provided (that is, the tax base ceded) by jurisdiction j in relation to industry k; and Non-Routinekj denotes the non-routine profits calculated for that jurisdiction as described above.

Equation 8

Equation 8

For any jurisdiction not identified as a relieving jurisdiction for a given industry grouping (that is, if ), Relkj is zero.

3. Overall Impact

The overall impact on a given jurisdiction’s pretax income in relation to a given industry grouping is a function of two effects — its allocation of amount A in its capacity as a market jurisdiction less any income it has to cede in its capacity as a relieving jurisdiction. This is shown in Equation 9 (ΔPBTkj denotes the change in jurisdiction j’s pretax income in a given industry k), which is derived from Equation 7 and Equation 8.

Equation 9

Equation 9

Since the overall tax base for a given industry should remain unchanged — this is among the stated objectives of pillar 1 — the sum total of Equation 9 across all relevant jurisdictions for industry k will be zero. Expressed differently, the overall change in pretax income across all jurisdictions that experience an increase in their tax base — that is, for which the first term on the right-hand side of Equation 9 is greater than the second — will exactly equal the total reduction in pretax income across the set of jurisdictions that experience a reduction in their tax base. Denoting the first set of jurisdictions in a given industry k by J(+)k and the second set by J(-)k, Equation 10 shows the total income reallocation for the industry (denoted by IRk) as a result of the proposed rules on amount A.

Equation 10

Equation 10

Table 7 shows the reallocated income by industry grouping together with the in-scope pretax income. For reference, the total amount A for each industry grouping is also shown.

Table 7. Income Reallocation, Amount A, and In-Scope PBT by Industry Grouping

 

Amount A (millions USD)

Amount A as Percentage of In-Scope PBT

Income Reallocation (millions USD)

Income Reallocation as Percentage of In-Scope PBT

Income Reallocation as Percentage of Amount A

Manufacturing

$32,929

8.2%

$17,977

4.5%

55%

Wholesale, retail, and related

$3,695

3.9%

$822

0.9%

22%

Information

$22,106

9.9%

$9,217

4.1%

42%

Professional services

$4,971

24.8%

$1,187

5.9%

24%

Other services

$1,732

2.7%

$361

0.6%

21%

Total (of the above)

$65,433

8.2%

$29,564

3.7%

45%

The total income reallocation across all industries is less than 50 percent (that is, $29.6 billion) of the amount A calculated ($65.4 billion) for the same industries. Said differently, for every dollar of amount A allocable to market jurisdictions across the five industry groupings, only 45 cents actually get reallocated among jurisdictions. The remaining 55 cents are retained within the jurisdictions because of their additional claims on pretax income in their capacity as eligible market jurisdictions offsetting their obligations to cede income as relieving jurisdictions.

This income reallocation effect is strongest for the manufacturing industry (that is, 55 cents of income reallocation for each dollar of amount A), which is also the industry with the highest absolute magnitude for amount A and the income reallocation. This is followed by the information industry grouping (42 cents of income reallocation for each dollar of amount A), which also is second in terms of absolute magnitudes. The three other industries have much smaller income reallocation effects.

III. Results

A. Impact on Countries’ Tax Bases

The results presented in Table 7 show that the introduction of new nexus and income allocation rules related to amount A will result in an estimated $29.6 billon of income being reallocated between jurisdictions across all industry groups. The figures below show how this income reallocation affects different country groupings: In particular, they show who benefits from the income reallocation, and by how much, and who bears the burden of the income reallocation in the form of a reduction in their tax bases.

Figure 1 shows the impact of the income reallocation — that is, the increase or decrease in the tax base as a result of amount A associated with U.S. MNEs — at the level of each country grouping as a percentage of that group’s pre-change tax base that is specifically related to in-scope U.S. MNEs. As shown, low-income, middle-income, and high-income countries (as respective groups) gain from the income reallocation, and the investment hub countries are the only group that loses income because of the reallocation. When measured as a percentage of each group’s pre-change tax base (again, related to the in-scope U.S. MNEs), the increase in tax base is most significant for low-income countries (22 percent), followed by middle-income countries (19.4 percent), and high-income countries (3.1 percent). Investment hubs stand to lose 11.6 percent of their pre-change tax bases (related to in-scope U.S. MNEs) because of the reallocation brought about by amount A. This pattern of gains and losses is very similar to what is shown in the OECD impact assessment.32 It is worth noting that individual country results within a country grouping can be directionally different from that for the overall country grouping. For example, there may be countries in the high-income group that, unlike the group overall, experience reductions in their tax bases because of the income reallocation effect of amount A for a given industry.

