## wpiea2019213-print-pdf — Section III: Assessment of the direct effect of global Formula Apportionment (FA)

## Source details

**Canonical URL:** [wpiea2019213-print-pdf — Section III: Assessment of the direct effect of global Formula Apportionment (FA)](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019213-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019213-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019213-print-pdf.pdf.json)

---

### Major contribution and approach
- Paper assesses the direct effect on countries’ tax revenue of replacing the current system by global formula apportionment (FA).
- Three data sources exploited:
  - ORBIS (firm-level database) with improved methods to assess cross-border loss consolidation and apportionment using fixed assets and employment weights.
  - Bureau of Economic Analysis (BEA) data on operations of U.S. MNEs abroad for 2011-2016, expanding country coverage from 29 to 52 and allowing five more weighting schemes including payroll, sales by destination, value added, a Cobb-Douglas formula and the CCCTB formula.
  - Internal Revenue Service (IRS) country-by-country reports for U.S.-based MNEs, expanding coverage to 140 jurisdictions; includes only large MNEs (global revenue exceeding $850 mln.) for year 2016 and lacks some disaggregated formula inputs (e.g. sales by destination).

### What FA is and implementation choices
- FA allocates a consolidated (unitary) tax base of an MNE across jurisdictions using apportionment factors (assets, payroll, employment, sales or combinations).
- Key implementation design choices include:
  - Definition of a unitary group and a common consolidated tax base.
  - Choice of apportionment formula (single-factor or multi-factor, e.g., Cobb-Douglas, CCCTB).
  - Nexus and treaty adjustments, especially for destination-based sales factors.
  - Sector-specific adjustments (e.g., financial sector: loans/deposits or gross receipts; extractive industries: origin-based sales or extraction factors).

### Two direct channels affecting global CIT revenue
- Loss consolidation:
  - Cross-border consolidation of losses reduces the overall tax base because losses in one subsidiary immediately offset profits elsewhere in the group.
  - ORBIS-based analyses indicate global tax base under FA could be reduced by up to 10 percent due to cross-border loss consolidation.
  - Robustness using EBIT: for companies with losses in 2011, the global CIT base under cross-border consolidation is estimated to be 14 percent smaller than under SA with loss carry forward in 2016.
- Reallocation of tax base (relocation effects):
  - FA can shift taxable base from low-tax to high-tax countries, potentially offsetting reductions from loss consolidation.
  - ORBIS relocation effects: boosts global CIT revenue by 2.2 percent if assets are used; 4.6 percent if employment is used.
  - IRS relocation estimates: revenue gain is 3 percent if employment is used; asset factor yields negative revenue effect driven by reallocation into the U.S. (U.S. gains an additional 7 per cent of the global tax base) combined with a relatively low effective tax rate of 18 percent in the IRS data.
  - BEA relocation estimates: with U.S. effective tax rate of 27.7 percent, relocation yields positive effects ranging between 6 and 15 percent depending on formula.

### Key quantitative findings (static, no behavioral responses)
- ORBIS data:
  - Final sample comprises 58,345 unique companies in 7,772 MNE groups during 2011-2016.
  - Global tax base under FA could be reduced by up to 10 percent due to cross-border loss consolidation.
  - Stock of loss carryforward (aggregate USD): under SA grows from USD 400 bln to USD 1,100 bln between start and end period; under FA rises from USD 200 bln to USD 500 bln.
- BEA and IRS data:
  - Corporate tax revenue losses from MNEs of up to 80 percent are incurred by countries with low tax rates and a high ratio of FDI-to-GDP (“investment hubs”).
  - BEA data suggest a doubling of CIT revenue in the median of the group of non-OECD countries when employment weight is large.
  - BEA data suggest developing countries might also benefit from apportionment by destination sales.
  - G7 countries typically gain from FA, although gains are smaller when the apportionment formula is based on the number of employees.
- Descriptive statistics (selected exact figures from Table 3):
  - ORBIS Panel A (2011-2016, firm-level):
    - Profit/loss before taxes: Observations 241,986; Mean 10.49 (million USD); Standard Deviation 190; Min -12,590; Max 31,260.
    - Tax base, SA: Observations 241,889; Mean 14.45 (million USD); Standard Deviation 201; Min 0; Max 31,260.
    - Fixed Assets: Observations 245,480; Mean 151.3 (million USD); Standard Deviation 1,799; Min 0; Max 129,100.
    - No. of employees: Observations 245,480; Mean 349; Standard Deviation 3,409; Min 0; Max 372,562.
    - Effective Tax Rate (mean, country-level): Observations 336; Mean 0.21; Standard Deviation 0.25; Min 0.00; Max 1.75.
  - BEA Panel B (2011–2016, aggregate per country/year):
    - Profit before taxes: Observations 337; Mean 34,604 (million USD); Standard Deviation 145,812; Min -5,246; Max 1,124,857.
    - Taxable income (tax paid/tax rate): Observations 293; Mean 33,041 (million USD); Standard Deviation 124,338; Min -5,246; Max 848,900.
    - Payroll: Observations 985; Mean 19,012 (million USD); Standard Deviation 160,194; Min 0; Max 2,196,665.
    - Employment ('000 persons): Observations 1,047; Mean 296; Standard Deviation 2,175; Min 0; Max 28,046.
    - Fixed Assets: Observations 335; Mean 249,150 (million USD); Standard Deviation 1,253,607; Min 915; Max 10,277,053.
    - Tax: Observations 286; Mean 9,538 (million USD); Standard Deviation 41,532; Min 4; Max 331,712.
    - Effective Tax Rate: Observations 284; Mean 0.29; Standard Deviation 0.18; Min 0.00; Max 0.96.
  - IRS Panel C (year 2016, aggregate per jurisdiction):
    - Profit before taxes: Observations 137; Mean 10,357 (million USD); Standard Deviation 74,879; Min -6,204; Max 867,852.
    - Profit, non-loss-making companies only: Observations 87; Mean 20,214; Standard Deviation 106,826; Min 16; Max 990,019.
    - Employment ('000 persons): Observations 137; Mean 201; Standard Deviation 1,406; Min 0; Max 16,377.
    - Fixed Assets: Observations 137; Mean 45,217 (million USD); Standard Deviation 362,056; Min 4; Max 4,223,575.
    - Tax: Observations 137; Mean 1,914 (million USD); Standard Deviation 15,846; Min 0; Max 185,540.
    - Effective Tax Rate: Observations 137; Mean 0.20; Standard Deviation 0.12; Min 0; Max 0.68.
- Notes: All values except number of employees and tax rates are in million USD. Observations with negative tax or an effective tax rate greater than 1 have been dropped.

