## ch3onlineannex

## Source details

**Canonical URL:** [ch3onlineannex](https://www.imf.org/-/media/files/publications/weo/2026/october/english/ch3onlineannex.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/weo/2026/october/english/ch3onlineannex.pdf.md)
- [Structured JSON version](/-/media/files/publications/weo/2026/october/english/ch3onlineannex.pdf.json)

---

### Decomposition of BETR changes for MNCs with Headquarters in the United States
- Periods analyzed: 2010–17 and 2018–23 (percentage point unless otherwise stated).
- BETR (start, percent): 29.3 (2010–17); 21.0 (2018–23).
- BETR (end, percent): 21.0 (2010–17); 18.9 (2018–23).
- ΔBETR (end-period value minus start-period value): –8.3 (2010–17); –2.0 (2018–23).
- Component contributions (percentage point):
  - Domestic Taxation: –3.3 (2010–17); –3.9 (2018–23)
    - Domestic Statutory: –0.1 (2010–17); –8.3 (2018–23)
    - Domestic Base: –3.2 (2010–17); 4.4 (2018–23)
  - Foreign Rates + Bases: –3.9 (2010–17); 2.5 (2018–23)
    - Foreign Statutory: –0.7 (2010–17); –0.0 (2018–23)
    - Foreign Base: –3.2 (2010–17); 2.5 (2018–23)
  - Profit Shifting: –1.7 (2010–17); –0.0 (2018–23)
  - Distribution: –1.5 (2010–17); –0.2 (2018–23)
  - Globalization: –0.1 (2010–17); 0.2 (2018–23)
  - Residual: 0.5 (2010–17); –0.6 (2018–23)
- Key interpretation points:
  - Pre-TCJA (2010–17) decline in BETR was mostly explained by foreign tax policy and profit shifting.
  - Post-TCJA (2018–23) the domestic statutory rate effect (reflecting the statutory rate cut from 35 to 21 percent) dwarfs every other component.
  - The domestic base effect turns positive post-TCJA, consistent with base-broadening provisions and profit repatriation under the new territorial regime.
  - The contribution of profit shifting is approximately zero in 2018–23.
- Data sources: U.S. Bureau of Economic Analysis (2009–2023) and IMF staff calculations.
- Note: BETR = backward-looking effective tax rate; ΔBETR = change in BETR (end-period value minus start-period value).

### Decomposition of BETR changes for MNCs with Headquarters in the European Union
- Sample subperiods: 2010–17 and 2018–21 (percentage point unless otherwise stated).
- BETR (start, percent): 34.9 (2010–17); 17.3 (2018–21).
- BETR (end, percent): 17.3 (2010–17); 17.1 (2018–21).
- ΔBETR: –17.4 (2010–17); –0.1 (2018–21).
- Component contributions (percentage point):
  - Domestic Taxation: –12.5 (2010–17); –0.4 (2018–21)
    - Domestic Statutory: –1.0 (2010–17); –0.7 (2018–21)
    - Domestic Base: –11.5 (2010–17); 0.3 (2018–21)
  - Foreign Rates + Bases: 1.4 (2010–17); 0.3 (2018–21)
    - Foreign Statutory: –0.5 (2010–17); –0.4 (2018–21)
    - Foreign Base: 1.9 (2010–17); 0.7 (2018–21)
  - Profit Shifting: −7.4 (2010–17); 0.1 (2018–21)
  - Distribution: –0.9 (2010–17); –0.1 (2018–21)
  - Globalization: –6.5 (2010–17); 0.2 (2018–21)
  - Residual: 1.0 (2010–17); –0.2 (2018–21)
- Memo and coverage:
  - Country reweighting: –0.3 (2010–17); 0.1 (2018–21)
  - Coverage (EU parent countries): 25.0 (2010–17); 26.0 (2018–21)
- Data sources: Orbis (2008–2021 vintages) and IMF staff calculations.
- Note: Percentage-point contributions from chained annual decomposition, PPP-GDP weights anchored at each annual end year. Croatia excluded due to data limitations.

### Robustness and cross-checks
- Country-by-country reporting (CbCR) data used to cross-check BEA and Orbis decompositions; results are consistent with BEA and Orbis for the later period.
- US industry-level results and additional robustness results available upon request (as reported by IMF staff).

### Local Projection Estimates of CIT and PIT Shock Spillovers (Online Annex 3.2)
- Objective: Examine cross-border spillovers of corporate income tax (CIT) and personal income tax (PIT) shocks originating in the United States, the United Kingdom, and Germany.
- Identification and instruments:
  - Narrative tax shocks used as instruments following Mertens and Ravn (2013).
  - Structural tax shocks (z_CIT,s,t and z_PIT,s,t) recovered and used in a local projection instrumental variable (LP-IV) framework.
- Shock sample coverage (quarters):
  - Germany: 1992Q1–2018Q2
  - United Kingdom: 1993Q1–2007Q4
  - United States: 1951Q1–2019Q4
- Recipient panel: 29 AEs and EMEs.
- Methodology highlights:
  - Stacked dyadic LP-IV specification per Jordà (2005).
  - Endogenous regressor: source country’s average corporate and personal income tax rate (A_CITR_s,t or A_PITR_s,t), instrumented by corresponding structural shock.
  - Controls: four lags of source country A_PITR and A_CITR; four lags of recipient output growth; contemporaneous + four lags of the Global Financial Cycle index; robustness checks include external demand and Chicago Fed’s Adjusted National Financial Conditions Index.
  - Impulse response to a tax cut traced by {−β_h}_h=0.
  - Standard errors: Driscoll and Kraay (1998) estimator.
- Power and validity:
  - Instruments jointly significant in the first stage with F-statistic for excluded instruments well above 20 at each horizon.
  - Re-estimation controlling for contemporaneous tax shocks in other source countries reduces sample size and power; U.S. spillovers remain qualitatively similar but less precise.
- Empirical findings:
  - Country-level estimates more precise for U.S. spillovers due to longer time series.
  - Spillovers from U.K. and German tax shocks are not statistically different from each other.
  - Spillovers from PIT shocks generally larger than from CIT shocks for Germany and the U.K.
  - Figures show cumulative impact of a one percentage point decrease in average PIT or CIT in Germany, the U.K., or the U.S., instrumented by narrative shocks, on real GDP and real investment in the rest of the world (90 percent confidence intervals reported in source).