Figure 1. Income Reallocation by Country Grouping (2017 All Industries): As a Percentage of Pre-Change Tax Base

Figure 2 paints a slightly different picture. This figure shows how the different groups of countries share in the gains and losses as a percentage of the total income reallocation — that is, how much each group receives (or cedes) as a percentage of every dollar that gets reallocated across all industries because of amount A. As expected, the investment hubs (as a group) bear the overwhelming burden of the reallocation, accounting for 85.6 cents for every dollar that gets reallocated. (The remaining 14.4 cents for every dollar of income that gets reallocated are ceded by countries from the three other groupings.) Among the winners, however, when viewed in terms of who claims the biggest share of the reallocated tax base, it is the high-income group that claims 55.6 cents of every dollar that gets reallocated between jurisdictions. This is followed by the middle-income group, which claims 23.5 cents per dollar, while the low-income group receives only 6.5 cents. That this small share of the income reallocation results in the most significant increase as a percentage of the pre-change tax base for low-income countries reflects the fact that this grouping accounts for the smallest share of the tax base under the status quo.

Figure 2. Income Reallocation by Country Grouping (2017 All Industries): Share of Total Income Reallocation

B. Incremental Tax Cost for U.S. MNEs

The final part of the analysis estimates the tax impact of the income reallocation. Pillar 1 is not intended to change the overall tax base related to in-scope MNEs but rather to reallocate a portion of that tax base among jurisdictions. The tax impact of this income reallocation will depend on the tax rate differentials between the jurisdictions that receive the income reallocation in their capacity as market jurisdictions and those that must cede that reallocation as relieving jurisdictions. Even more so than other parts of the analysis, the tax impact analysis is intended to be directional — that is, to estimate the rough order of magnitudes involved instead of precise revenue or cost estimates. In that spirit, this calculation is undertaken at a simple level.

The tax rate applicable to any additional income received or ceded by a jurisdiction that is not an investment hub is assumed to be the statutory CIT rate for that jurisdiction. The statutory CIT rates used in this calculation are sourced from the OECD Tax Database. For investment hub countries — several of which are known to provide preferential tax rates lower than their statutory tax rates to MNEs (via tax rulings and so forth) — an estimate of the effective tax rate for that jurisdiction was applied to any additional income received or ceded by it. Torslov, Wier, and Zucman was used as the source for these estimates.33

Table 8 shows the incremental tax impact by industry grouping and in the aggregate estimated to result from the income reallocation between jurisdictions because of amount A.

Table 8. Income Reallocation From Amount A: Tax Impact by Industry Grouping

Industry Grouping

Tax Impact of Income Reallocation (millions USD)

In-Scope PBT (millions USD)

Tax Impact as Percentage of In-Scope PBT

Manufacturing

$3,619

$401,582

0.9%

Wholesale, retail, and related

$70

$93,720

0.1%

Information

$1,914

$222,186

0.9%

Professional services

$59

$20,030

0.3%

Other services

$43

$63,596

0.1%

All Industries

$5,706

$801,114

0.7%

The effect of the incremental change in tax costs resulting from the introduction of amount A across industries amounts to 0.7 percent of the pre-change in-scope tax base. The impact is highest for the manufacturing and information industry groupings — approximately 0.9 percent. These two industry groupings account for approximately 97 percent of the incremental tax across all industries. In contrast, they account for just over 84 percent of the total amount A and 92 percent of the income reallocation income across all industries.