### Distributional impacts (summary of cross-country patterns)
- U.S. versus Rest of World (BEA and IRS):
  - U.S. tax revenue from MNEs expands under most formulas, except when the employment factor is used (U.S. tax revenue falls by 14 percent in the IRS data).
  - BEA: U.S. revenue increase ranges from 1 percent for employment factor to 20 percent for asset factor.
  - Median country in ROW: tax revenue falls under asset, payroll and CD formulas; expands under employment formula (effects range from decrease under asset factor of around 20 percent to increase under employment factor between 35 and 65 percent).
  - Under sales by destination formula, median country in ROW increases tax revenue.
- Advanced versus Developing economies:
  - ORBIS: median OECD country suffers loss in revenue from FA when asset or employment factor is used (partly reflecting 10 percent narrowing of global tax base).
  - Median non-OECD country shows revenue increase if employment factor is used, but not for asset factor.
  - Using BEA or IRS: OECD countries typically enjoy revenue expansion under most formulas, except pure asset factor.
  - Median OECD country experiences a 27 percent revenue increase from U.S.-based MNEs under value-added apportionment.
  - Median non-OECD country: revenue tends to decline under apportionment by assets or payroll; expands significantly (by nearly 85 percent) under apportionment by number of employees.
  - Apportionment by destination sales expands tax revenue in median non-OECD country.
- High versus Low-tax countries:
  - Countries with tax rates higher than 22.8 percent (sample average) tend to benefit from FA, except under asset formula.
  - Under asset formula, both median low-tax and high-tax countries lose revenue (explained by U.S. gain).
  - Most low-tax countries lose revenue from FA when production factors are used; gain when employment factor is used.

### Investment hubs: concentrated losses under FA
- Definition: Investment hubs characterized by high ratio of FDI to GDP.
- 2017 top ten inward FDI to GDP economies: Cyprus, Hong Kong SAR, Hungary, Ireland, Luxembourg, Malta, Mauritius, Netherlands, Singapore and Switzerland.
- Key findings:
  - Investment hubs lose significant tax revenue from FA, regardless of apportionment factor used.
  - Erosion of tax base can exceed 70 per cent of the MNE’s current tax payments under some formulas; corporate tax revenue losses from MNEs of up to 80 percent are observed for countries with low tax rates and high FDI-to-GDP.
  - ORBIS coverage limited; exception in ORBIS: median investment hub in Europe sees tax revenue increase under assets (reflects small ORBIS sample and positive effects in four of five covered countries).

### Modeling framework and formula specifications
- SA and FA tax base and allocation summarized with expressions for subsidiary profits, consolidated profits, apportionment shares and tax revenue; statutory tax rate ߬
௖௧ is assumed unchanged.
- Example multi-factor formulas preserved as presented:
  - Cobb-Douglas (CD) example: 1/3 ߙ_asset + 2/3 ߙ_payroll.
  - CCCTB example: 1/3 assets + 1/3 payroll + 1/3*(1/2 sales + 1/2 employment).
- Seven apportionment formulas explored using BEA data: employment, assets, payroll, sales by destination, value added, Cobb-Douglas, and CCCTB-style mixes.

### Dynamic effects, behavioral responses and policy implications
- Static analysis omits behavioral responses by firms and governments; dynamic effects can materially alter revenue and welfare conclusions.
- Under FA, CIT rates induce investment, employment and production location incentives depending on formula factors:
  - Asset-weighted formulas give firms incentive to locate assets in low-tax countries; dynamic relocation could reverse static gains for high-tax countries.
  - Sales by destination expected to produce smaller behavioral effects; risk of routing sales via low-tax sales hubs exists.
- Literature evidence:
  - Bettendorf et al. (2010): static CCCTB suggests low-tax countries lose base; including investment responses finds Ireland experiences net welfare gain.
  - Altshuler and Grubert (2009): firms shift from transfer pricing manipulation to organizational-structure manipulation; static revenue gains can disappear once behavioral responses are modeled.
- Policy implications:
  - Design choices matter: higher employment or destination-sales weights favor developing countries; asset weights favor capital-rich countries.
  - Tax competition may persist or intensify if formula factors are mobile; governments may respond by adjusting CIT rates or formula weights (U.S. states shifted toward sales factor historically).

