## 2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits (wpiea2020049-print-pdf)

## Source details

**Canonical URL:** [2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits (wpiea2020049-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020049-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020049-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020049-print-pdf.pdf.json)

---

### Introduction and motivation
- RPA schemes break multinational profits into routine profit (normal return) and residual profit (aggregate excess profit), allocating routine profit mechanically and residual profit by formula (e.g., destination-based sales).
- RPA departs from current norms by (i) eliminating scope for profit shifting via group-level allocation of residual profit and (ii) allocating taxing rights to jurisdictions (market/destination countries) that may currently have no corporate tax liability.
- Open empirical/policy questions motivating the analysis:
  - How large are routine and residual profits in practice?
  - How are they distributed across multinationals and countries?
  - Which countries would gain or lose revenue under RPA?
  - Likely impacts on investment and efficiency.
- Paper objective: establish core stylized facts and analytical features needed to assess RPA schemes, acknowledging substantial data and definitional challenges.

### Model primitives and implications for marginal effective tax rates (METRs)
- Key definitions and equations:
  - Multinational invests k_i in country i; aggregate capital K ≡ ∑_i k_i. Routine profit in i corresponds to k_i r.
  - Total earnings before financing costs: Ẑ(k). Residual profit: Π(k) ≡ Ẑ(k) − ∑_i k_i r. (Equation (1))
  - Under generic RPA, tax payable in i: T_i = t_i θ_i Π + τ_i k_i r. (Equation (2))
  - After-tax profits: (1 − t̄) Π(k) − ∑_i τ_i k_i r, where t̄(k,t) ≡ ∑_i t_i θ_i. (Equation (3))
  - METR under RPA: METR_i^RPA ≡ (∂Ẑ/∂k_i − r)/r = τ_i (1 − t̄) + t̄ ∂(θ·k)/∂k_i · Π/(r (1 − t̄)). (Equation (5))
    - First term: tax on routine profits amplified by residual taxation.
    - Second term: distortion from k-dependence of allocation weights θ_i; increases with size of residual profits.
- Under arm’s length pricing (ALP):
  - Tax liability: T_i = t_i π_i + (1 − t_i) h_i k_i r. (Equation (6))
  - Declared excess profits: π_i = π_i^*(k_i) + s_i (profit shifting s_i; ∑_i s_i = 0). (Equations (7), (8a))
  - Profit shifting cost specified: C(k_i, s_i) = 1/2 ϑ (s_i^2 / (k_i r)). (Equation (11))
  - METR under ALP with shifting: METR_i^ALP = h_i − (1/(ϑ 2)) (1 − t_i) (s_i / (k_i r))^2. (Equation (12))
  - Profit shifting s_i proportional to deviations of local tax rate from capital-weighted average tax elsewhere. (Equations (13)-(14))

### Data sources, construction, and key sample statistics
- Two datasets:
  - Firm-level consolidated MNE group data (S&P Capital IQ), top 10,000 largest MNEs for 2011–2017; baseline sample after cleaning: 8,854 MNEs.
    - Aggregate 2017 sample statistics:
      - Total assets: USD 658 trillion.
      - Total net fixed assets: USD 32 trillion.
      - Reported taxable profits: USD 5.3 trillion.
      - Total revenue of the smallest MNE in sample: USD 1.196 billion.
      - Average tax rate (total current tax / total pre-tax profit): around 21 percent.
    - Headquarters distribution (Table 1 summary):
      - Advanced Economies: Number of Companies 5,961; Share of Fixed Assets 57.5%; Share of EBIT 70.7%; Average Tax Rate 20.1%.
      - Emerging Markets: Number of Companies 2,044; Share of Fixed Assets 27.8%; Share of EBIT 19.8%; Average Tax Rate 22.2%.
      - Low-Income Developing Countries: Number of Companies 22; Share of Fixed Assets 0.1%; Share of EBIT 0.1%; Average Tax Rate 21.4%.
      - HQ Undisclosed: Number of Companies 827; Share of Fixed Assets 14.6%; Share of EBIT 9.4%; Average Tax Rate 19.7%.
    - Sectoral examples (Table 2 summary):
      - Manufacturing: Number of Companies 2,940; Share of Fixed Assets 29.5%; Share of EBIT 40.0%; Average Tax Rate 24%; Return on fixed assets 26%; Mark-up 13%; K/COGS 51%.
      - Transportation, Communication, Utilities: Number of Companies 1,345; Share of Fixed Assets 39.8%; Share of EBIT 22.1%; Average Tax Rate 19%; Return on fixed assets 11%; Mark-up 19%; K/COGS 179%.
      - Averages: Average, unweighted: Average Tax Rate 24%; Mark-up 12%; K/COGS 49%. Average, weighted (with fixed assets): Average Tax Rate 19%; Mark-up 13%; K/COGS 65%.
    - Note: consolidated accounts capture residual profits but lack jurisdictional detail to attribute routine profits by country or allocate residual by destination sales.
  - Country-level macro data for 129 countries (final sample), averaging variables between 2015 and 2017:
    - Capital stocks by jurisdiction from UN national accounts (51 countries), Bui and others (2019) (77 countries), and WEO (32 countries), adjusted to approximate corporate non-financial capital stocks.
    - Destination-based third-party sales proxied as GDP minus net exports from the IMF’s World Economic Outlook.
    - Country-level declared excess profits π_i estimated from T_i^CIT / t_i = π_i + (1 − t_i) h_i k_i r, using assumption (1 − t_i) h_i / t_i ≈ 1 and r = 7.5 percent.
    - Statutory tax rates from IMF’s WoRLD database.
- Data caveats: routine/residual concepts lack direct accounting counterparts; analysis relies on plausible but untested assumptions and eclectic data linkage.

### Methods to proxy routine profit
- Two proxy methods (residual = EBIT − routine):
  1. Tangibles-based approach: apply notional rate of return to tangible assets. Benchmark notional rates: 7.5 percent and 10 percent (also 5 percent used in sensitivity).
  2. COGS-based approach: apply markup to total direct COGS (group-level micro data only). Benchmark mark-ups: 3.75 percent and 5 percent (also 2 percent and 2.5 percent in sensitivity).
- Relationship: COGS nominally around twice tangible assets → notional return r on tangibles corresponds roughly to one-half the markup on COGS.
- Interest payments treated as deducted in calculating residual profit (alternatives discussed but not implemented due to data limits).

### Routine profit findings
- Illustration at 10 percent return on tangibles or 5 percent markup on COGS:
  - Routine profits ≈ 40-50 percent of total profits in absence of offset.
  - Routine profits ≈ 30 percent of total profit among firms earning positive profit (when negative residuals are offset).
- Country-level macro result:
  - At routine return r = 7.5 percent, routine profit exceeds current CIT base in 23 countries.
  - Robust negative relationship between routine profit / current CIT base and GDP per capita.
  - Top 10 countries by capital stock relative to GDP per capita are disproportionately emerging markets.