### GIMF Spillover Simulations (Online Annex 3.3)
- Model: IMF’s Global Integrated Monetary and Fiscal model (GIMF); 8-region setup (United States, euro area, Japan, Other Advanced Economies, China, Other Asia, Latin America, remaining economies).
- Key model features:
  - Profit-maximizing firms in monopolistic competition across non-tradable, tradable, and Global Value Chain (GVC) sectors; calibration to OECD Inter-Country Input-Output Tables (OECD 2021).
  - Frictions: sticky prices and wages, real adjustment costs, liquidity-constrained households, overlapping-generations (OLG) households.
  - Fiscal instruments: corporate and wage income taxes, government consumption, investment, transfers; governments can issue debt.
  - International capital flows: households can hold government debt and trade tradable U.S.-dollar-denominated obligations; capital mobile across regions via arbitrage in real returns.
  - Firms taxed on economic profits and capital rental income; depreciation and interest payments deductible.
- Transmission mechanisms emphasized:
  - CIT cuts raise after-tax marginal product of capital in implementing region, attracting savings and raising global real interest rates.
  - Higher global real interest rates increase user cost of capital elsewhere, weakening investment abroad — macroeconomic analogue to FDI-relocation channel.
  - PIT cuts stimulate labor supply, boosting disposable income and consumption; they exert smaller upward pressure on global real interest rates relative to CIT cuts.
- Quantitative simulation results:
  - A permanent unilateral 5-percentage point CIT cut in a major advanced economy (debt-financed for 40 years) leads to:
    - Larger increases in interest rates abroad and larger declines in investment abroad than a 5-percentage point PIT cut.
    - Negative GDP spillovers to other regions: higher global interest rates and reallocation of investment to implementing country more than offset positive external demand.
  - A permanent unilateral 5-percentage point PIT cut produces:
    - Smaller rise in global interest rates and smaller fall in investment abroad.
    - Overall external effect broadly neutral, due to stronger labor supply response and smaller demand-capacity gap.
  - Deficit financing amplifies these dynamics: when CIT cut is not offset by lower spending or higher taxes elsewhere, public borrowing rises and higher world real interest rates induce higher private saving and lower investment at home and abroad.
- Interpretation: Macroeconomic channels in GIMF, even absent MNC-specific relocation, can generate opposite-sign spillovers for CIT versus PIT cuts through global interest rate and investment reallocation effects.

### Strategic tax competition: estimation approach and key coefficients
- Model: peer effects specification estimating country i’s CIT response, τ_{i,t} = α + β1 τ_{i,t}^{peers} + β2 X_{i,t} + β3 Σ_w_{ij,t} X_{j,t} + φ_W + φ_{W,t} + θ_i + γ_t + u_{i,t}.
- Baseline peer weights: geographical distance; alternative weights: similarity in receiving FDI inflows (gravity model).
- Identification: instruments exploit peers’ political-cycle changes (time since election, share of opposition), with controls for domestic political shocks.
- Interpretation: β1 > 0 implies strategic complements; β1 < 0 implies strategic substitutes.
- Selected IV fixed-effects estimates (Peers Tax Rate coefficients and standard errors):
  - Statutory CIT (pooled): 0.404*** (SE 0.039); interacted: 0.431*** (SE 0.058).
  - EATR (pooled): 0.285*** (SE 0.137); interacted: 0.297*** (SE 0.115).
  - EMTR (pooled): 0.312** (SE 0.155); interacted: 0.345*** (SE 0.132).
  - Peers Tax Rate × (year >= 2017): -0.169** (SE 0.08) in one specification; other specifications show -0.026 (SE 0.059) and -0.040 (SE 0.07).
  - Post-2016 Response (Peers Tax Rate + Peers Tax Rate × (year >= 2017)): 0.262*** (SE 0.06), 0.270** (SE 0.122), 0.305** (SE 0.142).
- Estimation diagnostics:
  - Observations: 3,875; 3,870; 2,618 (depending on specification).
  - Countries: 152; 152; 151.
  - Cragg-Donald Wald F and Sanderson-Windmeijer first-stage F reported (examples): 80.4; 34.4; 39.4; 29.6; 31.9; 23.5.
- Controls: lagged dependent variable, own-country political variables and interactions, log GDP, log population, trade openness, spatial lags of peer log GDP, log population and trade openness; country and year fixed effects.
- Policy implication: Positive and sizable peer effects imply governments’ CIT choices are responsive to peers’ CIT levels, supporting cross-border interdependence in CIT setting.

### Spillovers from base shifting: bilateral FDI gravity and sectoral estimation
- Bilateral gravity specification for cumulative FDI (percent of destination GDP at time t−1): Y_{ij,t+h} = η (τ_{i,t} − τ_{j,t}) + ζ1 X_{i,t} + ζ2 X_{j,t} + ζ3 X_{ij,t} + ε_{ij,t}; η = semi-elasticity of FDI to tax differentials.
- Tax differential instrumented with peers’ political cycle.
- Bilateral FDI source: World Bank Harmonized Bilateral FDI Database (Steenbergen and others 2022).
- Sector-level specification allows heterogeneity by intangibles intensity x(s) = {tangible, intangible}; intangible sectors include Information, Legal Services, Management of Companies and Enterprises.
- Outcomes: (a) count of brownfield and greenfield projects (extensive margin); (b) level of investment as percent of GDP (intensive margin).
- Intensive margin: control function approach; extensive margin: 2SLS.

### Profit-shifting estimation: data, methodology, and main results
- Data:
  - Orbis unconsolidated balance sheet panel, 2009–23, real (2019) USD.
  - Unit: subsidiary f in country j, sector k, year t, linked to global ultimate owner (GUO) in country i.
  - Sample: about 500,000 MNE groups and over 3,000,000 subsidiary-year observations; covers over 100 host countries and headquarters in over 40 jurisdictions.
- Outcome and main independent variable:
  - Dependent variable: return on assets (ROA) = EBIT / assets; reported taxable income treated as censored at zero.
  - Tax differential at affiliate jurisdiction j: τ̅_{j,t} = τ_{j,t} − Σ_{l≠j} ω_{l,t} τ_{l,t}, where ω are asset weights across jurisdictions where the MNC has affiliates.
- Main specification for ROA:
  - ROA_{G,i,f,j,k,t} = β0 + β1 τ̅_{j,t} + γ′ X_{f,j,t} + δ′ X_{G,i,t} + ϵ′ X_{j,k,t} + φ_G + φ_k + φ_{j,t} + u_{...}.
  - Controls: affiliate-level (log turnover, log employment), industry-jurisdiction-level (turnover-based Herfindahl), conglomerate-level (MNC profitability); fixed effects: year, conglomerate, subsidiary-jurisdiction-year.
- Elasticity interpretation:
  - Estimated elasticity of reported income to tax differentials for an MNC with average ROA: −1.2 (Online Annex Table 3.5.1, column 1).
- Heterogeneity by intangibles intensiveness and time:
  - Intangibles intensiveness at GUO level: ratio of intangible fixed assets to total fixed assets; high-intangibles-intensive (I_G = 1) if GUO ratio ≥ 50 percent (75th percentile).
  - Post-2017 dummy D_{2017} = 1 for year ≥ 2017.
- Main estimates (baseline regressions, dependent variable: censored ROA; sample observations = 3,198,284; R² = 0.65):
  - τ − Σωτ coefficient: −0.327*** (SE 0.0894) in column (1); −0.190* (SE 0.101) in column (2); −0.239* (SE 0.126) in column (3).
  - (τ − Σωτ) × high-intangibles: −0.280*** (SE 0.0780) in column (2); −0.546*** (SE 0.126) in column (3).
  - (τ − Σωτ) × post 2017: 0.251** (SE 0.117) in column (2).
  - (τ − Σωτ) × post 2017 × high-intangibles: −0.368** (SE 0.143) in column (2).
  - (τ − Σωτ) × TP doc. HQ: 0.242** (SE 0.104) in column (3).
  - (τ − Σωτ) × TP doc. HQ × high-intangibles: 0.243* (SE 0.104) in column (3); reported −0.126 elsewhere.
  - TP doc. HQ (non-interacted): −1.662** (column (2)) and −0.705 (column (3)).
- Implied semielasticities (for an MNC with average ROA):
  - Overall profits semielasticity for average affiliate: Low II = −1.168*** (SE 0.319) in column (1); −0.679* (SE 0.361) in column (2); −2.148 (SE 0.353) in column (3).
  - Low II post-2017 / TP doc: 0.219 (SE 0.483) and −0.488 (SE 0.332) in alternative rows.
  - High II: −1.678*** (SE 0.348).
  - High II post-2017 / TP doc: −2.095*** (SE 0.426).
- Interpretation of heterogeneous effects:
  - Low-intangibles MNCs: semielasticity before 2017 ≈ −0.68; after 2017 semielasticity ≈ 0.2 (not statistically different from zero) — marked decline in profit-shifting responsiveness after 2017.
  - High-intangibles MNCs: semielasticity before 2017 ≈ −1.7; after 2017 ≈ −2.1 — persistent and larger sensitivity to tax differentials for intangibles-intensive firms.
  - Transfer-pricing documentation requirements (TPDRs): for low-intangibles MNCs, semielasticity is −0.85 without TPDRs and statistically indistinguishable from zero with TPDRs; for high-intangibles MNCs, semielasticity is −2.8 without TPDRs and falls (statistically significant difference of 1.7) with TPDRs.
  - Overall implication: profit-shifting responsiveness declined for low-intangibles firms after 2017 and with TPDRs, but high-intangibles-intensive MNEs remain more responsive to tax differentials, with continued profit-shifting despite anti-avoidance efforts and expansion of profit-based tax incentives for intangible-derived income after 2017.