### Conclusion (paper’s main messages)
- FA could reduce global profit-shifting and simplify international taxation, with particular potential benefits for some developing countries.
- New distortions and tax-planning opportunities would arise in allocation of formula factors; tax competition would not disappear and might intensify where factors are mobile.
- Distributional effects are large: investment hubs may face losses up to 80 percent of MNE tax payments; many large economies experience revenue increases.
- Developing countries gain primarily if employment receives large weight; sales-by-destination formulas also tend to favor consumption-rich developing countries.
- Limitations: lack of global MNE dataset covering low-income countries; static analysis ignores behavioral and general equilibrium effects critical for welfare assessment.

### Appendix highlights — revenue effects by economy (selected preserved figures)
- Appendix Table 1.1 (percentage change in CIT revenue from MNEs; summary statistics):
  - Median: -12.1; 23.1; 0.4; 56.9; 17.9; 21.2; -8.3; -23.6; 36.3; -9.6; 0.6
  - Average: -4.1; 46.6; 17.0; 114.0; 30.5; 40.7; 8.6; 94.6; 156.7; -11.0; 17.4
  - Aggregate: 11.0; 6.2; 14.5; 11.1; 10.3; 10.1; 13.3; -4.8; 3.2; -7.8; -5.4
- Appendix Table 1.2 (BEA, taxable income as tax base; selected country entries preserved):
  - Argentina: -44.2; -51.7; -60.8; -33.2; -34.4; -47.6; -55.3
  - Australia: 31.1; 2.3; 27.8; -0.8; 20.7; 15.7; 28.9
  - Austria: -1.1; 118.7; 95.7; 61.2; 39.1; 65.3; 63.4
  - Bermuda: 0.0; 0.0; 0.0; 0.0; 0.0; 0.0; 0.0
  - Brazil: 43.3; 101.1; 86.6; 241.6; 69.9; 102.8; 72.2
  - India: -53.9; -20.7; 23.0; 415.3; -5.4; 48.2; -2.6
  - Ireland: -60.4; -85.0; -84.4; -86.5; -24.6; -76.9; -76.4
  - United States: 40.5; 22.4; 37.4; 17.8; 29.1; 30.1; 38.4
  - Summary statistics (bottom of Appendix Table 1.2):
    - Median: -35.3; -4.5; -23.1; 3.3; -7.5; -5.7; -26.0
    - Mean: -23.0; 40.8; -3.0; 114.3; 9.9; 68.9; -9.6
    - Aggregate: 14.1; 6.9; 15.5; 8.4; 11.9; 11.1; 15.0
- Robustness using EBIT (Appendix 2 excerpts):
  - Loss consolidation (EBIT): global CIT base under cross-border consolidation is 14 percent smaller than under SA with loss carry forward in 2016 (compare: EBT-based reduction was 10 percent).
  - Relocation boosts global CIT revenue by 3 percent if assets used; by 5 percent if employment used.
  - Net global CIT tax revenue using EBIT: decrease by 11 percent under fixed asset formula; decrease by 9 percent under employment formula.

*Source: wpiea2019213-print-pdf — Section III and Appendix excerpts*

### References .............................................................................................................

### wpiea2019213-print-pdf - References .............................................................................................................

### I. Introduction and context
- Subnational corporate income taxes (CITs) commonly work by formula apportionment (FA).
- Under FA:
  - Accounts of all company’s affiliates are consolidated to generate a common tax base that is apportioned across jurisdictions on a formulaic basis.
  - Jurisdictions then apply their own tax rate to the apportioned base.
- This contrasts with separate accounting (SA) and arm’s-length pricing (ALP):
  - Under SA, the accounts of an affiliate terminate at the border and profit attribution to each affiliate is determined by ALP of intracompany transactions.
- Subnational FA exists in several countries: Canada, Germany, Japan, and the United States.

### II. Design features and national examples
- Canada:
  - Tax base of a corporation with a permanent establishment in more than one province is apportioned across those provinces based on an equal weighting of payroll and gross revenue.
  - Special weights or formulas apply to certain sectors (insurance, banking, transportation).
  - Provinces retain autonomy to apply their own tax rate and tax credits to the apportioned tax base.
- United States:
  - States can choose different weights for assets, payroll, and sales; weights may vary by sector.
  - Example: Alaska uses an origin-based sales factor for extractive industries.
- Variation in FA systems summarized in Table 1.

### III. Regional and global FA proposals
- European Union:
  - European Commission proposed a ‘Common Consolidated Corporate Tax Base’ (CCCTB) for the EU, implemented through a two-step approach:
    - First, a uniform common tax base is established across member states.
    - Second, EU-wide taxable profit is apportioned across member states by three equally-weighted factors: assets, sales by destination and labor (in turn equally weighting payroll and employees). 
  - See European Commission (2016) for a description and impact assessment of the CCCTB.
- Global proposals:
  - Worldwide formula apportionment has been proposed by the Independent Commission for the Reform of International Corporate Taxation (2018).
  - Scholarly advocacy examples include contributions in Picciotto (2016) and Avi-Yonah and Clausing (2008) (unilateral FA adoption by the U.S. discussed).
- Motivations:
  - Conceptual and practical difficulties of applying ALP in international taxation have increased interest in FA at regional or global levels.
  - Subnational experiences suggest FA may be better suited than ALP for dividing profits of related companies as economic and political integration proceeds.

### IV. Research agenda and paper scope
- Introduction of FA at the international level would be a significant departure from current practice.
- The paper aims to inform public debate by careful consideration and analysis of implications.
- A small but growing economic literature has assessed distributional and economic implications of FA; Section II discusses key findings.
- The paper benefits from comments and discussions with multiple experts and lists outstanding research assistance (footnote references 1–4).