### Residual profit findings (MNE-level)
- Definition: Π(k) = EBIT − routine profit.
- 2017, among MNEs with positive group-wide pre-tax profit (≈ 90 percent of groups):
  - Proportion with implied negative residual:
    - Tangibles-based, 10% return: 19 percent.
    - COGS-based, 5% markup: 28 percent.
  - Negative-residual firms (28 percent under mark-up) own around 12 percent of fixed assets and earn around 1 percent of taxable profits.
  - Sensitivity:
    - Tangibles-based, 5% return: negative residual falls to 8 percent.
    - COGS-based, 2% markup: negative residual falls to 11 percent.
- Duration of negative residual spells (examples from Table 5):
  - At 7.5% tangible return: 0 years = 71.1%, 7 years = 2.0%.
  - At 3.75% COGS markup: 0 years = 62.3%, 7 years = 5.5%.
- Present value negativity over 2011–2017 (Table 6):
  - Tangible-based (interest/notional return combos): e.g., 5% interest with 5.0% return → 2.7%; 7.5% interest with 5.0% return → 8.4%; 10% interest with 5.0% return → 14.8%.
  - COGS-based (interest = one-half tangible interest): e.g., 5% interest with 2.5% markup → 7.7%; 7.5% interest with 3.75% markup → 13.9%; 10% interest with 5% markup → 20.6%.
- Residual share among firms with positive residuals:
  - Tangible-based, 10% return: residual ≈ 70 percent of total profits for firms with positive residual.
  - Weighted by pre-tax profit: residual/total ≈ 71 percent (10% return) and ≈ 75 percent (7.5% return).
  - COGS-based: overall ratio ≈ 68 percent (5% markup) and 66 percent (3.75% markup).
- Concentration:
  - Largest 10 percent of corporate groups (by sales) account for >50 percent of all positive residual profits (tangibles-based, 7.5%).
  - Top 100 companies (~1% of sample) account for >28 percent of total residual profits; these have residual/total profit ratio of 85 percent and account for ~17 percent of tangible assets.
- Persistence (Table 7), at 7.5% tangible return:
  - Probability of staying in top quintile next year: 72.12 percent.
  - Probability of staying in lowest quintile next year: 88.18 percent.

### Geography, headquarters concentration (Table 8)
- Top HQ countries capture large shares of implied residual profit across specifications:
  - United States: 33% (7.5% tangibles), 36% (10% tangibles), 33% (3.75% COGS), 35% (5% COGS).
  - China: 10% (7.5%), 7% (10%), 12% (3.75%), 11% (5%).
  - Japan: 8% (7.5%), 8% (10%), 7% (3.75%), 6% (5%).
- Around 75-85 percent of aggregate residual profit accrues to companies headquartered in just 10 countries; more than half accrues to MNEs headquartered in the same four; United States alone ≈ one-third.

### Aggregate revenue effects of moving to destination-based RPA (no behavioral responses)
- Four effects when moving from ALP to RPA (equation (16)): (i) reallocation of residual relative to declared excess, (ii) change related to routine profit treatment, (iii) revenue change from investment responses, (iv) gain from efficiency increases. Sections quantify first two effects assuming universal adoption and no behavioral responses.
- Reallocation of excess profits under destination-based RPA (allocation weight = share of sales to third-party residents):
  - Aggregate estimated increase in total tax revenue: USD 168 billion ≈ 7.1 percent of total CIT revenue in the sample.
  - Aggregate CIT revenue in sample: USD 2370 billion.
  - Unweighted average change in tax revenue: increase of 15 percent.
- Country-level variation:
  - Largest proportionate increase: Central African Republic, 290 percent.
  - Largest reduction: Hong Kong SAR, -85 percent.
  - Lower-income countries tend to gain proportionally more than higher-income countries.
  - Around 30 percent of countries would lose at least 25 percent of current CIT revenue; around 30 percent would gain at least 25 percent.

### Regression drivers of reallocation effects (Table 9 summary)
- Dependent variable: simulated change in tax base (percent of current tax base).
- Statistically significant drivers include:
  - CIT rate: positive and significant. Examples of coefficients (destination-based):
    - Column 1: 2.083 (increase in tax base percent per percentage point higher CIT rate).
    - Column 2: 2.388***.
    - Capital stocks specifications: 1.680**, 1.899***.
  - log(GDP per cap.): negative and significant. Examples:
    - Destination-based: -15.436*** (Column 1), -18.484*** (Column 2).
    - Capital-based: -10.548*** (Column 3), -15.550** (Column 4).
  - FDI/GDP>1.5 (investment hubs): large negative and significant (e.g., -56.649***, -59.706***, -60.280***, -63.171***).
  - Resource rich: negative but not consistently significant.
  - Trade deficit: not significant.
- Intercepts reported and statistically significant (e.g., 20.281***, 22.001***, 15.266***, 17.445***).

### Decomposition of reallocation effects (Table 10) — elimination of profit shifting vs reallocation of purged profits
- Destination-based sales allocation, cost-function based identification:
  - Country-specific average (percent of revenue): Profit shifting -0.43; Reallocation of purged profits 5.44; Total 5.02.
  - Global revenue change (percent of revenue): Profit shifting 2.61; Reallocation 4.48; Total 7.09.
  - Global revenue change (billion USD): 62 (profit shifting), 106 (reallocation), 168 total.
- Destination-based sales allocation, activity-based identification:
  - Country-specific average: Profit shifting 3.9; Reallocation 1.12; Total 5.02.
  - Global revenue change (percent): Profit shifting 4.61; Reallocation 2.48; Total 7.09.
  - Global revenue change (billion USD): 109 (profit shifting), 59 (reallocation), 168 total.
- Capital-based allocation, cost-function based identification:
  - Country-specific average: Profit shifting -0.43; Reallocation 4.33; Total 3.9.
  - Global revenue change (percent): Profit shifting 2.61; Reallocation 2; Total 4.61.
  - Global revenue change (billion USD): 62 (profit shifting), 48 (reallocation), 109 total.
- Interpretation:
  - Total global revenue effect from reallocation of excess profits is robust at USD 168 billion (7.09 percent of global CIT revenue) under destination-based allocation, but composition between elimination of profit shifting and reallocation of purged profits depends on identification method.