### Microeconometric results: effect of EMTR differentials on investment rate
- Baseline: a 1 percent increase in the tax differential reduces the investment rate by 0.148.
- Key reported coefficients (Online Annex Table 3.7.1):
  - Effective Marginal Tax Rate Differentials: −0.148**, −0.117*, −0.223**, −0.163** (standard errors: (0.060), (0.062), (0.097), (0.069)).
  - Effective Marginal Tax Rate Differentials × Post 2017: −0.134**, −0.385***, −0.056 (standard errors: (0.068), (0.088), (0.090)).
  - Observations: 6,235,788; 6,235,788; 6,235,782; 4,422,576 (columns 1–4 respectively).
  - R²: 0.13 across reported specifications.
  - Significance codes: *** p<0.01, ** p<0.05, * p<0.1.
- Interactions with origin-country anti-avoidance regulations and intangible intensity:
  - Earnings stripping rules interaction: −0.511*** (SE (0.073)).
  - Transfer pricing documentation requirements interaction: −0.607*** (SE (0.072)).
  - Controlled foreign corporation (CFC) rules interaction: −0.005 (SE (0.079)).
  - IP box regimes interaction: 0.242*** (SE (0.066)).
  - Intangible-intensive firms interaction on investment margin: −0.141 (SE (0.090)).
  - Observations for interaction regressions: 6,340,678 (columns 1–4) and 6,222,510 (column 5).
- Interpretation:
  - Stricter earnings stripping and TPDRs increase sensitivity of investment to destination taxes.
  - IP box regimes reduce tax-sensitivity of investment, suggesting these regimes lower the effective tax burden and diminish relocation incentives.
  - Intangible intensity does not show significant difference on the investment margin in these interactions.

### Structural model of multinational production, profit shifting, and tax competition (overview)
- Model type: multi-region spatial general equilibrium extending Dyrda, Hong, and Steinberg (2024a, 2024b) to a seven-region world and incorporating productive public capital.
- Regions: major AE; other AEs; major EME; G20 EMEs; productive low-tax jurisdictions (PLTJs); rest of world (ROW: other productive EMDEs); unproductive low-tax jurisdictions (ULTJs).
- Firm heterogeneity: productivity draws from a Pareto distribution; firms decide export destinations, affiliate locations, intangible capital production, and share of profits shifted to LTJs.
- Intangible capital: produced via domestic R&D labor; non-rival use across affiliates → increasing returns to R&D; MNCs undertake vast majority of intangible investment.
- Profit shifting: firms can locate ownership of intangible assets in PLTJs and ULTJs, transfer licensing rights after paying fixed costs, and choose share of licensing income earned by LTJ affiliates; profit shifting subject to convex cost capturing tax-planning resource use.
- Productive public capital: effective TFP influenced by public capital stock; accumulation financed by a share of CIT revenue (remainder from other tax sources fixed in counterfactuals).

### Defining low-tax jurisdictions (classification thresholds)
- Four thresholds used: (1) CIT abuse index (Tax Justice Network); (2) statutory CIT rate (including special regimes for MNCs); (3) FDI stocks (average of assets and liabilities) as a share of GDP; (4) population size.
- ULTJ criteria:
  - CIT abuse indicator above 60;
  - FDI stocks larger than the size of the economy;
  - statutory CIT rate lower than 10 percent;
  - population size less than 2 million.
- Economies satisfying three criteria—including a low CIT rate and low population size—are also classified as ULTJs.
- PLTJ classification: economies that satisfy two criteria, or three criteria including small population size or low CIT rate.

### Transmission channels of CIT rate changes (model implications)
- Direct effects on the tax-cutting economy:
  - Lower CIT rates stimulate intangible investment and innovation by reducing user cost of capital.
  - Lower CIT rates narrow tax differentials, reducing incentive to shift profits and freeing resources for production and R&D.
  - Lower CIT rates reduce CIT revenue and public capital investment, lowering resources for productive public capital and impacting TFP.
  - Net direct effect depends on fiscal productivity (output elasticity of public capital) and share of public investment financed by CIT revenue; higher calibrated values in EMDEs imply fiscal cost of tax competition disproportionately borne by EMEs.
- Spillover effects to other regions:
  - Base-shifting channel: the tax-cutting economy becomes relatively more attractive for MNC investment and profit booking, eroding corporate tax base of other economies.
  - Intangible capital channel: lower tax rate in a tax-cutting country reduces user cost of capital, incentivizing additional R&D and larger accumulation of intangible capital across MNC affiliates, raising productivity of foreign affiliates.

### Calibration, model-implied elasticities, and alignment with empirical estimates
- Key calibration parameters preserved exactly:
  - Elasticity of Substitution Between Varieties (ρ) = 4.
  - Labor Share in Production (γ) = 0.85.
  - Intangible Capital Share in Production (ψ) = 0.1.
  - TFP Elasticity to Public Capital (δ) = 0.05.
  - Share of Public Capital Investment in Government Revenue = 0.09–0.47.
  - Pareto Shape Parameter (η) = 2.9–3.1.
- Model-implied semi-elasticities (Online Annex Table 3.8.2):
  - Extensive margin (location choice): microeconometric estimate −1.4; model-implied −1.6.
  - Intensive margin (affiliate scale): microeconometric estimate −0.2; model-implied −0.3 to −0.8.
  - Aggregate FDI: micro estimate −2.0; model-implied −2.0 to −2.4.
  - Profit shifting (reported profits): microeconometric estimate −1.2 to −2.0; model-implied −0.9.
- Note: model’s profit-shifting semi-elasticity ≈ −0.9 is within the literature range (−0.8 to −1.1) and close to microeconometric estimate −1.2.

### Counterfactual policy experiments and sensitivity
- Counterfactual 1 (reciprocated CIT cuts):
  - Major AE cuts statutory CIT rate by 3 percentage points; other AEs cut by 1.2 percentage points in response; all EMEs assumed to cut by 1.2 percentage points.
- Counterfactual 2 (anti-avoidance enforcement instead of EME cuts):
  - AEs implement same CIT cuts as scenario 1; all EMEs strengthen anti-avoidance regulation instead of matching tax cuts.
  - Enforcement improvements calibrated to target reductions in profit shifting between 10 and 20 percent, comparable to enforcement gains achieved by AEs over the past decade.
- Robustness/sensitivity:
  - Varying the CIT rate of a large economy while holding others constant generates a standard Laffer curve for revenue and an inverted U-shape for domestic output—reductions in CIT can boost revenues and output when initial rates are very high, but beyond certain points further cuts lower tax collections and can undermine economic activity.