### V. Document structure and supporting materials
- Tables listed:
  - Table 1. FA in Selected Federal States
  - Table 2. Number of Firms in MNE Groups in ORBIS, by Location and Owner Country Groups
  - Table 3. Descriptive Statistics for Key Variables
- Figures listed:
  - Figure 1. Allocation of the Tax Base Under Various Formulas
  - Figure 2. Stock of Loss Carryforward, by Year of Initial Losses
  - Figure 3. Revenue Effects of FA from Reallocating the Tax Base
  - Figure 4. Summary of Revenue Effects by Country Groups
  - Figure 5. Summary of Revenue Effects for Investment Hubs
- Appendixes listed:
  - Appendix 1. Revenue Effects of FA by Economy
  - Appendix 2. Revenue Effects of FA using EBIT to Measure Taxable Profit
- Page and section markers appearing in the source:
  - References section begins on page 36.
  - Introductory material spans pages referenced as 1, 4, and 5 in headings and footnotes.

*Source: wpiea2019213-print-pdf - References .............................................................................................................*

### Section III then presents the main contribution of this paper: an assessment of the direct effect

### Section III then presents the main contribution of this paper: an assessment of the direct effect

### Major contribution and approach
- The paper assesses the direct effect on countries’ tax revenue of replacing the current system by global formula apportionment (FA).
- Three data sources are exploited:
  - ORBIS (firm-level database), used with improved methods to assess cross-border loss consolidation and apportionment using fixed assets and employment weights.
  - Bureau of Economic Analysis (BEA) data on operations of U.S. MNEs abroad for 2011-2016, expanding country coverage from 29 to 52 and allowing five more weighting schemes including payroll, sales by destination, value added, a Cobb-Douglas formula and the CCCTB formula.
  - Internal Revenue Service (IRS) country-by-country reports for U.S.-based MNEs, expanding coverage to 140 jurisdictions; includes only large MNEs (global revenue exceeding $850 mln.) for year 2016 and lacks some disaggregated formula inputs (e.g. sales by destination).

### What is Formula Apportionment (FA)
- FA allocates a consolidated (unitary) tax base of an MNE across jurisdictions using apportionment factors (assets, payroll, employment, sales or combinations).
- Key implementation design choices include:
  - Definition of a unitary group and a common consolidated tax base (potentially via accounting rules).
  - Choice of apportionment formula (single-factor or multi-factor, e.g., Cobb-Douglas, CCCTB).
  - Nexus and treaty adjustments, especially for destination-based sales factors.
- Sector-specific adjustments may be required (financial sector: loans/deposits or gross receipts; extractive industries: origin-based sales or extraction factors).

### Revenue effects: two direct channels
- Moving from Separate Accounting (SA) to FA affects aggregate global CIT revenue via:
  - Loss consolidation: cross-border consolidation of losses reduces the overall tax base because losses in one subsidiary immediately offset profits elsewhere in the group (no separate subsidiary-level loss carryforward constraint under FA).
  - Reallocation of tax base: FA can shift taxable base from low-tax to high-tax countries, potentially offsetting reductions from loss consolidation.
- Evidence and prior findings summarized in the paper:
  - ORBIS-based analyses indicate global tax base under FA could be reduced by up to 10 percent due to cross-border loss consolidation.
  - Studies cited: Fuest, Hemmelgarn and Ramb (2007) report >20 percent EU-wide base reduction from loss consolidation; Cobham and Loretz (2014) report ~10 percent global; Bettendorf et al (2010) CGE model for Europe finds ~7.5 percent long-term reduction.
  - Reallocation effects can offset base reduction: Devereux and Loretz (2008) find EU-wide revenues rise by 2 percent; Cobham and Loretz (2014) report global revenue increases between 2 and 4 percent under FA.

### Key quantitative findings from this paper’s analysis (static, no behavioral responses)
- ORBIS data:
  - Global tax base under FA could be reduced by up to 10 percent due to cross-border loss consolidation.
  - Associated loss in global CIT revenue is much smaller and can even be positive because of reallocation from low- to high-tax countries.
- BEA and IRS data:
  - Effects on revenue distribution can be large.
  - Corporate tax revenue losses from MNEs of up to 80 percent are incurred by countries with low tax rates and a high ratio of FDI-to-GDP (so-called “investment hubs”).
  - Developing countries are more likely to benefit from FA if the number of employees has a large weight in the formula; BEA data suggest a doubling of CIT revenue in the median of the group of non-OECD countries when employment weight is large.
  - BEA data suggests developing countries might also benefit from apportionment by destination sales.
  - G7 countries typically gain from FA, although gains are smaller when the apportionment formula is based on the number of employees.

### Modeling framework (expressions summarized)
- Under Separate Accounting (SA):
  - Subsidiary reported profit: ߨ
௜௝௖௧
.
  - Taxable income for subsidiary under SA: ߚ
௜௝௖௧
ௌ
ߨൣݔܽܯൌ
௜௝௖௧
ߣ൅
௜௝௖௧ିଵ
ௌ
;0൧൒0
  - Country aggregate tax base: ߚ
௖௧
ൌ
ൌ෍෍ߚ
௜௝௖௧
.
  - Country tax revenue: ߩ
௖௧
ௌ
߬ൌ
௖௧
ߚ
௖௧
.
- Under Consolidation / FA:
  - Group consolidated profit: ߨ
௝௧
ൌ෍෍ߨ
௜௝௖௧
.
  - Consolidated tax base under FA: ߚ
௝௧
ி
ߨሾݔܽܯൌ
௝௧
ߣ	൅
௝௧ିଵ
ி
;0ሿ൒0
  - Apportionment share for subsidiary i (factor ߙ
௜௝௖௧
): ߱
௜௝௖௧
ൌ
ߙ
௜௝௖௧
∑∑
ߙ
௜௝௖௧௖௜
.
  - Allocated subsidiary tax base under FA: ߚ
௜௝௖௧
ி
߱ൌ
௜௝௖௧
ߚ
௝௧
ி
.
  - Tax paid uses country statutory tax rate ߬
௖௧
, assumed unchanged.