### Identification strategies to estimate profit shifting
- Cost-function based approach:
  - Estimate parameter ߠ of cost function calibrating to consensus semi-elasticity of reported gross profits to home-foreign tax rate differential → choose ߠ = 2.55.
  - Implied ߚ௜ from ߚ௜ ߨ = (ݏെݜ)/Π relation.
- Activity-based approach:
  - Assume purged excess returns proportional to country-specific capital stocks: ߚ௜ = ݇௜/∑݇.
  - Profit-shifting then estimated as residual ݏ௜ = ߨ௜ − ߚ௜ Π.
- Comparison:
  - Cost-function centers on statutory CIT differentials as driver; activity-based captures shifts tied to distribution of real activity and special tax regimes.
  - Empirical decomposition (Figure 8) shows dominance of elimination of shifting vs reallocation of purged profits depends strongly on identification approach.

### Drivers of decomposed effects (summary)
- Elimination of profit shifting:
  - Under cost-function identification, CIT rates are main driver of revenue changes from elimination of profit shifting; GDP per capita not significant.
- Reallocation of purged profits:
  - Generates revenue losses in investment hubs and benefits less advanced economies.
- Capital-based allocation reduces reallocation-of-purged-profits effect compared with destination-based allocation.
- Overall: higher statutory CIT rates → larger gains under RPA reallocation; lower-income countries tend to gain proportionally more; investment hubs (high FDI/GDP) lose large shares of tax base.

### Routine-profit treatment and revenue implications
- Two benchmark treatments of routine profits:
  1. Routine profits taxed at current statutory rates (ݐ௜ = current statutory rate):
     - Revenue effects (percent of current revenue) can be large and sometimes negative where closed-economy METR exceeds grossed-up statutory rate.
     - Examples:
       - Estonia: METR 31 percent; statutory rate 20 percent → loss of 31 percent of current tax revenue if routine returns taxed at statutory rate.
       - Argentina: statutory rate 35 percent; closed economy METR 27 percent → revenue would increase by 101 percent.
     - Conclusion: taxing routine profits at statutory rates would substantially increase revenue in virtually all countries.
  2. Routine profits taxed so as to leave marginal tax on investment unchanged (ݐ௜ set so second term in (16) vanishes):
     - No impact on revenue through treatment of routine profit.
- Allowing financing costs against routine profits would reduce revenue impacts and shift effects toward residual component.

### METR, investment responses, and distributional effects
- Current METRs under ALP:
  - METRs average around 15 percent with wide variation; profit-shifting term can dominate in countries with very high or very low statutory rates.
  - Example: Bulgaria’s closed economy METR is sufficiently low that profit shifting makes overall METR negative.
- METR under capital-based RPA:
  - Requires assumptions on routine profit tax rate, allocation weighting (capital weighting as in (19)), and ratio of residual to routine profit (routine return assumed 7.5 percent).
  - Under capital-based weighting, METR under RPA (equation (20)) includes a “weighting” term that raises (lowers) METR for investment in countries with relatively high (low) statutory tax rates on residual profits.
- Empirical METR results (Figures 12–13):
  - METR is higher when routine profits taxed at statutory rates than when taxed to leave effective tax on capital unchanged.
  - Reweighting effect often reduces METR and can encourage investment; reweighting can reduce METR by 5 percentage points or more in several cases.
  - In four countries (Switzerland, Estonia, Germany and Canada) reweighting effect is so strong that METR falls on moving to RPA when routine profits taxed at statutory rates.
  - Overall, capital-based RPA raises METR in almost all countries if routine profits taxed at statutory rates.
  - If routine-tax rate set to leave tax on capital unchanged, METR is lower under RPA for most countries; nonetheless, high statutory tax countries in 2017 (U.S., France, Argentina, Belgium, Brazil, India, Mexico and Australia) see METR increase under capital-based RPA.

### Efficiency effects (capital-weighted METR) and interpretation (Table 12)
- Reported averages:
  - Average METR: Current = 13 ; RPA (routine taxed at statutory rate) = 28 ; RPA (routine taxed at current effective rate) = 11.
  - Capital-weighted average METR: Current = 17 ; RPA (routine taxed at statutory rate) = 38 ; RPA (routine taxed at current effective rate) = 18.
- Interpretation:
  - Capital-weighted average METR increases substantially (efficiency declines) if routine profits taxed at statutory rate.
  - If tax on routine profits set to leave effective tax on capital unchanged, weighted average METR and efficiency essentially unchanged.
  - Aggregate efficiency gains are unlikely unless routine profits taxed at a low rate or residual profits allocated by a factor markedly less sensitive to investment than capital itself.
  - Same implications hold if allocation weights are wholly unaffected by investment (e.g., destination-based RPA under proportionality assumptions).

### Additional substantive findings and policy-relevant points
- Residual profits are large in aggregate — perhaps ≈ 70 percent of total profits of the very largest MNE groups — and highly concentrated in firms headquartered in a handful of countries (notably the United States).
- Many large multinationals appear to have negative residual profits with persistence; handling residual losses is practically important.
- Low-income countries tend to gain revenue under RPA; this is not primarily because they run trade deficits but because under current arrangements they lose more, as a share of corporate tax revenue, than advanced economies.
- For many countries, routine profits exceed the current base of their CIT, implying potentially significant revenue gains from taxing routine profits under RPA.
- Key uncertainties and open issues:
  - Treatment of interest expense and financing costs in defining routine vs residual profits.
  - Identification of routine and residual profits from available data is challenging.
  - Tax-setting incentives and strategic government behavior under RPA remain to be analyzed.
  - Practical RPA proposals vary widely; operationalization may be difficult given available data.

### Appendix I — country-level variables and selected entries
- Final sample: 129 countries; variables averaged between 2015 and 2017.
- Selected country-level descriptive statistics (Table 3):
  - CIT revenue (billion USD): Minimum 0.01; Mean 17.45; Maximum 449.13; Sum 2250.79.
  - CIT base (billion USD): Minimum 0.02; Mean 67.72; Maximum 1789.37; Sum 8735.87.
  - Capital stock (billion USD): Minimum 0.27; Mean 565.75; Maximum 20235.09; Sum 72981.27.
  - CIT rate (percent): Minimum 9.00; Mean 24.29; Maximum 55.00.
  - Sales weight (percent): Minimum 0.00; Mean 0.78; Maximum 26.60.
  - Capital stock to GDP (percent): Minimum 21.40; Mean 104.14; Maximum 273.31.
  - CIT base to GDP (percent): Minimum 2.69; Mean 17.42; Maximum 98.52.
  - FDI to GDP (percent): Minimum 3.89; Mean 75.93; Maximum 1546.33.
- Representative country rows (Appendix I):
  - Malawi — Low income — GdpPc 308.09 — CitRev 0.21 — CitRate 30 — Salesweight 0.01 — Assets to Gdp 38.20 — Profit shifting share 0.98 — Reallocation share -32.43 — Total -31.46.
  - Central African Republic — Low income — GdpPc 359.58 — CitRev 0.01 — CitRate 30 — Salesweight 0.00 — Assets to Gdp 48.84 — Profit shifting share 6.40 — Reallocation share 277.40 — Total 283.80.
  - India — Lower middle income — GdpPc 1826.85 — CitRev 78.92 — CitRate 35 — Salesweight 3.22 — Assets to Gdp 114.73 — Profit shifting share 14.27 — Reallocation share 21.95 — Total 36.22.
  - China — Upper middle income — GdpPc 8422.08 — CitRev 449.13 — CitRate 25 — Salesweight 15.35 — Assets to Gdp 173.78 — Profit shifting share -7.34 — Reallocation share 20.14 — Total 12.79.
  - United States — High income — GdpPc 58332.70 — CitRev 394.54 — CitRate 39 — Salesweight 26.60 — Assets to Gdp 59.77 — Profit shifting share 22.53 — Reallocation share 47.03 — Total 69.56.
  - Luxembourg — High income — GdpPc 103688.99 — CitRev 2.91 — CitRate 28 — Salesweight 0.05 — Assets to Gdp 72.63 — Profit shifting share -0.20 — Reallocation share -51.22 — Total -51.42.

