## _wp15118 - References

## Source details

**Canonical URL:** [_wp15118 - References](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15118.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15118.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15118.pdf.json)

---

### Major themes and objectives
- Provide the first systematic empirical evidence on whether BEPS and tax competition matter for developing countries, and quantify possible effects.
- Two distinct cross-border fiscal externalities analyzed:
  - Base spillovers: impact of one country’s tax policy on the tax bases of other countries (via real activities and/or profit shifting).
  - Strategic rate spillovers: impact on a country’s policy choices of tax changes abroad (tax competition).
- Data exploited: aggregate data on corporate tax bases for a panel of 120 countries (and tax rates for 173 countries), over the period 1980–2013.
- Emphasis: concentrate primarily on base spillovers because of direct relevance to BEPS and to enable an estimate—acknowledged as very speculative—of revenue impact of international corporate tax avoidance involving tax ‘havens’.

### Theoretical framework (key constructs and implications)
- World of n countries; country i has population share ℎi; world population normalized to unity (∑ℎj = 1).
- Multinational with one affiliate per country; real activity in i described by fi(ki) with fi′>0 and fi′′<0.
- Base in country i: bi = hi ki + ∑sij (equation (1)).
- Multinational’s after-tax profit Π given by expression (2), with tax rates ti and base-shifting costs cij(sij).
- First-order condition for real capital allocation: fi′(ki) = ρ + ti (equation (3)); ki implicitly a function ki(t1,..,tn).
- For quadratic production functions, tax changes imply:
  - ∂ki/∂ti = −(1 − hi) < 0 (equation (4)).
  - ∂ki/∂tj = hj > 0 (equation (5)).
  - Size (hi) matters critically for real-investment channel: effects from j on i larger the larger is country j.
- Base-shifting first-order condition (for sij): cij′(sij) + cji′(sij) = tj − ti when tj > ti (equation (6)); with quadratic costs cij(sij) = (1/2) Δij sij^2 this yields responsiveness ∂sij/∂tj = 1/δ(i,j) > 0 (equation (7)), where δ(i,j) ≡ Δij + Δji.
- Profit-shifting channel is independent of country sizes and depends on ease of shifting (δ(i,j)); even small jurisdictions can induce shifting from large ones.
- Combined small change impact on per capita base:
  - dbi/hi = βi (dti − ∑ωij dtj) (equation (8)), where βi and ωij defined in (9) and (10).
  - If effects operate only through real investment, weights become ωij = hj/(1 − hi) (equation (11)) — “GDP-weighted”.
  - If effects operate only through profit shifting, weights become ωij = (1/δ(i,j)) / ∑p≠i (1/δ(i,p)) (equation (12)) — “haven/shiftability-weighted”.

### Empirical specification and identification
- Base spillover dynamic specification:
  - bit = λ bi,t−1 + φ τit + γ W−i τ−i,t + ζ′ Xit + αi + μt + εit (equation (13)),
    - bit: CIT base for country i at time t; τit: domestic CIT rate; W−i τ−i,t: weighted average of foreign statutory CIT rates (∑ωij τjt, ∑ωij = 1); Xit: controls; αi, μt: fixed effects.
  - Short-run own-rate effect φ expected negative; long-run impact θ(φ) ≡ φ/(1 − λ).
  - Spillover coefficient γ expected positive; long-run θ(γ) ≡ γ/(1 − λ) (equation (14)).
- Weighting matrices to distinguish channels:
  - GDP-weighted rates: tax rate in j weighted by j’s GDP as share of total GDP of all countries other than i (captures real investment channel).
  - Haven-weighted rates: unweighted average of rates in jurisdictions on a commonly-used list of ‘tax havens’ (captures profit-shifting channel).
  - Inverse-distance weighting considered to capture geographic proximity effects.
- Identification issues and solutions:
  - Haven-weight variable does not vary across non-haven countries and is collinear with time effects—resolved by imposing a common linear time trend or by imposing φ = −γ (testable restriction derived from theory).
  - Use of statutory rates versus Average Effective Tax Rates (AETR): AETRs more relevant for investment but data limited; AETRs pursued where available.

### Empirical strategy for strategic rate spillovers
- Rate-setting (strategic) regressions:
  - τit = b W−i τ−i,t + ζ′ Xit + ai + ct + εit (equation (15)),
    - same three weighted-average constructs used (GDP-weighted, haven-weighted, inverse-distance).
  - Specification includes country effects and a common time trend; lagged dependent variable omitted.

### Contextual and literature-based quantitative benchmarks preserved from source
- Data scope: panel of 120 countries for tax bases; tax rates for 173 countries; period 1980–2013.
- Prior findings cited:
  - De Mooij and Ederveen (2008): a 10 percentage point reduction in a country’s average effective tax rate increases its stock of FDI, on average and in the long run, by over 30 percent.
  - Heckemeyer and Overesch (2013): consensus semi-elasticity of −0.8, implying that a 10 percentage point higher tax rate will reduce reported profit in an affiliate by 8 percent.
  - Devereux et al. (2008): among OECD countries, a 10 percentage point decrease in statutory CIT rates in other countries generates, on average, a cut of 7 percentage points in response.
  - Klemm and Van Parys (2012): among Sub-Saharan African and Caribbean countries, estimated strategic interactions of 2.5 to 3 points in response to a 10 point tax cut abroad.
- Figure 1 source and note retained: Source: IMF Staff estimates; data from IMF’s Fiscal Affairs Department database. Note: Total tax revenue including grants and excluding social contributions; resource-rich countries excluded.