### Data sources and sample facts
- ORBIS:
  - Firm-level; final sample comprises 58,345 unique companies in 7,772 MNE groups during 2011-2016.
  - Sample restriction: include subsidiaries for which the sum of employment within each MNE group represents at least 70 percent of total employment reported in consolidated accounts.
  - ORBIS used to assess loss consolidation and provides fixed assets and employment; payroll and sales by destination are often missing.
- BEA:
  - Aggregate data for majority-owned foreign affiliates of U.S. parents across 199 countries; detailed info for affiliates in 52 countries (covers around 92 percent of worldwide profits of majority-owned affiliates).
  - Data years used: 2011-2016.
  - BEA provides fixed assets, payroll, employment and sales by destination (with about 10 percent of sales to unaffiliated persons having unspecified destination in BEA; benchmark region allocation and bilateral exports used to allocate to countries).
- IRS:
  - Aggregate country-by-country reports covering 142 tax jurisdictions; data for year 2016 includes large MNEs (global revenue > $850 mln).
  - Expands coverage to 140 jurisdictions for the paper’s analysis; limited in detailed formula factor inputs (e.g., lacks sales by destination for formula calculations).

### Descriptive statistics (selected exact figures from Table 3)
- Panel A: ORBIS dataset, 2011-2016 (firm-level)
  - Observations (Profit/loss before taxes): 241,986; Mean: 10.49 (million USD); Standard Deviation: 190; Min: -12,590; Max: 31,260.
  - Tax base, SA: Observations 241,889; Mean: 14.45 (million USD); Standard Deviation: 201; Min: 0; Max: 31,260.
  - Fixed Assets: Observations 245,480; Mean: 151.3 (million USD); Standard Deviation: 1,799; Min: 0; Max: 129,100.
  - No. of employees: Observations 245,480; Mean: 349; Standard Deviation: 3,409; Min: 0; Max: 372,562.
  - Effective Tax Rate (mean, country-level): Observations 336; Mean: 0.21; Standard Deviation: 0.25; Min: 0.00; Max: 1.75.
- Panel B: BEA dataset, 2011–2016 (aggregate per country/year)
  - Profit before taxes: Observations 337; Mean: 34,604 (million USD); Standard Deviation: 145,812; Min: -5,246; Max: 1,124,857.
  - Taxable income (tax paid/tax rate): Observations 293; Mean: 33,041 (million USD); Standard Deviation: 124,338; Min: -5,246; Max: 848,900.
  - Payroll: Observations 985; Mean: 19,012 (million USD); Standard Deviation: 160,194; Min: 0; Max: 2,196,665.
  - Employment ('000 persons): Observations 1,047; Mean: 296; Standard Deviation: 2,175; Min: 0; Max: 28,046.
  - Fixed Assets: Observations 335; Mean: 249,150 (million USD); Standard Deviation: 1,253,607; Min: 915; Max: 10,277,053.
  - Tax: Observations 286; Mean: 9,538 (million USD); Standard Deviation: 41,532; Min: 4; Max: 331,712.
  - Effective Tax Rate: Observations 284; Mean: 0.29; Standard Deviation: 0.18; Min: 0.00; Max: 0.96.
- Panel C: IRS dataset, year 2016 (aggregate per jurisdiction)
  - Profit before taxes: Observations 137; Mean: 10,357 (million USD); Standard Deviation: 74,879; Min: -6,204; Max: 867,852.
  - Profit, non-loss-making companies only: Observations 87; Mean: 20,214; Standard Deviation: 106,826; Min: 16; Max: 990,019.
  - Employment ('000 persons): Observations 137; Mean: 201; Standard Deviation: 1,406; Min: 0; Max: 16,377.
  - Fixed Assets: Observations 137; Mean: 45,217 (million USD); Standard Deviation: 362,056; Min: 4; Max: 4,223,575.
  - Tax: Observations 137; Mean: 1,914 (million USD); Standard Deviation: 15,846; Min: 0; Max: 185,540.
  - Effective Tax Rate: Observations 137; Mean: 0.20; Standard Deviation: 0.12; Min: 0; Max: 0.68.
- Notes: All values except number of employees and tax rates are in million USD. Observations with negative tax or an effective tax rate greater than 1 have been dropped.

### Formulas and multi-factor specifications used
- Cobb-Douglas (CD) formula (example used): ߱
஼஽,௝,௧
 =
 1/3 ߙ_asset + 2/3 ߙ_payroll (weights preserved as presented).
- CCCTB formula (example used): 1/3 assets + 1/3 payroll + 1/3*(1/2 sales + 1/2 employment) as specified in the paper.
- Seven apportionment formulas are explored using BEA data, including employment, assets, payroll, sales by destination, value added, Cobb-Douglas, and CCCTB-style mixes.

### Distributional and policy-relevant implications highlighted
- FA reallocates the tax base toward countries with larger shares in the chosen apportionment factors; for the BEA aggregates:
  - U.S. share of profits: 70 percent; investment hubs: 10 percent; rest of world (ROW): 20 percent.
  - U.S. share in formula weights ranges from 66 percent (employment) to 80 percent (assets).
  - For investment hubs, formula shares are typically 2–3 percent of global aggregate (much smaller than their share of reported profits), implying likely base losses under FA.
- Net global revenue effect depends on interplay between loss consolidation and reallocation:
  - Loss consolidation tends to reduce the global tax base.
  - Reallocation from low-tax to high-tax jurisdictions can raise aggregate revenue and may offset base reduction.
- Sectoral and design choices matter for distributional outcomes (e.g., destination-based sales favors consumer countries; employment-weighted formulas favor labor-rich developing countries).