*Source: IMF staff, “2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits,” wpiea2020049-print-pdf.*

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

### References

### Tables
- 1. Descriptive Statistics by HQ Location
- 2. Descriptive Statistics by Sector
- 3. Descriptive Statistics: Global Sample
- 4. Proportion of Firms with Negative Residuals, by Sector
- 5. Duration of Spells of Implied Negative Residual Profit
- 6. Proportion of Firms with Negative PV of Residual Profit
- 7. Persistence of Residual Profit
- 8. Implied Residual Profit by Country of Headquarters, 2017
- 9. Drivers of the Revenue Impact of Moving to Destination- or Capital-Based RPA
- 10. Decomposition of Average Residual Reallocation Effects
- 11. Drivers of the Components of Reallocation of Excess Returns
- 12. Efficiency Effects of Moving to Capital-Based RPA

### Figures
- 1. Share of Routine Profit Varied by Routine Rate of Return
- 2a. Routine Profits in Percent of the Total Tax Base and its Correlation with Income Per Capita
- 2b. Capital Stocks divided by GDP
- 3. Proportion of Companies with Negative Residual Profit, 2017
- 4. Residual Relative to Total Profits for Firms with Positive Implied Residual
- 5. Distribution of Residual Profits by Firm Size
- 6. Revenue Gains from the Reallocation of Excess Profits under Destination-Based
- 7. Patterns of Revenue Effects from Reallocation Excess Profits in Moving to Destination-Based RPA
- 8. Country-Specific Decomposition of Revenue Effects of Moving to Destination-Based RPA
- 9. Reallocation of Excess Returns Under Destination-Based RPA by Income Group: Total and Decomposition
- 10. Revenue Impact of Changed Treatment of Routine Profits Under RPA
- 11. Current METRs, with Profit Shifting
- 12 METR RPA under Alternative Assumptions on the Tax Rate on Routine Profits
- 13. Effects on METRs of Moving to Capital-Based RPA

### Appendix
- 1. Country-Level Variables

*Source: wpiea2020049-print-pdf - References .........................................................................................................*

### 2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits ............................................................

### 2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits

### Introduction: context and motivation
- Recent policy openness to fundamentally new international corporate tax approaches has focused attention on schemes within the family of “Residual Profit Allocation” (RPA).
- RPA schemes:
  - Break multinational profits into routine profit (a normal return on activities/functions allocated to jurisdictions) and residual profit (aggregate excess profit).
  - Allocate routine profit by arm’s length or mechanical markup; allocate residual profit across jurisdictions by formulaic rules (e.g., destination-based sales).
  - Mark a departure from current norms by (i) eliminating scope for profit shifting via group-level allocation of residual profit and (ii) allocating taxing rights to jurisdictions (market/destination countries) that may currently have no corporate tax liability.
- Variants and related proposals cited in the text include:
  - Allocation by destination-based sales (Avi-Yonah, Clausing and Durst (2009)).
  - Allocation by adjusted destination-based sales to attribute residual to jurisdictions incurring sales-related costs (Auerbach and others (2019)).
  - Modified RPAs allocating only part of residual mechanically; examples: U.K. proposals allocating user-participation-derived residual and U.S. GILTI-like elements allocating part attributable to “marketing intangibles.”
- Key open empirical and policy questions motivating the paper:
  - How large are routine and residual profits in practice?
  - How are they distributed across multinational groups and countries?
  - Which countries would gain or lose revenue under RPA?
  - What are likely impacts on investment and efficiency?
- Paper objective: establish core stylized facts and analytical features needed to meaningfully assess RPA schemes, acknowledging substantial data and definitional challenges.

### Analyzing Residual Profit Allocation schemes: model setup and implications
- Model primitives:
  - Multinational invests k_i in country i, i = 1,..,N; aggregate capital K ≡ ∑_i k_i.
  - Financial costs and safe pre-tax return denoted r (treated as fixed); routine profit in country i corresponds to k_i r.
  - Total earnings before financing costs denoted Ẑ(k); residual profit Π(k) defined as Π(k) ≡ Ẑ(k) − ∑_i k_i r. (Equation (1))
- Under a generic RPA:
  - Jurisdiction i taxes routine profit at rate τ_i and is allocated share θ_i of residual profit Π, taxed at rate t_i.
  - Tax payable in i: T_i = t_i θ_i Π + τ_i k_i r. (Equation (2))
  - Aggregate weighted average tax on residual: t̄(k,t) ≡ ∑_i t_i θ_i.
  - After-tax profits: (1 − t̄) Π(k) − ∑_i τ_i k_i r. (Equation (3))
  - First-order condition yields the marginal effective tax rate (METR) in i under RPA: METR_i^RPA ≡ (∂Ẑ/∂k_i − r)/r = τ_i (1 − t̄) + t̄ ∂(θ·k)/∂k_i · Π/(r (1 − t̄)). (Equation (5) in text; second term shows investment distortion from k-dependence of allocation weights)
  - Implications highlighted:
    - Tax on routine profits amplified by residual taxation.
    - Distortions from jurisdictional taxation of residual when allocation weights θ_i depend on k (e.g., via destination sales affected by investment).
    - Company-specific effects; strength of second effect increases with size of residual profits.
- Under current arm’s length pricing (ALP):
  - Tax liability in i decomposed into: tax on declared excess profit π_i at rate t_i, and tax on routine return at rate h_i (deductible against t_i): T_i = t_i π_i + (1 − t_i) h_i k_i r. (Equation (6))
  - Purged excess profits π_i^* depend only on k_i; declared excess profits π_i = π_i^*(k_i) + s_i where s_i denotes profit shifting into i; ∑_i s_i = 0. (Equation (7) and constraint (8a))
  - Profit shifting is modeled with a nondeductible convex cost C(s_i, k_i); a specific functional form used later: C(k_i, s_i) = 1/2 ϑ (s_i^2 / (k_i r)). (Equation (11))
  - METR under ALP (allowing for profit shifting) is lower than the closed-economy METR h_i because profit shifting reduces the effective tax burden and affects investment incentives. (Equation (10))
  - With the specified shifting cost, METR under ALP becomes: METR_i^ALP = h_i − (1/(ϑ 2)) (1 − t_i) (s_i / (k_i r))^2. (Equation (12))
  - Profit shifting first-order condition implies s_i proportional to deviations of local tax rate from capital-weighted average tax rate elsewhere. (Equations (13)-(14))