### Figure: Legislation of Transfer Pricing Documentation Requirements (Annex Figure 3.5.2)
- Depiction: number of countries per year (flow) and cumulative count (right scale) of jurisdictions enacting transfer pricing documentation requirements.
- Comparative categories: EMDEs (emerging market and developing economies) and AEs (advanced economies).
- Primary data source: International Tax Institutions Database; calculations: IMF staff calculations.
- Purpose: shows annual incidences and accumulated adoption over time, distinguishing contributions from EMDEs and AEs.

### Output–CIT Rate Curve (sensitivity analysis and mechanisms)
- Sensitivity to elasticity of TFP to public capital (δ):
  - Baseline calibration: δ = 0.05.
  - Range analyzed: δ = 0; δ = 0.02; δ = 0.05; δ = 0.1; δ = 0.15.
  - Findings:
    - Laffer curve (revenue–CIT rate) broadly invariant to δ.
    - Output–CIT rate relationship sensitive to δ:
      - Higher δ (> 0.05) steepens upward-sloping part of output–CIT rate curve, increasing likelihood that CIT rate cuts reduce output.
      - Lower δ (< 0.05) weakens link between revenue and productivity, making output gains from CIT rate cuts more common.
      - Extreme case δ = 0 implies CIT rate cuts unambiguously increase output.
  - Calibration note: output and revenue outcomes normalized to 100 in baseline.
- Public investment needs and composition of spending:
  - CIT rate cuts more likely to reduce output when:
    - public investment needs are large (δ large); or
    - government has high propensity to spend additional CIT revenues on public investment.
  - Mechanism: lower CIT revenues can reduce public capital and productivity, offsetting positive effects from lower tax distortions and intangible investment.
- Sensitivity to FDI elasticity with respect to CIT rate:
  - Baseline FDI elasticity = 1.08.
  - Scenarios: Baseline × 1.25; Baseline × 1.5.
  - Findings:
    - Higher FDI elasticity strengthens domestic output gains from tax cuts because MNCs more likely to reallocate or expand investment to lower-tax jurisdictions.
    - With higher FDI elasticity, output–CIT rate curve displays steeper decline in output as tax rates rise and flatter increase at lower tax rates.
    - Revenues are lower at higher CIT rates when FDI is more responsive to taxation.
- Tax competition in absence of public investment (δ = 0):
  - Effects of tax competition on revenues and shifted profits similar to baseline.
  - In absence of link between CIT revenues and public capital:
    - Effects of CIT cuts are positive but small for major AEs (0.01 percent).
    - Magnitudes relatively small and positive for EMEs.
    - Magnitudes negative and small for other AEs due to base-shifting effect.
  - Tax competition assumptions for illustrative scenario:
    - “AE tax competition”: major AE reduces CIT by 3 percentage points; “Other AEs” reduce by 1.2 percentage points.
    - “EM tax competition”: EMEs reduce CIT rates by 1.2 percentage points.
- Mechanisms emphasized:
  - Base-shifting and profit shifting: lower CIT rates encourage intangible investment and make economy more attractive for multinational production, reinforced by profit shifting.
  - Closed-economy benchmark: without trade, FDI, or profit shifting, intangible investment is fully deductible and positive CIT rate does not distort R&D; CIT revenue rises almost proportionally with the rate and real GDP largely unaffected.

*Source: World Economic Outlook, International Monetary Fund | October 2026.*

### 29.3 percent to 21.0 percent over 2010–17,  and fell a further

### ch3onlineannex - 29.3 percent to 21.0 percent over 2010–17,  and fell a further

### Decomposition of BETR changes for MNCs with Headquarters in the United States
- Periods analyzed: 2010–17 and 2018–23 (percentage point unless otherwise stated).
- BETR (start, percent): 29.3 (2010–17); 21.0 (2018–23).
- BETR (end, percent): 21.0 (2010–17); 18.9 (2018–23).
- ΔBETR (end-period value minus start-period value): –8.3 (2010–17); –2.0 (2018–23).

- Component contributions (percentage point):
  - Domestic Taxation: –3.3 (2010–17); –3.9 (2018–23)
    - Domestic Statutory: –0.1 (2010–17); –8.3 (2018–23)
    - Domestic Base: –3.2 (2010–17); 4.4 (2018–23)
  - Foreign Rates + Bases: –3.9 (2010–17); 2.5 (2018–23)
    - Foreign Statutory: –0.7 (2010–17); –0.0 (2018–23)
    - Foreign Base: –3.2 (2010–17); 2.5 (2018–23)
  - Profit Shifting: –1.7 (2010–17); –0.0 (2018–23)
  - Distribution: –1.5 (2010–17); –0.2 (2018–23)
  - Globalization: –0.1 (2010–17); 0.2 (2018–23)
  - Residual: 0.5 (2010–17); –0.6 (2018–23)

- Key interpretation points stated:
  - Pre-TCJA (2010–17) decline in BETR was mostly explained by foreign tax policy and profit shifting.
  - Post-TCJA (2018–23) the domestic statutory rate effect (reflecting the statutory rate cut from 35 to 21 percent) dwarfs every other component.
  - The domestic base effect turns positive post-TCJA, consistent with base-broadening provisions and profit repatriation under the new territorial regime.
  - The contribution of profit shifting is approximately zero in 2018–23.
- Data sources: U.S. Bureau of Economic Analysis (2009–2023) and IMF staff calculations.
- Note: BETR = backward-looking effective tax rate; ΔBETR = change in BETR (end-period value minus start-period value).

### Decomposition of BETR changes for MNCs with Headquarters in the European Union
- Two sample subperiods reported (percentage point unless otherwise stated): 2010–17 and 2018–21.
- BETR (start, percent): 34.9 (2010–17); 17.3 (2018–21).
- BETR (end, percent): 17.3 (2010–17); 17.1 (2018–21).
- ΔBETR: –17.4 (2010–17); –0.1 (2018–21).

- Component contributions (percentage point):
  - Domestic Taxation: –12.5 (2010–17); –0.4 (2018–21)
    - Domestic Statutory: –1.0 (2010–17); –0.7 (2018–21)
    - Domestic Base: –11.5 (2010–17); 0.3 (2018–21)
  - Foreign Rates + Bases: 1.4 (2010–17); 0.3 (2018–21)
    - Foreign Statutory: –0.5 (2010–17); –0.4 (2018–21)
    - Foreign Base: 1.9 (2010–17); 0.7 (2018–21)
  - Profit Shifting: −7.4 (2010–17); 0.1 (2018–21)
  - Distribution: –0.9 (2010–17); –0.1 (2018–21)
  - Globalization: –6.5 (2010–17); 0.2 (2018–21)
  - Residual: 1.0 (2010–17); –0.2 (2018–21)

- Memo and coverage:
  - Country reweighting: –0.3 (2010–17); 0.1 (2018–21)
  - Coverage (EU parent countries): 25.0 (2010–17); 26.0 (2018–21)
- Data sources: Orbis (2008–2021 vintages) and IMF staff calculations.
- Note: Percentage-point contributions from chained annual decomposition, PPP-GDP weights anchored at each annual end year. Croatia excluded due to data limitations. BETR = backward-looking effective tax rate; ΔBETR = change in BETR (end-period value minus start-period value); EU = European Union.