C. Caveats and Limitations

The analysis and results presented in this article are subject to several limitations. One immediate caveat is that the analysis focuses specifically on U.S. MNEs and all estimates involve only those MNEs. The ultimate impact of amount A on jurisdictions will depend on the income reallocation effects of all in-scope MNEs. Several other limitations of the analysis are likely to be present in any analysis aimed at estimating the impact of the proposed rules on pillar 1 (for example, they also affect, to varying degrees, the OECD impact assessment).

The analysis relies on data from 2017 and predates the changes to the U.S. tax system implemented as part of the Tax Cuts and Jobs Act. The snapshot of U.S. MNEs’ worldwide footprint (for example, in terms of reported pretax income) as reflected in the 2017 data and used as the basis for the analysis may not be representative of what more recent data may depict. Similarly, the 2017 data may not fully reflect the impact of the measures adopted as part of changes and refinements to the international tax system under the BEPS project that the OECD concluded in 2015.

The analysis is also not able to account for some aspects of the proposed rules discussed in the pillar 1 blueprint, most notably involving segmentation (specifically, when amount A is determined and allocated for in-scope segments of a firm rather than on a consolidated basis). For sourcing firm revenues to the location of consumers or users, the analysis relies on macro (country-level) data in the absence of firm-level data. Also, the analysis assumes a value of 10 percent for the profitability threshold and 20 percent for the reallocation percentage, but these parameters for amount A have not been formally set. The analysis does not account for any version of loss carryforward rules for amount A discussed in section 5.4 of the pillar 1 blueprint.

Finally, the analysis is static: It does not account for behavioral responses on the part of taxpayers or taxing authorities. Furthermore, the impact of the proposed rules is estimated relative to the status quo rather than an alternative global regime that might exist if pillar 1 is not adopted (for example, a world in which digital services taxes are ubiquitous and retaliatory trade measures might be adopted).

IV. Conclusions

This article analyzes the impact of amount A in the pillar 1 proposal on U.S. MNEs, with specific attention devoted to the proposals involving scope, nexus, income allocation, and prevention of double taxation. The analysis relies on several data sources, both firm-level and aggregated, and it analyzes the effects on different industry and country groupings. While the analysis finds variation in the impact on different industries and countries, ultimately the overall impact is quite modest.

For all in-scope U.S. MNEs, this study estimates that the amount A — or, to use the pillar 1 blueprint’s terminology, the allocable tax base under amount A — totals about $65.4 billion. By way of comparison, considering the corresponding total for amount A across all (that is, U.S. and non-U.S.) in-scope MNEs, the executive summary of the OECD’s impact analysis estimates that “taxing rights on about USD 100 billion could reallocated to market jurisdictions under the Pillar One rules.”34

Importantly, for every dollar of the total amount A allocable tax base for in-scope U.S. MNEs across the five industry groupings, the analysis finds that only 45 cents actually are reallocated between jurisdictions. The remaining 55 cents are retained in the same jurisdictions to which income is being allocated under existing nexus and income allocation rules. For these 55 cents, the jurisdictions’ claim to pretax income in their capacity as eligible market jurisdictions is offset by their obligation to cede income as relieving jurisdictions under the proposed amount A rules.

The analysis finds that as a group investment hub countries lose income, while high-income, middle-income, and low-income countries — as respective groups — all benefit from the reallocation. This is similar to the OECD’s conclusions in its impact assessment. The study contextualizes these gains in two different ways. First, when the amounts reallocated to each group are expressed as a percentage of the countries’ pre-change tax base from in-scope U.S. MNEs, the low-income countries appear to benefit the most, with the amount of income reallocated to them representing 22 percent of their pre-change tax base from in-scope U.S. MNEs. The amount of income reallocated to high-income countries represents the smallest percentage (3.1 percent) of the countries’ pre-change tax base from in-scope U.S. MNEs. This ordering is similar to that in the OECD impact assessment, which shows that when viewed as a percentage of the respective countries’ GDP, the income reallocated to low-income countries under amount A rules represents the biggest amount while that for high-income countries is the smallest.

Alternatively, when the gains for each group of countries are expressed as a share of the total amount of income reallocation (that is, $29.6 billion), the high-income countries account for the highest share (55.6 percent) of the reallocated income while the low-income countries account for the smallest share (6.5 percent). The OECD impact assessment does not consider how the three groups of countries share in every dollar of income that gets reallocated between jurisdictions because of amount A.