### Data: sources, construction, and descriptive statistics
- Two datasets used to empirically explore routine and residual profits:
  - Firm-level consolidated MNE group data (S&P Capital IQ) covering top 10,000 largest MNEs for 2011–2017; baseline sample after cleaning: 8,854 MNEs.
    - Aggregate sample-level statistics for 2017:
      - Total assets: USD 658 trillion.
      - Total net fixed assets (including property, plant, equipment): USD 32 trillion.
      - Reported taxable profits: USD 5.3 trillion.
      - Total revenue of the smallest MNE in sample: USD 1.196 billion.
      - Average tax rate (total current tax divided by total pre-tax profit): around 21 percent.
    - Headquarters distribution (Table 1 summary):
      - Advanced Economies: Number of Companies 5,961; Share of Fixed Assets 57.5%; Share of EBIT 70.7%; Average Tax Rate 20.1%.
      - Emerging Markets: Number of Companies 2,044; Share of Fixed Assets 27.8%; Share of EBIT 19.8%; Average Tax Rate 22.2%.
      - Low-Income Developing Countries: Number of Companies 22; Share of Fixed Assets 0.1%; Share of EBIT 0.1%; Average Tax Rate 21.4%.
      - HQ Undisclosed: Number of Companies 827; Share of Fixed Assets 14.6%; Share of EBIT 9.4%; Average Tax Rate 19.7%.
    - Sectoral distribution and selected indicators (Table 2 summary; columns reported in source):
      - Example sector entries:
        - Agriculture, Forestry and Fishing: Number of Companies 30; Share of Fixed Assets 0.1%; Share of EBIT 0.1%; Average Tax Rate 24%; Return on fixed assets 24%; Mark-up 10%; K/COGS 40%.
        - Manufacturing: Number of Companies 2,940; Share of Fixed Assets 29.5%; Share of EBIT 40.0%; Average Tax Rate 24%; Return on fixed assets 26%; Mark-up 13%; K/COGS 51%.
        - Transportation, Communication, Utilities: Number of Companies 1,345; Share of Fixed Assets 39.8%; Share of EBIT 22.1%; Average Tax Rate 19%; Return on fixed assets 11%; Mark-up 19%; K/COGS 179%.
      - Averages:
        - Average, unweighted: Average Tax Rate 24%; Mark-up 12%; K/COGS 49%.
        - Average, weighted (with fixed assets): Average Tax Rate 19%; Mark-up 13%; K/COGS 65%.
    - Notes:
      - Firm-level consolidated accounts capture the level at which residual profits are defined but lack jurisdictional detail needed to attribute routine profits by country or allocate residual by destination sales.
  - Country-level macro data assembled to attribute routine profit and to simulate destination-based RPA revenue impacts:
    - Capital assets by jurisdiction from three sources:
      - UN national accounts statistics for gross fixed capital formation (perpetual inventory method) for 51 countries (approximate corporate non-financial capital stocks).
      - Private capital stock series from Bui and others (2019) for additional 77 countries, adjusted by the median ratio of non-financial corporate capital to non-financial private capital (to remove housing).
      - National accounts gross fixed capital formation from the WEO, used to provide information for an additional 32 countries (adjusted similarly).
    - Destination-based third-party sales proxied as GDP minus net exports from the IMF’s World Economic Outlook.
    - Country-level declared excess profits π_i estimated by rearranging the national-level corporate income tax revenue identity:
      - From T_i^CIT / t_i = π_i + (1 − t_i) h_i k_i r, so under assumption (1 − t_i) h_i / t_i ≈ 1, estimate π_i as (T_i^CIT / t_i) − k_i r, taking r = 7.5 percent (explicit assumption stated).
    - Statutory tax rate information taken from the IMF’s WoRLD database.
- Data challenges and caveats emphasized:
  - Routine and residual profit concepts have no direct accounting counterparts; combined consolidated and unconsolidated data required for full attribution are not consistently available.
  - Empirical work relies on plausible but untested assumptions and eclectic use of distinct data sources to construct approximations.

*Source: IMF staff, “2. Inferring ࣂ using the Semi-Elasticity of Taxable Profits,” wpiea2020049-print-pdf.*

### Appendix I lists, and provides key macro variables for all countries in the sample). Table 3

### Appendix I lists, and provides key macro variables for all countries in the sample). Table 3

### Sample and Descriptive Statistics
- Final sample comprises 129 countries.
- Variables averaged between 2015 and 2017 to reduce influence of outliers.
- Key country-level descriptive statistics (Table 3):
  - CIT revenue (in billion USD): Minimum 0.01, Mean 17.45, Maximum 449.13, Sum 2250.79
  - CIT base (in billion USD): Minimum 0.02, Mean 67.72, Maximum 1789.37, Sum 8735.87
  - Capital stock (in billion USD): Minimum 0.27, Mean 565.75, Maximum 20235.09, Sum 72981.27
  - CIT rate (in percent): Minimum 9.00, Mean 24.29, Maximum 55.00
  - Sales weight (in percent): Minimum 0.00, Mean 0.78, Maximum 26.60
  - Capital stock to GDP (in percent): Minimum 21.40, Mean 104.14, Maximum 273.31
  - CIT base to GDP (in percent): Minimum 2.69, Mean 17.42, Maximum 98.52
  - FDI to GDP (in percent): Minimum 3.89, Mean 75.93, Maximum 1546.33
- Note: Due to missing corporate tax revenue information, not all countries with capital stock data are included.

### Methods to Proxy Routine Profit
- Two methods to proxy routine profit (used to infer residual profit = earnings before interest and tax minus routine profit):
  - Tangibles-based approach (modeled on U.S. GILTI): apply notional rate of return to tangible assets (K at group-level, ݇௜ at country-level).
    - Benchmark notional rates emphasized: 7.5 percent and 10 percent.
  - COGS-based approach (group-level micro data only): apply markup to total direct COGS.
    - Benchmark mark-ups emphasized: 3.75 percent and 5 percent.
- Assumption: notional return on tangibles or markup on COGS is the same for all multinationals (noted as heroic).
- Relationship between methods: COGS nominally around twice tangible assets → notional return r on tangibles corresponds very roughly to one-half the markup on COGS.
- Caveat: Interest payments treated as deducted in calculating residual profit (alternative allocations discussed but not implemented due to data limits).