### Robustness and cross-checks
- Country-by-country reporting (CbCR) data used to cross-check BEA and Orbis decompositions; results are consistent with BEA and Orbis for the later period.
- US industry-level results and additional robustness results are available upon request (as reported by IMF staff).

### Local Projection Estimates of CIT and PIT Shock Spillovers (Online Annex 3.2)
- Objective: Examine cross-border spillovers of corporate income tax (CIT) and personal income tax (PIT) shocks originating in the United States, the United Kingdom, and Germany.
- Identification and instruments:
  - Narrative tax shocks used as instruments following Mertens and Ravn (2013) and other cited authors.
  - Structural tax shocks (z_CIT,s,t and z_PIT,s,t) recovered and used in a local projection instrumental variable (LP-IV) framework.
- Shock sample coverage (quarters):
  - Germany: 1992Q1–2018Q2
  - United Kingdom: 1993Q1–2007Q4
  - United States: 1951Q1–2019Q4
- Recipient panel: 29 AEs and EMEs (list provided in source).
- Methodology highlights:
  - Stacked dyadic LP-IV specification per Jordà (2005).
  - Endogenous regressor: source country’s average corporate and personal income tax rate (A_CITR_s,t or A_PITR_s,t), instrumented by corresponding structural shock.
  - Controls: four lags of source country A_PITR and A_CITR; four lags of recipient output growth; contemporaneous + four lags of the Global Financial Cycle index; robustness checks include external demand and Chicago Fed’s Adjusted National Financial Conditions Index.
  - Impulse response to a tax cut traced by {−β_h}_h=0.
  - Standard errors computed using Driscoll and Kraay (1998) estimator.
- Power and validity:
  - Instruments jointly significant in the first stage with F-statistic for excluded instruments well above 20 at each horizon.
  - Re-estimation controlling for contemporaneous tax shocks in other source countries reduces sample size and power; results for U.S. spillovers remain qualitatively similar but less precise.
- Empirical findings summarized:
  - Country-level estimates more precise for U.S. spillovers due to longer time series.
  - Spillovers from U.K. and German tax shocks are not statistically different from each other; spillovers from PIT shocks generally larger than from CIT shocks for those countries.
  - Figures show cumulative impact of a one percentage point decrease in average PIT or CIT in Germany, the U.K., or the U.S., instrumented by narrative shocks, on real GDP and real investment in the rest of the world (90 percent confidence intervals reported in source).

### GIMF Spillover Simulations (Online Annex 3.3)
- Model: IMF’s Global Integrated Monetary and Fiscal model (GIMF); 8-region setup (United States, euro area, Japan, Other Advanced Economies, China, Other Asia, Latin America, remaining economies).
- Key model features:
  - Profit-maximizing firms in monopolistic competition across non-tradable, tradable, and Global Value Chain (GVC) sectors; calibration to OECD Inter-Country Input-Output Tables (OECD 2021).
  - Frictions: sticky prices and wages, real adjustment costs, liquidity-constrained households, overlapping-generations (OLG) households.
  - Fiscal instruments: corporate and wage income taxes, government consumption, investment, transfers; governments can issue debt.
  - International capital flows: households can hold government debt and trade tradable U.S.-dollar-denominated obligations; capital mobile across regions via arbitrage in real returns.
  - Firms taxed on economic profits and capital rental income; depreciation and interest payments deductible.
- Transmission mechanisms emphasized:
  - CIT cuts raise after-tax marginal product of capital in implementing region, attracting savings and raising global real interest rates.
  - Higher global real interest rates increase user cost of capital elsewhere, weakening investment abroad — a macroeconomic analogue to FDI-relocation channel.
  - PIT cuts stimulate labor supply, boosting disposable income and consumption; they exert smaller upward pressure on global real interest rates relative to CIT cuts.
- Quantitative simulation results (as reported):
  - A permanent unilateral 5-percentage point CIT cut in a major advanced economy (debt-financed for 40 years) leads to:
    - Larger increases in interest rates abroad and larger declines in investment abroad than a 5-percentage point PIT cut.
    - Negative GDP spillovers to other regions: higher global interest rates and reallocation of investment to implementing country more than offset positive external demand.
  - A permanent unilateral 5-percentage point PIT cut produces:
    - Smaller rise in global interest rates and smaller fall in investment abroad.
    - Overall external effect broadly neutral, due to stronger labor supply response and smaller demand-capacity gap.
  - Deficit financing amplifies these dynamics: when CIT cut is not offset by lower spending or higher taxes elsewhere, public borrowing rises and higher world real interest rates induce higher private saving and lower investment at home and abroad.
- Figures in source illustrate:
  - Panel 1: Impact on interest rates and investment (Years 0–9) for PIT vs. CIT.
  - Panel 2: Impact on GDP (Years 0–5) for debt-financed CIT and PIT cut abroad.
- Interpretation: Macroeconomic channels in GIMF, even absent MNC-specific relocation, can generate opposite-sign spillovers for CIT versus PIT cuts through global interest rate and investment reallocation effects.

*Source: World Economic Outlook, International Monetary Fund | October 2026.*

### 1.  Transmission of Corporate Tax

### 1. Transmission of Corporate Tax

### Strategic tax competition: estimation approach and key coefficients
- Model: peer effects specification (equation (1)) estimating country i’s CIT response, τ_{i,t} = α + β1 τ_{i,t}^{peers} + β2 X_{i,t} + β3 Σ_w_{ij,t} X_{j,t} + φ_W + φ_{W,t} + θ_i + γ_t + u_{i,t}.
- Peers’ weighted average defined as τ_{i,t}^{peers} = Σ_w_{ij,t} τ_{j,t}.
- Baseline peer weights: geographical distance; alternative weights: similarity in receiving FDI inflows (gravity model).
- Identification: instruments exploit peers’ political-cycle changes (time since election, share of opposition), with controls for domestic political shocks.
- Interpretation: β1 > 0 implies strategic complements; β1 < 0 implies strategic substitutes.
- Online Annex Table 3.4.1 (Fixed-effects two-stage least squares, endogenous regressor: Peers Tax Rate) — selected estimates:
  - Statutory CIT (columns (1)-(2)): Peers Tax Rate = 0.404*** (SE 0.039) pooled; 0.431*** (SE 0.058) interacted.
  - EATR (columns (3)-(4)): Peers Tax Rate = 0.285*** (SE 0.137) pooled; 0.297*** (SE 0.115) interacted.
  - EMTR (columns (5)-(6)): Peers Tax Rate = 0.312** (SE 0.155) pooled; 0.345*** (SE 0.132) interacted.
  - Peers Tax Rate × (year >= 2017): -0.169** (SE 0.08) in one specification; other specifications show -0.026 (SE 0.059) and -0.040 (SE 0.07).
  - Post-2016 Response (Peers Tax Rate + Peers Tax Rate × (year >= 2017)): 0.262*** (SE 0.06), 0.270** (SE 0.122), 0.305** (SE 0.142).
  - Memo: Pre-2017 Subsample estimates: 0.431***, 0.431***, 0.358***, 0.358***, 0.337**, 0.337** (various SEs reported).
- Estimation diagnostics:
  - Observations: 3,875; 3,870; 2,618 (depending on specification).
  - Countries: 152; 152; 151.
  - Cragg-Donald Wald F: 80.4; 34.4; 39.4; 29.6; 31.9; 23.5 (by column).
  - First-stage F (Sanderson-Windmeijer) reported similarly (e.g., 80.4, 34.8, 39.4).
- Controls: lagged dependent variable, own-country political variables and interactions, log GDP, log population, trade openness, spatial lags of peer log GDP, log population and trade openness; country and year fixed effects.