### Analytical scope and limitations (as stated)
- The analysis is static and ignores behavioral responses (no firm or government re-optimization).
- Data limitations:
  - ORBIS: stronger for European MNEs, missing payroll and sales-by-destination in many cases.
  - BEA: detailed and includes sales by destination for 52 countries, but destination unspecified for ~10 percent of sales to unaffiliated persons (allocated by region and bilateral exports).
  - IRS: broader country coverage (142 jurisdictions) but restricted to large MNEs in 2016 and limited formula inputs.
- The paper assumes that the statutory tax rate ߬
௖௧
 remains the same under FA when computing tax paid.

*Source: wpiea2019213-print-pdf - Section III then presents the main contribution of this paper: an assessment of the direct effect*

### 30.8 percent, on average; and for Investment hubs, it is only 14.7 percent, on average.

### wpiea2019213-print-pdf - 30.8 percent, on average; and for Investment hubs, it is only 14.7 percent, on average.

### Allocation factors and automatic stabilization
- Allocation factors will depend on country-specific macro-economic conditions, including cyclical conditions, and thus affect automatic stabilization of the CIT.
- The consolidated profit might be relatively more robust and less dependent on asymmetric shocks.

### Data sources and scope (BEA, IRS, ORBIS)
- Panel C of Table 3: IRS country-by-country reports for the 2016 tax year.
  - “Profit” reflects profit according to prevailing rules of the country (financial statements or tax records).
  - Tax variable includes income taxes and withholding taxes paid on payments received by the subsidiary.
  - Profitable U.S. MNEs in 2016 reported total profits of USD 1,759 billion across all jurisdictions.
  - Over half of these profits were reported in the U.S.: USD 990 billion.
  - Loss-making U.S. MNEs reported USD 338 billion in losses, of which USD 122 billion was reported in the U.S.
  - BEA-covered employment: 42.3 million workers employed by U.S. MNEs.
  - IRS-covered employment: 27.6 million workers covered by the IRS data.
- Note: Due to aggregate nature of data, estimating revenue effects of cross-border loss consolidation within company groups is not possible (data also offset losses between company groups).

### Effects on global tax revenue: loss carryforward and consolidation
- Method: ORBIS data used to assess impact of cross-border loss consolidation on global corporate tax base; start with loss carry forward in 2011 based on reported losses in 2010; analyze under SA and consolidation with FA.
- Dynamics of loss carryforward stock (2011–2016):
  - Initial growth as new losses exceed offsets; growth levels off as offsets grow.
  - Between 2015 to 2016: stock declines under SA and is virtually stable under FA.
  - Assessment best based on 2015 and 2016 when convergence achieved.
- Magnitude:
  - Under SA, stock of loss carryforward grows from an aggregate USD 400 bln to 1100 bln between start and end period.
  - Under FA, it rises from a total of USD 200 bln to 500 bln.
- Aggregate impact on global CIT base:
  - For all companies that incurred some losses in 2011, global CIT base under cross-border consolidation is estimated to be 10 percent smaller than under SA with loss carryforward in 2016.
  - This 10 percent represents an upper bound since some countries already allow domestic relief/loss consolidation not captured here.
- Assumption: analysis ignores limitations in period of loss carryforward; period explored is 5 years and almost all countries have periods exceeding this, so assumption of unlimited carryforward does not affect results.

### Revenue effects from reallocating the tax base (relocation effects)
- ORBIS estimates:
  - Relocation effect boosts global CIT revenue by 2.2 percent if assets are used in the formula.
  - Relocation effect boosts global CIT revenue by 4.6 percent if employment is used.
- IRS data estimates:
  - Revenue gain is 3 percent if employment is used as a factor.
  - For the asset factor, there is a negative revenue effect driven by reallocation of the tax base into the U.S. (which gains an additional 7 per cent of the global tax base) combined with a relatively low effective tax rate of 18 percent in the IRS data.
- BEA data estimates:
  - BEA reports an effective tax rate in the U.S. of 27.7 percent.
  - Relocation of the tax base yields much larger positive effects on global revenue, ranging between 6 and 15 percent depending on the formula used.

### Effects on the distribution of tax revenue (summary findings)
- Figure 4 summarizes effects of FA on distribution of CIT revenue paid by MNEs across country groups: (a) US versus ROW; (b) OECD versus non-OECD; (c) below and above the median tax rate.
- United States versus Rest of World (Panel A):
  - Results for the U.S. from BEA and IRS (not ORBIS).
  - U.S. tax revenue from MNEs expands under most formulas, except when the employment factor is used (U.S. tax revenue falls by 14 percent in the IRS data).
  - BEA data: U.S. revenue increase ranges from 1 percent for the employment factor to 20 percent for the asset factor.
  - Median country in ROW mirrors U.S. effects: tax revenue falls under asset, payroll and CD formulas; expands under employment formula.
  - Effects range from a decrease under the asset factor of around 20 percent in both BEA and IRS data, to an increase under employment factor of between 35 and 65 percent.
  - Under sales by destination formula, median country in ROW increases tax revenue.
- Advanced versus Developing Economies (Panel B):
  - ORBIS: median OECD country suffers a loss in revenue from FA when asset or employment factor is used (partly reflecting 10 percent narrowing of global tax base).
  - Median non-OECD country shows revenue increase if employment factor is used, but not for asset factor.
  - Using U.S. MNE data (BEA or IRS): OECD countries typically enjoy revenue expansion under most formulas, except pure asset factor.
  - Median OECD country experiences a 27 percent revenue increase from U.S.-based MNEs under value-added apportionment.
  - Median non-OECD country: revenue tends to decline under apportionment by assets or payroll; expands significantly (by nearly 85 percent) under apportionment by number of employees.
  - Apportionment by destination sales expands tax revenue in median non-OECD country (possibly due to high consumption relative to production).
- High versus Low-Tax Countries (Panel C):
  - Countries with relatively high tax rates (higher than 22.8 percent, average in sample) tend to benefit from FA, except if the asset formula is used.
  - Under asset formula, both median low-tax and high-tax countries lose revenue (explained by gain in the U.S.).
  - Most low-tax countries lose revenue from FA when production factors are in the formula (asset, payroll, Cobb-Douglas), but gain when employment is used.
- Investment hubs (Figure 5; summary here and detailed below).