### Routine Profit Findings
- Illustration: at 10 percent return on tangibles or 5 percent markup on COGS:
  - Routine profits account for around 40-50 percent of total profits in absence of offset.
  - Routine profits account for about 30 percent of total profit of those earning positive profit (when negative residuals are offset).
- Country-level macro results:
  - Using a routine return on capital of 7.5 percent, routine profit exceeds current CIT base in 23 countries.
  - Robust negative relationship between ratio of routine profit to current CIT base and GDP per capita.
  - Top 10 countries by capital stock relative to GDP per capita are disproportionately emerging markets.

### Residual Profit Findings (MNE-level)
- Definition: residual profit Π(k) = earnings before interest and taxes − routine profit.
- In 2017, among MNEs with positive group-wide pre-tax profit (accounting for 90 percent of groups in sample):
  - Proportion with implied negative residual profit:
    - Tangibles-based, 10% return: 19 percent
    - COGS-based, 5% markup: 28 percent
  - Negative residual firms under mark-up approach (28 percent) own around 12 percent of fixed assets and earn around 1 percent of taxable profits.
  - Sensitivity to assumed routine return/markup:
    - Tangibles-based, 5% return: negative residual falls to 8 percent
    - COGS-based, 2% markup: negative residual falls to 11 percent
- Duration of negative residual spells (Table 5), tangible- and COGS-based examples:
  - At 7.5% tangible return: 0 years = 71.1%, 7 years = 2.0%
  - At 3.75% COGS markup: 0 years = 62.3%, 7 years = 5.5%
- Present value negativity (Table 6): proportion of firms with negative PV of residual profits over 2011-2017
  - Tangible-based (notional interest rate / notional rate of return):
    - 5% interest: 2.7% (5.0% return), 2.6% (7.5%), 2.7% (10.0%)
    - 7.5% interest: 8.4% (5.0%), 8.4% (7.5%), 8.5% (10.0%)
    - 10% interest: 14.8% (5.0%), 14.8% (7.5%), 14.9% (10.0%)
  - COGS-based (interest rate = one-half of tangible interest): 7.7% (2.5% markup, 5% interest), 13.9% (3.75%, 7.5%), 20.6% (5%, 10%)
- Residual profit share among firms with positive residuals (Figure 4):
  - Tangible-based, 10% return: residual averages about 70 percent of total profits for firms with positive residual.
  - Weighted by pre-tax profit, overall ratio residual/total ≈ 71 percent (10% return) and ≈ 75 percent (7.5% return).
  - COGS-based: overall ratio ≈ 68 percent (5% markup) and 66 percent (3.75% markup).
- Concentration of positive residual profits (Figure 5):
  - Largest 10 percent of corporate groups (by sales) account for more than half of all positive residual profits (tangibles-based, 7.5%).
  - Top 100 companies (~1% of sample) account for >28 percent of total residual profits, with residual/total profit ratio of 85 percent; these firms account for about 17 percent of tangible assets.
- Persistence (Table 7), at 7.5% tangible return:
  - Probability of staying in top quintile next year: 72.12 percent
  - Probability of staying in lowest quintile next year: 88.18 percent

### Geography and Headquarters Concentration (Table 8)
- Under various assumptions (tangibles 7.5% and 10%; COGS 3.75% and 5%), the top countries by headquarters of MNE parent capture large shares of implied residual profit:
  - United States: 33% (7.5% tangibles), 36% (10% tangibles), 33% (3.75% COGS), 35% (5% COGS)
  - China: 10% (7.5% tangibles), 7% (10% tangibles), 12% (3.75% COGS), 11% (5% COGS)
  - Japan: 8% (7.5% tangibles), 8% (10% tangibles), 7% (3.75% COGS), 6% (5% COGS)
  - United Kingdom, Germany, France, South Korea, Switzerland, Taiwan Province, PRC, Hong Kong SAR also in top 10 across specifications.
- Around 75-85 percent of aggregate residual profit accrues to companies headquartered in just 10 countries; more than half accrues to MNEs headquartered in the same four; United States alone accounts for around one-third.

### Revenue Effects of Moving to a Residual Profit Allocation (RPA) Scheme — Aggregate Results
- Framework: four effects identified when moving from current ALP to RPA (equation (16)): reallocation of residual profit relative to declared excess, change in revenue related to routine profit, change in revenue from investment responses, gain from efficiency increases. Sections quantify first two effects (reallocation of excess profits and routine rate effects) assuming universal adoption and no behavioral responses.
- Reallocation of excess profits under destination-based RPA (allocation weight ߱௜ = share of multinational’s sales to third-party residents of jurisdiction i):
  - Aggregate estimated increase in total tax revenue: USD 168 billion ≈ 7.1 percent of total CIT revenue in the sample.
  - Aggregate CIT revenue in sample: USD 2370 billion.
  - Unweighted average change in tax revenue: increase of 15 percent.
- Country-level variation (Figure 6 example results):
  - Largest proportionate increase: Central African Republic, 290 percent.
  - Largest reduction: Hong Kong SAR, -85 percent.
  - Broad tendency: lower income countries tend to gain proportionally more than higher income countries.
- Distributional incidence (Figure 7):
  - Around 30 percent of countries would lose at least 25 percent of their current CIT revenue.
  - Around 30 percent would gain at least 25 percent.
  - More large revenue gainers than large losers in count.

### Regression Drivers of Reallocation Effects (Table 9 summary)
- Dependent variable: simulated change in tax base (as percent of current tax base).
- Key statistically significant drivers (destination-based and capital-based allocations):
  - CIT rate: positive and significant. Approximate magnitudes:
    - Destination-based (Column 1): tax base increases by 2.083 (i.e., 2.083 percent of current tax base) for each percentage point higher CIT rate.
    - Destination-based (Column 2): 2.388***
    - Capital stocks (Column 3): 1.680**
    - Capital stocks (Column 4): 1.899***
  - log(GDP per cap.): negative and significant. Approximate magnitudes:
    - Destination-based: -15.436*** (Column 1), -18.484*** (Column 2)
    - Capital stocks: -10.548*** (Column 3), -15.550** (Column 4)
  - FDI/GDP>1.5 (investment hubs) : large negative and significant, e.g., -56.649*** (Column 1), -59.706*** (Column 2), -60.280*** (Column 3), -63.171*** (Column 4)
  - Resource rich: negative but not statistically significant in reported specifications.
  - Trade deficit: not significant.
- Intercepts reported (statistically significant): e.g., 20.281***, 22.001***, 15.266***, 17.445*** with reported standard errors.