### Spillovers from base shifting: bilateral FDI gravity and sectoral estimation
- Gravity specification for bilateral cumulative FDI (as percent of destination GDP at time t−1) (equation (2)):
  - Y_{ij,t+h} = η (τ_{i,t} − τ_{j,t}) + ζ1 X_{i,t} + ζ2 X_{j,t} + ζ3 X_{ij,t} + ε_{ij,t}.
  - η = semi-elasticity of FDI to tax differentials.
  - Tax differential instrumented with peers’ political cycle.
  - Bilateral FDI source: World Bank Harmonized Bilateral FDI Database (Steenbergen and others 2022).
  - Controls include population, GDP, bilateral trade, contextual effects, and lagged bilateral inflows.
- Sector-level specification (equation (3)) for outcomes Y_{ij,s,x(s),t+h}:
  - Y_{ij,s,x(s),t+h} = η^{x(s)} (τ_{i,s,t} − τ_{j,s,t}) + θ X_{ij,t} + δ_s + ε_{ij,s,t}.
  - Heterogeneity across sectoral intangibles intensity x(s) = {tangible, intangible}; intangible sectors include Information, Legal Services, Management of Companies and Enterprises.
  - Outcomes: (a) count of brownfield and greenfield projects (extensive margin); (b) level of investment as percent of GDP (intensive margin).
  - Intensive margin uses control function approach; extensive margin uses 2SLS.

### Profit-shifting estimation: data, methodology, and main results
- Data:
  - Orbis unconsolidated balance sheet panel, 2009–23, real (2019) USD.
  - Unit of observation: subsidiary f in country j, sector k, year t, linked to global ultimate owner (GUO) in country i.
  - Sample: about 500,000 MNE groups and over 3,000,000 subsidiary-year observations, covering over 100 host countries and headquarters in over 40 jurisdictions.
  - MREID dataset used for sectoral MNC activities across 185 countries and 25 industries (for other FDI analysis).
- Outcome and main independent variable:
  - Dependent variable: return on assets (ROA) = EBIT / assets; reported taxable income treated as censored at zero.
  - Tax differential at affiliate jurisdiction j: τ̅_{j,t} = τ_{j,t} − Σ_{l≠j} ω_{l,t} τ_{l,t}, where ω are asset weights across jurisdictions where the MNC has affiliates.
- Main specification (equation (1) for ROA):
  - ROA_{G,i,f,j,k,t} = β0 + β1 τ̅_{j,t} + γ′ X_{f,j,t} + δ′ X_{G,i,t} + ϵ′ X_{j,k,t} + φ_G + φ_k + φ_{j,t} + u_{...}.
  - Controls: affiliate-level (log turnover, log employment), industry-jurisdiction-level (turnover-based Herfindahl), conglomerate-level (MNC profitability), fixed effects: year, conglomerate, subsidiary-jurisdiction-year.
- Elasticity interpretation:
  - Estimated elasticity of reported income to tax differentials for an MNC with average ROA: −1.2 (Online Annex Table 3.5.1, column 1).
- Heterogeneity by intangibles intensiveness and time (equation (2)):
  - Intangibles intensiveness at GUO level measured as ratio of intangible fixed assets to total fixed assets.
  - High-intangibles-intensive (I_G = 1) if GUO ratio ≥ 50 percent (75th percentile).
  - Post-2017 dummy D_{2017} = 1 for year ≥ 2017.
- Main estimates (Online Annex Table 3.5.1, baseline regressions, dependent variable: censored ROA; sample observations = 3,198,284; R² = 0.65):
  - τ − Σωτ coefficient: -0.327*** (SE 0.0894) in column (1); -0.190* (SE 0.101) in column (2); -0.239* (SE 0.126) in column (3).
  - (τ − Σωτ) × high-intangibles: -0.280*** (SE 0.0780) in column (2); -0.546*** (SE 0.126) in column (3).
  - (τ − Σωτ) × post 2017: 0.251** (SE 0.117) in column (2).
  - (τ − Σωτ) × post 2017 × high-intangibles: -0.368** (SE 0.143) in column (2).
  - (τ − Σωτ) × TP doc. HQ: 0.242** (SE 0.104) in column (3).
  - (τ − Σωτ) × TP doc. HQ × high-intangibles: 0.243* (SE 0.104) in column (3); reported -0.126 elsewhere.
  - TP doc. HQ (non-interacted): -1.662** (column (2)) and -0.705 (column (3)).
- Implied semielasticities reported in table (for an MNC with average ROA):
  - Overall profits semielasticty for average affiliate: Low II = -1.168*** (SE 0.319) in column (1); -0.679* (SE 0.361) in column (2); -2.148 (SE 0.353) in column (3) — table reports multiple reported variants.
  - Low II post-2017 / TP doc: 0.219 (SE 0.483) and -0.488 (SE 0.332) in alternative rows.
  - High II: -1.678*** (SE 0.348).
  - High II post-2017 / TP doc: -2.095*** (SE 0.426).
- Interpretation of heterogeneous effects:
  - Low-intangibles MNCs: semielasticity before 2017 ≈ −0.68; after 2017 semielasticity ≈ 0.2 (not statistically different from zero) — indicating marked decline in profit-shifting responsiveness after 2017.
  - High-intangibles MNCs: semielasticity before 2017 ≈ −1.7; after 2017 ≈ −2.1 — indicating persistent and larger sensitivity to tax differentials for intangibles-intensive firms.
  - Transfer-pricing documentation requirements (TPDRs): for low-intangibles MNCs, semielasticity is −0.85 without TPDRs and statistically indistinguishable from zero with TPDRs; for high-intangibles MNCs, semielasticity is −2.8 without TPDRs and falls (statistically significant difference of 1.7) with TPDRs.
  - Overall implication: profit-shifting responsiveness declined for low-intangibles firms after 2017 and with TPDRs, but high-intangibles-intensive MNEs remain more responsive to tax differentials, with continued profit-shifting despite anti-avoidance efforts and the expansion of profit-based tax incentives for intangible-derived income after 2017.

### Quantitative summaries and diagnostics
- Sample sizes and coverage:
  - Strategic competition: up to 3,875 observations, 152 countries.
  - Profit-shifting: 3,198,284 subsidiary-year observations; ~500,000 MNE groups; over 100 host countries; headquarters in over 40 jurisdictions.
- Key elasticities/semielasticities preserved exactly:
  - Estimated elasticity of reported income to tax differentials: −1.2 (average MNC).
  - Semielasticities: low-intangibles pre-2017 = −0.68; low-intangibles post-2017 = 0.2; high-intangibles pre-2017 ≈ −1.7; high-intangibles post-2017 ≈ −2.1.
  - High-intangibles without TPDRs semielasticity = −2.8; TPDR introduction reduces semielasticity (difference of 1.7).
- Statistical significance notation preserved: ***, **, * denote p<0.01, p<0.05, p<0.1 respectively.