### Investment hubs: concentrated losses under FA
- Definition: Investment hubs characterized by a high ratio of FDI to GDP (Damgaard and Elkjaer 2017).
- 2017 top ten inward FDI to GDP economies: Cyprus, Hong Kong SAR, Hungary, Ireland, Luxembourg, Malta, Mauritius, Netherlands, Singapore and Switzerland.
- ORBIS coverage limited to five: Hungary, Ireland, Luxembourg, Malta and the Netherlands.
- Key findings:
  - Investment hubs lose significant tax revenue from the introduction of FA, regardless of apportionment factor used.
  - The erosion of the tax base can exceed 70 per cent of the MNE’s current tax payments under some formulas.
  - Results illustrate large differences between currently reported profits by U.S. MNEs in these countries and reported factor shares used in FA apportionment.
  - Exception: when ORBIS data are used and apportionment is by assets, the median investment hub in Europe sees tax revenue increase — reflecting small ORBIS sample and positive effects in four of five countries where MNEs report large asset shares (exception Ireland).
  - This ORBIS-based positive asset result does not carry over to BEA or IRS data for U.S. MNEs.

### Dynamic effects and behavioral responses
- Static analysis omission: results above are static and do not account for changes in behavior of MNEs or corporate tax policy changes by countries.
- Under FA, CIT rates induce incentives for MNE investment, employment and production depending on formula factors:
  - If assets are part of formula, MNEs will have incentive to locate more assets in low-tax countries and less in high-tax countries to shift taxable share to low-rate jurisdictions.
  - Dynamic relocation of assets would benefit low-CIT countries and harm high-CIT countries, opposite to static effects, mitigating revenue effects.
  - Relocation of capital may boost labor productivity in low-tax countries and reduce it in high-tax countries, affecting GDP and welfare.
- Evidence from literature:
  - Bettendorf et al. (2010): CCCTB static analysis suggests low-tax countries like Ireland lose tax base; accounting for dynamic investment effects finds Ireland experiences net welfare gain.
  - Altshuler and Grubert (2009): Under FA, firms shift from transfer pricing manipulation to manipulating organizational structure (outsourcing/merging activities). Simulations show static revenue gains for the U.S. under an equal-weight formula of assets, payroll and sales virtually disappear once behavioral responses are modeled.
- Policy and design implications:
  - Sales by destination expected to produce much smaller behavioral effects since consumers are less responsive to CIT rate differences than production factors.
  - Risk of sales factor: low-tax countries could become sales hubs by having MNEs route sales through third parties located in these hubs to minimize global tax liability.
  - Governments can respond: if assets receive significant weight and MNEs relocate assets, governments have incentive to reduce CIT rates to attract investment.
  - Bettendorf et al. (2010) simulations suggest welfare gain of a CIT rate cut by an individual country is larger under FA than under current system.
  - Experience from the U.S.: tax competition led States to modify formulas by moving away from production factors towards larger share of sales by destination.
- Conclusion on dynamics:
  - Dynamic effects can significantly change assessment of revenue and wider economic impact of FA.
  - Such effects depend critically on magnitude of behavioral responses.
  - Dynamic effects likely more important if production factors receive large weight; less important if sales by destination receives large weight.

### Conclusion (summary of paper’s main messages)
- Paper explores economic and distributional impact of global formula apportionment (FA).
- Key points:
  - Existing literature emphasizes benefits from elimination of profit shifting within MNE groups and simplification gains; particularly important for developing countries.
  - New distortions will arise in allocation of formula factors; opportunities for tax planning will persist in different forms.
  - Tax competition will not disappear under FA and might intensify if formula factors are mobile across borders.
  - Distributional effects are large; largest revenue decreases apparent in investment hubs (losses of up to 80 percent of MNE tax payments reported).
  - Many large economies experience increase in CIT revenue.
  - Developing countries gain mostly if employment receives large weight in formula; also tend to benefit on average from formula based on sales by destination.
- Broader relevance:
  - Analysis informs debates on allocation of taxing rights and profit attribution by showing how profit would be allocated if it resembled allocation of production factors or sales.
  - Sheds light on revenue implications of incremental reforms that partly use formulary elements (e.g., residual profit allocation proposals).
- Limitations:
  - Lack of a global dataset for MNEs covering low-income countries; potential future resolution via country-by-country report database.
  - Unclear what proportion of total corporate tax payments covered MNEs in the data.
  - Analysis is static and ignores behavioral responses by firms and governments and general equilibrium effects (e.g., exchange rate effects when tax base shifts from source to destination).
  - These economic effects are critical for understanding welfare implications and desirability of FA.