### Decomposition of Reallocation of Excess Profits (Elimination of Profit Shifting vs. Reallocation of Purged Profits) — Key Quantitative Results (Table 10)
- Destination-based sales allocation, cost-function based identification:
  - Country-specific average (in percent of revenue): Profit shifting -0.43, Reallocation of purged profits 5.44, Total 5.02
  - Global revenue change (in percent of revenue): Profit shifting 2.61, Reallocation 4.48, Total 7.09
  - Global revenue change (in billion USD): 62 (profit shifting), 106 (reallocation), 168 total
- Destination-based sales allocation, activity-based identification:
  - Country-specific average (in percent of revenue): Profit shifting 3.9, Reallocation 1.12, Total 5.02
  - Global revenue change (in percent of revenue): Profit shifting 4.61, Reallocation 2.48, Total 7.09
  - Global revenue change (in billion USD): 109 (profit shifting), 59 (reallocation), 168 total
- Capital-based allocation, cost-function based identification:
  - Country-specific average (in percent of revenue): Profit shifting -0.43, Reallocation 4.33, Total 3.9
  - Global revenue change (in percent of revenue): Profit shifting 2.61, Reallocation 2, Total 4.61
  - Global revenue change (in billion USD): 62 (profit shifting), 48 (reallocation), 109 total
- Interpretation:
  - Total global revenue effect from reallocation of excess profits is robust at USD 168 billion (7.09 percent of global CIT revenue) under destination-based allocation, regardless of identification approach; composition between elimination of profit shifting and reallocation of purged profits changes with identification method.
  - Cost-function based approach attributes more of the global revenue change to reallocation of purged profits; activity-based approach attributes more to elimination of profit shifting.

### Identification Strategies to Estimate Profit Shifting
- Two approaches used to estimate profit-shifting component (needed to decompose effects):
  - Cost-function based approach:
    - Estimate parameter ߠ of the cost function; under assumption Ε(ߚ௜ | ݇௜) = constant, calibrate ߠ using consensus estimates of semi-elasticity of reported gross profits with respect to home-foreign tax rate differential.
    - Leads to choice ߠ = 2.55.
    - ߚ௜ implied from ߚ௜ ߨ = (ݏെݜ)/Π relation.
  - Activity-based approach:
    - Assume excess (purged) returns are proportional to country-specific capital stocks: ߚ௜ = ݇௜/∑݇.
    - Profit-shifting then estimated as residual ݏ௜ = ߨ௜ − ߚ௜ Π.
- Comparison:
  - Cost-function based approach centers on statutory CIT differentials as driver of shifting; activity-based approach captures shifts tied to distribution of real activity/capital and possibly special tax regimes not reflected in statutory rates.
  - Empirical decomposition (Figure 8) shows that which effect dominates (elimination of shifting vs. reallocation of purged profits) depends strongly on identification approach.

### Regression Drivers of Decomposed Effects (summary)
- Elimination of profit shifting:
  - Under cost-function based identification, CIT rates are the main driver of revenue changes due to elimination of profit shifting; other variables including GDP per capita are not significant.
- Reallocation of purged profits:
  - Tends to generate revenue losses in investment hubs and benefits less advanced economies.
- Capital-based allocation (vs. destination-based) tends to reduce the reallocation-of-purged-profits effect compared with destination-based allocation.
- Overall pattern: higher statutory CIT rates → larger gains under RPA reallocation; lower-income countries tend to gain proportionally more; investment hubs (high FDI/GDP) lose large shares of tax base.

*Source: IMF working paper content unit (wpiea2020049-print-pdf).*

### conclusions apply under capital-based RPA, though the country-specific effects are of course

### wpiea2020049-print-pdf - conclusions apply under capital-based RPA, though the country-specific effects are of course

### Reallocation of Excess Returns and Revenue Effects
- Moving to a destination-based RPA increases global corporate tax revenue by around 8 percent.
- Decomposition results (Table 11 / Figure 9):
  - Elimination of profit shifting (PS) tends to benefit countries with high CIT rates; this correlation is empirically observed under the activity-based approach as well.
  - Investment hubs lose revenue, in part or entirely through the elimination of profit shifting depending on specification.
  - Developing countries tend to gain more from the elimination of profit shifting than advanced economies.
  - Reallocation of purged profits (Realloc.) benefits developing countries significantly, though the scale of the effect varies by identification approach.
- Empirical examples and magnitudes:
  - Table 11 regressions show statistical relationships between components and covariates including CIT rate, log(GDP per cap.), resource richness, and FDI/GDP>1.5; adjusted R-squared values reported include 0.703, 0.139, 0.703, 0.08, 0.111, 0.078 across columns.
  - Observations: 129 in each reported regression.

### Routine Profit Effect (treatment and revenue implications)
- Two benchmark assumptions on the tax treatment of routine profits:
  1. Routine profits taxed at current statutory rates (ݐ௜ = current statutory rate).
     - Revenue effects expressed as percentage of current revenue can be large and may be negative where closed economy METR (݉௜) exceeds grossed-up statutory rate.
     - Estonia example: METR is 31 percent while statutory tax rate is 20 percent, leading to losses of 31 percent of current tax revenue if routine returns were taxed at the statutory rate.
     - Argentina example: statutory rate 35 percent while closed economy METR is 27 percent: revenue would then increase by 101 percent.
     - Key point: revenue would substantially increase in virtually all countries if routine profits were taxed at their current statutory rates.
  2. Routine profits taxed so as to leave the marginal tax on investment at its current level (ݐ௜ = (1 − ݐ1) ݉௜ in the notation provided; equation (18)).
     - Under this assumption, there is no impact on revenue through the treatment of routine profit: the second term in (16) vanishes.
- If some allowance for financing costs is made against routine profits, revenue impacts would be smaller; effects might shift toward the residual component.

### Impact on Investment and METRs
- Current METRs under ALP:
  - METRs average around 15 percent with wide variation.
  - The profit-shifting term is generally smaller than the closed-economy term but can dominate in countries with very high or very low statutory tax rates.
  - Example: Bulgaria’s closed economy METR is sufficiently low that the profit shifting effect turns the overall METR negative.
- Calculating METR under RPA:
  - Three assumptions required: rate at which routine profits are taxed; weighting scheme for allocation of residual profit (focus on capital stock ݇௜ weighting as in (19)); ratio of residual to routine profit (routine return assumed 7.5 percent).
  - Under capital-based weighting, METR under RPA (equation (20)) includes a “weighting” term that raises (lowers) the METR for investment in countries with relatively high (low) statutory tax rates on residual profits.
- Figure 12 and Figure 13 findings:
  - METR is higher when routine profits are taxed at current statutory rate than when taxed to leave effective tax on capital unchanged.
  - The reweighting effect (investment changing the overall average rate) often reduces the METR and can encourage investment; in several cases the reweighting effect reduces the METR by 5 percentage points or more.
  - In four countries (Switzerland, Estonia, Germany and Canada) the reweighting effect is so strong that the METR actually falls on moving to the RPA scheme when routine profits are taxed at current statutory rates.
  - Overall, moving to capital-based RPA raises the METR in almost all countries under routine profits taxed at statutory rates.
  - If routine tax rate is set so as to leave tax on capital unchanged, for most countries the METR is lower under RPA than under current arrangements; nonetheless, some high statutory tax countries (in 2017: U.S., France, Argentina, Belgium, Brazil, India, Mexico and Australia) see METR increase under capital-based RPA.