### Policy-relevant interpretations and implications (from the analysis)
- Strategic tax competition:
  - Positive and sizable peer effects (e.g., Peers Tax Rate coefficients around 0.285–0.431 in various specifications) indicate governments’ CIT choices are responsive to peers’ CIT levels, implying cross-border interdependence in CIT setting.
  - Post-2016 dynamics: interaction with year >= 2017 shows changes in the strength of peer effects (e.g., Peers Tax Rate × (year >= 2017) coefficients include -0.169** and smaller estimates), with reported Post-2016 Response remaining positive in pooled specifications.
- Profit shifting and intangibles:
  - High-intangibles-intensive MNCs exhibit larger reported-income sensitivity to jurisdictional tax differentials than low-intangibles MNCs, implying intangibles amplify profit-shifting incentives.
  - Anti-avoidance measures such as TPDRs appear to reduce sensitivity for low-intangibles MNCs and to have a mitigating effect for high-intangibles MNCs, but high-intangibles responsiveness remains substantial even after 2017 and after TPDR introduction.
  - The persistence of large semielasticities for intangibles-intensive firms suggests that international tax reforms and documentation requirements have not fully eliminated profit-shifting channels associated with intangible assets.
- Implication for policy design:
  - Coordination and multilateral measures targeting income allocation for intangibles may be more effective than unilateral statutory rate adjustments alone, given strategic tax competition and base-shifting spillovers.
  - Strengthening documentation, transfer-pricing enforcement, and designing rules addressing the allocation of intangible-derived income could meaningfully reduce profit shifting, especially for high-intangibles MNCs, though substantial responsiveness may remain.

*Source: International Monetary Fund, World Economic Outlook, October 2026 (Online Annexes 3.4–3.5).*

### Annex Figure 3.5.2.

### Annex Figure 3.5.2.

### Figure summary
- Title: Legislation of Transfer Pricing Documentation Requirements
- Depiction: Number of countries per year (flow) and cumulative count (right scale) of jurisdictions enacting transfer pricing documentation requirements.
- Comparative categories: EMDEs = emerging market and developing economies; AEs = advanced economies.

### Data and sources
- Primary data source: International Tax Institutions Database (https://www.rsit-uni-tuebingen.de/data).
- Calculations: IMF staff calculations.

### Purpose and interpretation
- The figure shows both annual incidences (number of countries per year) and the accumulated adoption over time (cumulative series on the right-hand scale) for transfer pricing documentation legislation.
- The presentation distinguishes contributions from EMDEs and AEs to the annual and cumulative totals.

*Source: International Monetary Fund, Chapter 3, “Intangible Yet Real: Spillovers from Corporate Income Taxation,” October 2026; International Tax Institutions Database; IMF staff calculations.*

### 0. IP box regimes, by contrast, reduce the effective tax

### 0. IP box regimes, by contrast, reduce the effective tax

### Intangibles-intensity interaction and regulatory indicators
- The intangibles-intensity interaction tests whether MNCs with a large stock of intangible assets respond differently to tax differentials.  
- Intangible assets are nonrival within the MNC and facilitate profit-shifting through transfer pricing and strategic location of IP; historically, intangibles-intensive MNCs may have been less sensitive to EMTRs.  
- As anti-avoidance regulations curtail these channels, the sign of the interaction becomes an empirical question.  
- All regulatory indicators are lagged one year and entered separately.

### Microeconometric results: effect of EMTR differentials on investment rate
- Baseline specification (column 1): a 1 percent increase in the tax differential reduces the investment rate by 0.148.  
- Differential and weighted differential specifications show consistently negative coefficients across specifications.  
- The differential effect post 2017 is negative but estimated with more uncertainty.  
- Results are robust to alternative specifications, including controlling for destination-time fixed effects and adding GUO-time fixed effects.  
- Estimation notes: standard errors clustered at GUO level. Sample: affiliates of MNCs over the 2010–20 period, with headquarters in over 40 countries and subsidiaries in over 100 countries. Dependent variable: investment rate. Main explanatory variable: tax differential in sector-specific EMTR in the destination country relative to the asset-weighted group average EMTR across all subsidiaries of the MNC group.

Key reported coefficients from Online Annex Table 3.7.1:
- Effective Marginal Tax Rate Differentials: −0.148**, −0.117*, −0.223**, −0.163** (standard errors: (0.060), (0.062), (0.097), (0.069)).  
- Effective Marginal Tax Rate Differentials x Post 2017: −0.134**, −0.385***, −0.056 (standard errors: (0.068), (0.088), (0.090)).  
- Observations: 6,235,788; 6,235,788; 6,235,782; 4,422,576 (columns 1–4 respectively).  
- R²: 0.13 across reported specifications.  
- Significance codes: *** p<0.01, ** p<0.05, * p<0.1.

### Interactions with origin-country anti-avoidance regulations and intangible intensity
- MNCs headquartered in countries with stricter earnings stripping rules are more sensitive to destination taxes: interaction coefficient −0.511*** (standard error (0.073)).  
- Stricter transfer pricing documentation requirements increase sensitivity: interaction coefficient −0.607*** (standard error (0.072)).  
- Controlled foreign corporation (CFC) rules show no significant differential effect: interaction coefficient −0.005 (standard error (0.079)).  
- Firms from countries with IP box regimes are less tax-sensitive, suggesting these regimes reduce the effective tax burden and diminish the incentive to relocate investment: interaction coefficient 0.242*** (standard error (0.066)).  
- Firms more intensive in intangible investment do not exhibit significant differences in their investment margin: interaction coefficient −0.141 (standard error (0.090)).  
- Observations for interaction regressions: 6,340,678 (columns 1–4), 6,222,510 (column 5).

### Structural model of multinational production, profit shifting, and tax competition (model overview)
- Model type: multi-region spatial general equilibrium extending Dyrda, Hong, and Steinberg (2024a, 2024b) to a seven-region world and incorporating productive public capital.  
- Regions: (1) a major advanced economy (AE); (2) other AEs; (3) a major EME; (4) G20 EMEs; (5) productive low-tax jurisdictions (PLTJs); (6) the rest of the world (ROW: other productive EMDEs); plus unproductive low-tax jurisdictions (ULTJs) where no production occurs but shell affiliates can be established.  
- Region heterogeneity: population, TFP, trade costs, FDI costs, CIT rates, efficiency of tax enforcement.  
- Firm heterogeneity: idiosyncratic productivity drawn from a Pareto distribution; firms decide export destinations, foreign affiliate locations, intangible capital production, and share of profits shifted to LTJs; exporting and FDI decisions subject to fixed and variable costs; largest MNCs shift profits via transfer (mis)pricing of intangible assets (small share of firm population).  
- Production: firm output depends on aggregate TFP in production location, firm idiosyncratic productivity, stock of intangible capital, and local labor; intermediate goods aggregated through a CES demand structure with common elasticity of substitution.  
- Intangible capital: produced via domestic R&D labor; non-rival use across affiliates → increasing returns to R&D; MNCs undertake vast majority of intangible investment in equilibrium.  
- Profit shifting: firms can locate ownership of intangible assets in PLTJs and ULTJs, transfer licensing rights after paying fixed costs, and choose share of licensing income earned by LTJ affiliates; profit shifting subject to a convex cost capturing real resources devoted to tax planning.  
- Productive public capital: effective TFP influenced by public capital stock; accumulation financed by a share of CIT revenue (remainder from other tax sources fixed in counterfactuals).