As discussed, the analysis is subject to various limitations that are not trivial. Nonetheless, the analysis can be a helpful addition to ongoing efforts to better understand the impact of the proposed rules and concepts underlying pillar 1. Furthermore, the data and analytical framework underlying the analysis may be useful in analyzing alternative proposals aimed at refining and simplifying pillar 1.35

FOOTNOTES

1 OECD, “Tax Challenges Arising From Digitalisation — Report on Pillar One Blueprint” (Oct. 14, 2020); and OECD, “Tax Challenges Arising From Digitalisation — Report on Pillar Two Blueprint” (Oct. 14, 2020).

3 Kartikeya Singh, Murphy, and 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.

4 IRS, “SOI Tax Stats — Country by Country Report” (last accessed Oct. 15, 2020).

5 Bureau van Dijk, Orbis (as updated in 2020).

6 Given the different assumptions and approximations underlying the analysis, any results should ideally be presented in the form of ranges. The article presents point estimates for simplicity. However, these should be viewed as approximations intended to convey the direction of impact and orders of magnitude involved.

7 Each industry in this case refers to an industry group that corresponds to the industry categorization used in the IRS SOI data — for example, manufacturing or information. The SOI categorization covers seven industry groups. In turn, each industry group in this analysis represents a collection of four-digit SIC codes.

8 This range is for those industries that account for a non-trivial number of firms — and thus a non-trivial amount of pretax income — that will likely fall within the scope of amount A. As discussed herein, the figure is zero (or close to zero) for two industry groupings.

9 Again, this is only taking into account industries that have a non-trivial number of in-scope firms and associated amounts of in-scope pretax income. For those industries, the aggregate amount A represents less than 5 percent of total pretax income — that is, when the total income is for in-scope firms as well as those outside the scope of amount A.

10 The estimates for amount A in this analysis correspond to what chapter 6 of the pillar 1 blueprint calls the allocable tax base, determined based on the application of the profitability threshold (for example, 10 percent) and the reallocation percentage (for example, 20 percent).

11 The corresponding range for amount A as a percentage of the total income across all — in-scope and out-of-scope — firms in an industry ranges from less than 1 percent to almost 19 percent.

12 As discussed herein, the net amount of income reallocation is lower than the amount A because some of the amount A is allocable to the very same jurisdictions that will have to cede some income — in their capacity as relieving jurisdictions — that is currently allocated to them under existing rules.

13 Torslov, Wier, and Zucman, “The Missing Profit of Nations,” National Bureau of Economic Research Working Paper No. 24701 (June 2018; revised Apr. 2020).

14 World Bank, “GDP Per Capita, PPP (Current International $)” (last accessed Mar. 1, 2021); World Bank, “Population, Total” (last accessed Mar. 1, 2021); and World Bank, “Individuals Using the Internet (% of Population)” (last accessed Mar. 1, 2021).

15 World Bank, “Final Consumption Expenditure (Current US$)” (last accessed Mar. 2021); and World Bank, “General Government Final Consumption Expenditure (Current US$)” (last accessed Mar. 2021).

16 OECD, “OECD.Stat” (last accessed Mar. 2021).

17 Firms mapped to two of the seven industry groupings were assumed to be outside the scope of amount A, namely agriculture, mining, and others; and finance and related. This assumption was validated further based on a firm-level scope categorization.

18 Throughout the analysis, all items (for example, PBT and revenues) under the “stateless” category in the SOI CbC reporting data were excluded and industry aggregates were revised downward to reflect this exclusion. Items reported as stateless in CbC reporting data are susceptible to double-counting as these may stem from flow-through entities that are also accounted for elsewhere in the data. Excluding stateless data from this analysis is consistent with what users of CbC reporting data have done for purposes of research and is consistent with the approach adopted in the OECD impact assessment.

19 One possible reason is the underrepresentation of unlisted firms in the Orbis sample compared with the SOI CbC reporting data.