### Efficiency Effects
- Efficiency measured via capital-weighted average METR:
  - Table 12 reports:
    - Average METR: Current = 13 ; RPA with routine profits taxed at current statutory rate = 28 ; RPA with routine profits taxed at current effective rate = 11
    - Capital-weighted average METR: Current = 17 ; RPA with routine profits taxed at current statutory rate = 38 ; RPA with routine profits taxed at current effective rate = 18
  - Interpretation:
    - Capital-weighted average METR increases substantially (efficiency declines) if routine profits are taxed at the statutory tax rate.
    - If the tax on routine profits is set so as to leave the effective tax on capital unchanged, the weighted average METR, and hence efficiency, is essentially unchanged.
    - In neither case is aggregate efficiency manifestly increased by moving to the RPA scheme under the assumptions used.
    - Aggregate efficiency gains are unlikely unless routine profits are taxed at a low rate or residual profits are allocated by a factor markedly less sensitive to investment than capital itself.
  - Note: The implications for aggregate efficiency are the same in the opposite polar case in which weights are wholly unaffected by investment (e.g., destination-based RPA under proportionality assumptions), given the capital-weighted calculations.

### Additional substantive findings and policy-relevant points
- Residual profits are, in aggregate, considerable—perhaps 70 percent or so of the total profits of the very largest multinational groups—and are highly concentrated in companies headquartered in a handful of countries, notably the United States.
- Many large multinationals appear to have negative residual profits with some persistence; handling residual losses is a practical issue of real importance.
- Low income countries tend to gain revenue under RPA; this is not primarily because they run trade deficits but because under current arrangements they lose more, as a share of total corporate tax revenue, than more advanced economies.
- For many countries, routine profits exceed the current base of their CIT, implying potentially significant revenue gains from taxing routine profits under an RPA.
- Key uncertainties and open issues:
  - Treatment of interest expense and financing costs in defining routine vs residual profits.
  - Identification of routine and residual profits from available data is challenging.
  - Tax-setting incentives and strategic behavior by governments under RPA remain to be analyzed.
  - Practical RPA proposals vary widely in extent and manner of residual allocation; operationalization may be difficult given available data.

*Source: Authors’ computations and analysis as presented in the supplied PDF content.*

### Appendix I. Country-Level Variables

### Appendix I. Country-Level Variables

### Column headings
- Country
- Income group
- GdpPc (Unit (in USD))
- CitRev (in billion USD)
- CitRate (in percent)
- Salesweight (in percent of CIT base)
- Assets to Gdp
- Profit shifting share
- Reallocation share
- Total

### Country-level entries (first rows as listed)
- Malawi — Low income — 308.09 — 0.21 — 30 — 0.01 — 38.20 — 0.98 — -32.43 — -31.46
- Central African Republic — Low income — 359.58 — 0.01 — 30 — 0.00 — 48.84 — 6.40 — 277.40 — 283.80
- Mozambique — Low income — 477.51 — 0.98 — 32 — 0.02 — 52.82 — 1.63 — -59.25 — -57.61
- Afghanistan — Low income — 582.02 — 0.24 — 20 — 0.04 — 40.18 — -11.05 — 66.72 — 55.66
- Togo — Low income — 589.36 — 0.08 — 29 — 0.01 — 98.22 — 1.78 — 96.44 — 98.22
- Uganda — Low income — 674.57 — 0.28 — 30 — 0.04 — 90.94 — 7.52 — 209.41 — 216.93
- Gambia — Low income — 699.02 — 0.02 — 31 — 0.00 — 58.03 — 6.31 — 103.09 — 109.40
- Rwanda — Low income — 755.16 — 0.12 — 30 — 0.01 — 53.92 — 3.58 — 98.96 — 102.55
- Ethiopia — Low income — 760.13 — 1.40 — 30 — 0.11 — 55.31 — 2.53 — 37.91 — 40.44
- Nepal — Low income — 798.09 — 0.71 — 25 — 0.04 — 93.55 — -5.00 — 6.85 — 1.85
- Benin — Low income — 809.50 — 0.14 — 30 — 0.01 — 112.92 — 6.75 — 156.35 — 163.09
- Mali — Low income — 813.27 — 0.37 — 30 — 0.02 — 82.31 — 2.94 — 28.09 — 31.03

### Additional representative rows (selected mid- and high-income economies)
- India — Lower middle income — 1826.85 — 78.92 — 35 — 3.22 — 114.73 — 14.27 — 21.95 — 36.22
- China — Upper middle income — 8422.08 — 449.13 — 25 — 15.35 — 173.78 — -7.34 — 20.14 — 12.79
- Brazil — Upper middle income — 9387.30 — 69.91 — 34 — 2.66 — 133.83 — 13.49 — 23.12 — 36.61
- United States — High income — 58332.70 — 394.54 — 39 — 26.60 — 59.77 — 22.53 — 47.03 — 69.56
- Japan — High income — 36456.47 — 204.43 — 31 — 6.34 — 81.46 — 2.89 — -28.62 — -25.73
- Luxembourg — High income — 103688.99 — 2.91 — 28 — 0.05 — 72.63 — -0.20 — -51.22 — -51.42

### Key data features evident from the table
- The table pairs country-level GDP per capita (GdpPc) with measures of corporate income tax revenue (CitRev), statutory CIT rate (CitRate), and the salesweight share (Salesweight) used in the CIT base.
- Two adjustment columns capture distributional impacts from profit shifting and reallocation: Profit shifting share and Reallocation share; a Total column aggregates their net effect.
- Values include negative and positive effects; examples:
  - Central African Republic: Profit shifting share 277.40, Reallocation share 283.80 (large positive net Total).
  - Japan: Profit shifting share -28.62, Reallocation share -25.73 (negative net Total).
- CitRev is reported in billion USD; many large economies show high CitRev values (e.g., China 449.13, United States 394.54, Japan 204.43).

*Appendix I lists country-level variables as shown in the source table.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020049-print-pdf.pdf_