*Source: wpiea2019213-print-pdf*

### Appendix 1. Revenue Effects of FA by Economy

### Appendix 1. Revenue Effects of FA by Economy

### Overview
- Appendix presents the change in CIT revenue collected from MNEs under global adoption of formula apportionment (FA), reported as percentage change.
- Results are shown for various apportionment factors (Assets, Sales, Payroll, Employment, VA, CCCTB, CD) and using three datasets: BEA, IRS, ORBIS.
- Appendix Table 1.1 uses the standard profit variable; Appendix Table 1.2 uses taxable income as the proxy for the tax base.
- The “Employment” column indicates change in total CIT revenue from MNEs if the share of employees in each economy is used to allocate consolidated profit.

### Appendix Table 1.1 — Key findings and summary statistics (Percentage Change in CIT Revenue from MNEs)
- Table covers individual-economy percentage changes under FA across apportionment factors and datasets (BEA, IRS, ORBIS).
- Examples noted in the table text:
  - Argentina (multiple entries across factors and datasets): includes values such as 2.8; -11.6; -27.5; 21.6; 19.4; -3.9; -17.4; -28.4; 68.6.
  - Australia: 84.1; 43.1; 80.6; 40.8; 66.5; 62.6; 81.8; 148.7; 28.2; -4.1; -1.1.
  - Austria: 50.0; 232.9; 193.7; 139.2; 107.7; 149.8; 145.8; -17.2; 36.1; -28.0; -75.5.
  - Many small-economy and tax-haven entries show large negative changes (e.g., Bermuda: -79.5; -91.7; -94.4; -96.6; -75.6; -88.9; -89.5; -92.6; -99.9).
- Aggregate summary statistics reported at bottom of Appendix Table 1.1:
  - Median: -12.1; 23.1; 0.4; 56.9; 17.9; 21.2; -8.3; -23.6; 36.3; -9.6; 0.6
  - Average: -4.1; 46.6; 17.0; 114.0; 30.5; 40.7; 8.6; 94.6; 156.7; -11.0; 17.4
  - Aggregate: 11.0; 6.2; 14.5; 11.1; 10.3; 10.1; 13.3; -4.8; 3.2; -7.8; -5.4

### Appendix Table 1.2 — Revenue Effects using taxable income as tax base (Percentage Change in CIT Revenue; Dataset: BEA)
- Appendix Table 1.2 reports percentage changes in CIT revenue when taxable income is used as the proxy for the tax base, across apportionment factors (Assets, Sales, Payroll, Employment, VA, CCCTB, CD) for BEA.
- Selected country entries (exact values preserved from table):
  - Argentina: -44.2; -51.7; -60.8; -33.2; -34.4; -47.6; -55.3
  - Australia: 31.1; 2.3; 27.8; -0.8; 20.7; 15.7; 28.9
  - Austria: -1.1; 118.7; 95.7; 61.2; 39.1; 65.3; 63.4
  - Barbados: 3.3; -65.7; -88.5; -76.0; 137.8; -57.6
  - Belgium: -16.2; 28.1; 31.0; -4.6; 44.1; 8.4; 15.3
  - Bermuda: 0.0; 0.0; 0.0; 0.0; 0.0; 0.0; 0.0
  - Brazil: 43.3; 101.1; 86.6; 241.6; 69.9; 102.8; 72.2
  - Canada: 46.0; 34.7; 16.4; 40.6; 28.1; 36.4; 26.3
  - China: -45.5; 17.6; -41.4; 163.8; -25.1; 11.1; -42.8
  - Honduras: 199.9; 1530.7; 441.6; 3733.4; 382.6; 3387.0; 361.1
  - India: -53.9; -20.7; 23.0; 415.3; -5.4; 48.2; -2.6
  - Ireland: -60.4; -85.0; -84.4; -86.5; -24.6; -76.9; -76.4
  - United States: 40.5; 22.4; 37.4; 17.8; 29.1; 30.1; 38.4
  - Additional entries for many economies are listed in the table (values preserved in the source).
- Summary statistics at bottom of Appendix Table 1.2:
  - Median: -35.3; -4.5; -23.1; 3.3; -7.5; -5.7; -26.0
  - Mean: -23.0; 40.8; -3.0; 114.3; 9.9; 68.9; -9.6
  - Aggregate: 14.1; 6.9; 15.5; 8.4; 11.9; 11.1; 15.0
- Note: Aggregate for BEA is calculated between 2014 - 2016 due to a break in the data

### Robustness and related results using EBIT (Appendix 2 excerpts)
- Rationale: EBIT excludes income from equity investment in affiliates and avoids double-counting of subsidiary income; interest income/payment remain in EBIT.
- Impact of loss consolidation (EBIT measure):
  - For companies with losses in 2011, the global CIT base under cross-border consolidation is estimated to be 14 percent smaller than under SA with loss carry forward in 2016. (Compare: EBT-based reduction was 10 percent.)
- Incorporating reallocation (EBIT measure):
  - Relocation effect boosts global CIT revenue by 3 percent if assets are used in the formula.
  - Relocation effect boosts global CIT revenue by 5 percent if employment is used in the formula.
  - Net global CIT tax revenue using EBIT:
    - Decrease by 11 percent under the fixed asset formula.
    - Decrease by 9 percent under the employment formula.
  - The larger size of reduction relative to prior assessment is mainly due to a smaller global CIT base after loss consolidation using EBIT.

*Appendix 1. Revenue Effects of FA by Economy (from wpiea2019213-print-pdf)*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019213-print-pdf.pdf_