### Defining low-tax jurisdictions (classification thresholds)
- Four thresholds used: (1) CIT abuse index (Tax Justice Network); (2) statutory CIT rate (including special regimes for MNCs); (3) FDI stocks (average of assets and liabilities) as a share of GDP; (4) population size.  
- ULTJ classification criteria: CIT abuse indicator above 60; FDI stocks larger than the size of the economy; statutory CIT rate lower than 10 percent; population size less than 2 million. Economies satisfying three of these criteria—including a low CIT rate and low population size—are also classified as ULTJs.  
- PLTJ classification: economies that satisfy two criteria, or three criteria including the small population size or low CIT rate criterion.

### Transmission channels of CIT rate changes
- Direct effects on the tax-cutting economy:
  - Lower CIT rates stimulate intangible investment and innovation by reducing user cost of capital.  
  - Lower CIT rates narrow tax differentials, reducing incentive to shift profits and freeing resources for production and R&D.  
  - Lower CIT rates reduce CIT revenue and public capital investment, lowering resources for productive public capital and impacting TFP.  
  - Net direct effect depends on fiscal productivity (output elasticity of public capital) and share of public investment financed by CIT revenue; higher calibrated values in EMDEs imply the fiscal cost of tax competition is disproportionately borne by EMEs.
- Spillover effects to other regions:
  - Base-shifting channel: when one economy lowers CIT rates, it becomes relatively more attractive for MNC investment and profit booking, eroding the corporate tax base of non-tax-cutting economies.  
  - Intangible capital channel: lower tax rate in a tax-cutting country reduces user cost of capital, incentivizing additional R&D and larger accumulation of intangible capital across MNC affiliates, raising productivity of foreign affiliates.

### Calibration, model-implied elasticities, and alignment with empirical estimates
- Calibration: parameters common across baseline and counterfactuals; region-specific calibration to match production, trade, FDI, and profit shifting. Key assigned parameters include: Elasticity of Substitution Between Varieties (ρ) = 4; Labor Share in Production (γ) = 0.85; Intangible Capital Share in Production (ψ) = 0.1; TFP Elasticity to Public Capital (δ) = 0.05; Share of Public Capital Investment in Government Revenue = 0.09–0.47; Pareto Shape Parameter (η) = 2.9–3.1. Region-specific corporate tax rates, population, and other moments taken from data sources cited in the model calibration.  
- Model-implied semi-elasticities (Online Annex Table 3.8.2):
  - Extensive margin (location choice): microeconometric estimate −1.4; model-implied −1.6.  
  - Intensive margin (affiliate scale): microeconometric estimate −0.2; model-implied −0.3 to −0.8.  
  - Aggregate FDI: micro estimate −2.0; model-implied −2.0 to −2.4.  
  - Profit shifting (reported profits): microeconometric estimate −1.2 to −2.0; model-implied −0.9.  
- The model’s profit-shifting semi-elasticity of approximately −0.9 (reported pre-tax profits with respect to the tax differential) is within the literature range (−0.8 to −1.1) and close to the chapter’s microeconometric estimate of −1.2.

### Counterfactual policy experiments and sensitivity
- Counterfactual 1 (reciprocated CIT cuts): major AE cuts statutory CIT rate by 3 percentage points; other AEs cut by 1.2 percentage points in response; all EMEs assumed to cut by 1.2 percentage points.  
- Counterfactual 2 (anti-avoidance enforcement instead of EME cuts): AEs implement the same CIT cuts as in scenario 1; all EMEs strengthen anti-avoidance regulation instead of matching tax cuts. Enforcement improvements calibrated to target reductions in profit shifting between 10 and 20 percent, comparable to enforcement gains achieved by AEs over the past decade.  
- Robustness/sensitivity: varying the CIT rate of a large economy while holding others constant generates a standard Laffer curve for revenue and an inverted U-shape for domestic output—reductions in CIT can boost revenues and output when initial rates are very high, but beyond certain points further cuts lower tax collections and can undermine economic activity.

*Source: International Monetary Fund (World Economic Outlook, October 2026); IMF staff calculations and model described in Online Annexes 3.7 and 3.8.*

### 1. Output–CIT Rate Curve

### Output–CIT Rate Curve

### Sensitivity to the elasticity of TFP to public capital (δ)
- Baseline calibration uses an elasticity of δ = 0.05 (based on Bom and Ligthart (2014)).
- Range of elasticities analyzed: δ = 0; δ = 0.02; δ = 0.05; δ = 0.1; δ = 0.15.
- Findings:
  - The Laffer curve (revenue–CIT rate relationship) is broadly invariant to different δ values.
  - The output–CIT rate relationship is sensitive to δ:
    - Higher δ (> 0.05) makes the upward-sloping part of the output–CIT rate curve steeper, increasing the likelihood that CIT rate cuts reduce output (green and light blue cases).
    - Lower δ (< 0.05) weakens the link between revenue and productivity, making output gains from CIT rate cuts more common.
    - The extreme case δ = 0 (equivalent to a model without public capital) implies CIT rate cuts unambiguously increase output (red).
- Calibration note: Output and revenue outcomes in the baseline calibration are normalized to 100.

### Public investment needs and composition of spending
- CIT rate cuts are more likely to be linked to reductions in output when:
  - (i) public investment needs are large, such that the elasticity of TFP to public capital (δ) is large; or
  - (ii) the government has a high propensity to spend additional CIT revenues on public investment rather than current spending (modelled as household transfers).
- Mechanism: Lower CIT revenues can reduce public capital and productivity, offsetting any positive effects of lower tax distortions and base-shifting-induced intangible investment.

### Sensitivity to the elasticity of FDI with respect to the CIT rate
- Baseline FDI elasticity = 1.08.
- Scenarios considered: Baseline FDI elasticity; Baseline × 1.25; Baseline × 1.5.
- Findings:
  - Higher FDI elasticity strengthens domestic output gains from tax cuts because MNCs are more likely to reallocate or expand investment to lower-tax jurisdictions.
  - With higher FDI elasticity, the output–CIT rate curve displays a steeper decline in output as tax rates rise and a flatter increase at lower tax rates.
  - Revenues are lower at higher CIT rates when FDI is more responsive to taxation, because the adverse effects of a tax increase on the CIT base are larger.

### Tax competition in the absence of public investment (counterfactual)
- Counterfactual abstracts from public capital (δ = 0) to illustrate effects when governments do not link CIT revenues to public capital and productivity.
- Findings:
  - Effects of tax competition on revenues and shifted profits are similar to the baseline calibration.
  - In the absence of a link between CIT revenues, public capital, and productivity:
    - Effects of CIT cuts are positive but small for the major AEs (0.01 percent).
    - Magnitudes are relatively small and positive for EMEs.
    - Magnitudes are negative and small for other AEs due to the base-shifting effect.
- Tax competition assumptions for illustrative scenario:
  - “AE tax competition” assumes a reduction in a major AE’s corporate income tax rate of 3 percentage points and a reduction in the corporate income tax rate of “Other AEs” of 1.2 percentage points.
  - “EM tax competition” assumes a reduction in the corporate income tax rates of emerging market economies by 1.2 percentage points.

### Mechanisms emphasized
- Base-shifting and profit shifting:
  - Lower CIT rates encourage intangible investment and make the economy a more attractive destination for multinational production, reinforced by profit shifting.
- Closed-economy benchmark:
  - In a closed economy without trade, FDI, or profit shifting, intangible investment is fully deductible and a positive CIT rate does not distort R&D decisions; CIT revenue rises almost proportionally with the rate, and real GDP is largely unaffected.

---


_Source: https://www.imf.org/-/media/files/publications/weo/2026/october/english/ch3onlineannex.pdf_