20 The coverage ratio was defined based on revenues instead of PBT given the greater potential for double-counting of profits (for example, based on how dividends may get reported) in the CbC reporting data.

21 This is surprisingly high given that the Orbis sample covers only 37 firms within this industry compared with 127 in the SOI data.

22 Pillar 1 blueprint, chapter 2, sections 2.2.1 and 2.2.2.

23 Id. at chapter 6.

24 Id. at chapter 4.

25 Using country-level characteristics to estimate how revenues would be sourced under the destination-based principles for ADS and CFBs may not yield reliable estimates for individual companies because of idiosyncratic, firm-specific factors. However, the estimates are likely to be more reliable when dealing with industry aggregates across firms as in the present analysis.

26 This share is higher than what would have been allocated to the United States by the RSFs (for ADS and CFBs) based on country-level statistics described above. This is assumed to be a reasonable approximation for the present analysis because the split of U.S.-headquartered MNEs’ sales revenues (on a destination basis) can be expected to show a home country bias — that is, the United States accounts for a higher share of the MNEs’ worldwide revenues than what would be predicted purely by its structural factors relative to other markets (for example, based on population or average purchasing power). The skew of U.S. MNEs’ revenues toward the United States seen in the SOI CbC reporting data was also validated by reference to the Bureau of Economic Analysis’s “Activities of U.S. Multinational Enterprises, 2017” (2019).

27 Household (that is, private) consumption expenditure for a country is the difference between total consumption expenditure and government consumption expenditure (using 2017 data).

28 The pillar 1 blueprint discusses avoiding double taxation via an exemption or a credit mechanism. This analysis assumes an exemption approach.

29 For this approximation, the impact of the market connection priority test is disregarded. First, incorporating the impact of this feature of the four-part process is not feasible given the use of aggregated industry-level data in this analysis. Second, the pro rata allocation that follows the market connection priority step serves as a backstop to the market connection priority test. The pro rata allocation step will likely dilute the effect of the market connection priority test by allocating the burden of relief among a wider set of relieving jurisdictions when those deemed to have sufficient connection to a set of market jurisdictions do not have adequate residual profits to make up the full allocation of amount A to the relevant market jurisdictions.

30 The normal profits are estimated as the sum of two components. The first is a return on tangible assets in the jurisdiction. The second is a markup on employee costs estimated for each jurisdiction. The SOI CbC reporting data show only the number of employees by jurisdiction for each industry and do not show employee costs because companies are only required to report employee headcount in their CbC reports. To estimate the total employee compensation costs, the “average cost-per-employee” for each industry was derived using Bureau of Economic Analysis data from “Activities of U.S. Multinational Enterprises, 2017.” The data show total compensation as well as the number of employees for different U.S. multinationals by industry, for the parent jurisdiction (the United States), and for several other jurisdictions. The return on tangible assets was assumed to be 10 percent and the markup on employee costs was assumed to be 7 percent (a rate that is often used as an arm’s-length markup percentage for routine services).

31 For each industry, the total non-routine profits across all relieving jurisdictions was greater than the aggregate amount A estimated for that industry. Thus, the amount of relief (that is, income ceded) by a given jurisdiction is never more than that jurisdiction’s non-routine profits.

32 See OECD impact assessment, supra note 2, at Annex 2.C, Panel A. The OECD analysis expresses the change in the tax base by country groupings as a percentage of the groupings’ GDP.

33 Supra note 13.

34 Notably, $65.4 billion is the total amount A estimated for in-scope U.S. MNEs after rescaling the corresponding estimate derived for the Orbis sample ($56.6 billion). Thus, the total amount A tax base for in-scope U.S. companies estimated purely based on the Orbis sample represents 56.6 percent of the global total per the OECD impact assessment. Separately from the analysis described in this article, using Orbis data on U.S. and non-U.S. firms (without any rescaling) for 2016 to 2020, U.S. MNEs’ share of the worldwide allocable tax base under amount A was estimated to be 56 percent.

35 See, e.g., Michael J. Graetz, “A Major Simplification of the OECD’s Pillar 1 Proposal,” Tax Notes Int’l, Jan. 11, 2021, p. 199.

END FOOTNOTES

Copy RID