### Key methodological caveats highlighted in the source
- Difficulty disentangling real investment and profit-shifting channels precisely; many identification challenges due to data limitations.
- AETR data limitations: not available for as many countries or as long a period as statutory rates; calculated AETRs depend on tax base elements producing endogeneity concerns.
- Measured GDP may be affected by profit shifting, making GDP an imperfect size indicator for weighting.
- The approach for revenue impact of BEPS and havens is acknowledged as “very speculative”.

### Appendix 2 — Identification and estimation issues
- Equations (13) and (15) estimated by system generalized method of moments (GMM), using only internal instruments.
- Endogeneity concerns:
  - Shocks affecting domestic tax base may also affect contemporaneous tax rate choice.
  - Estimation of the CIT base by dividing revenues by the main statutory rate can give rise to measurement error when multiple CIT rates apply.
  - Tax rates are jointly determined across countries in the strategic spillover equation.
- Alternative W−i as simple average raises identification challenges; country-specific trends yield very similar results to common trend.

### Appendix 2 — Tax-haven effects and data limitations
- Attractions of tax ‘havens’ derive from special regimes and arrangements for which descriptive data are unavailable; identification of haven effects depends on a plausible but untestable correlation between movements in statutory rates and special regimes.
- Over the full sample period, the average CIT rate in the ‘havens’ is around 17 percent, compared to 32 percent for the full sample.
- Many havens are small; regressing CIT rate on country size and a haven dummy shows tax-havens actually have, on average, a significantly higher CIT rate than otherwise similar countries.

### Data, sample, and construction
- Sample: unbalanced panel comprising 173 countries over 1980–2013.
- Country listing and haven classification in Appendix 1.
- Data on CIT revenues and statutory tax rates from IMF’s Fiscal Affairs Department database; country coverage of CIT rates is full though unbalanced in time.
- Linear interpolation used only for constructing balanced panels of weighted average tax rates; own tax rates in regressions are actual values, not interpolations.
- Resource-rich countries excluded from dependent-variable treatment of tax bases; their tax rates included in weighted averages.
- CIT base in percent of GDP, bi, calculated by dividing CIT revenue by the standard CIT rate; lack of revenue data means this is possible for only 121 countries.
- AETR data are far more limited.

### Appendix 2 (for 43 countries) — Data and descriptive statistics
- Sample composition: 43 countries enumerated in the source text.
- Controls: (the log of) GDP per capita, share of agriculture in value-added, trade openness (sum of non-resource exports plus imports, relative to GDP), and inflation.
- Data sources: Agricultural value-added (WDI); Trade openness (IFS); GDP per capita constant (2000) U.S. dollars (WDI); Inflation (IFS).
- Key descriptive statistics (Table 1):
  - Statutory CIT Rate, in percent: Obs. 3037; Mean 32.15; Max. 61.80; Min. 0.00; Std. Dev. 10.85
  - GDP-weighted average tax rate, in percent: Obs. 3037; Mean 39.18; Max. 48.04; Min. 26.98; Std. Dev. 5.28
  - Haven-weighted average CIT rate, in percent: Obs. 3037; Mean 17.09; Max. 24.46; Min. 11.08; Std. Dev. 3.55
  - Inverse-distance-weighted average CIT rate, in percent: Obs. 4771; Mean 32.08; Max. 42.18; Min. 18.60; Std. Dev. 4.60
  - CIT revenue, percent of GDP: Obs. 2161; Mean 2.64; Max. 13.37; Min. 0.00; Std. Dev. 1.53
    - OECD countries: Obs. 913; Mean 2.76; Max. 8.02; Min. 0.26; Std. Dev. 1.28
    - Non-OECD countries: Obs. 2354; Mean 2.42; Max. 18.40; Min. 0.01; Std. Dev. 1.98
  - CIT base, percent of GDP: Obs. 2161; Mean 8.59; Max. 29.99; Min. 0.00; Std. Dev. 5.45
    - OECD countries: Obs. 893; Mean 8.75; Max. 29.99; Min. 1.06; Std. Dev. 4.61
    - Non-OECD countries: Obs. 1268; Mean 8.47; Max. 29.73; Min. 0.00; Std. Dev. 5.97
  - AETR, in percent: Obs. 508; Mean 22.23; Max. 40.27; Min. -11.61; Std. Dev. 9.24
  - GDP per capita, 2000 USD: Obs. 1970; Mean 13349; Max. 87716; Min. 126; Std. Dev. 15353
  - Trade openness, percent of GDP: Obs. 1974; Mean 79.04; Max. 436.95; Min. 6.32; Std. Dev. 45.66
  - Inflation, in percent: Obs. 1925; Mean 36.46; Max. 11749.64; Min. -4.47; Std. Dev. 368.39
- Stylized trend: pronounced decline in mean statutory CIT rates by 15 to 20 percentage points over the last three decades for both OECD and non-OECD groups; haven-weighted averages interpolated are substantially lower.

### Base spillovers (estimation of equation (13)) — main empirical findings
- Estimation approach: System GMM (one step, robust) with alternative weighting matrices: GDP-weighted, haven-weighted, inverse-distance-weighted. Dependent variable: CIT base.
- Own-rate short-run effects:
  - Column (1) GDP-weighted: short-run marginal coefficient on own CIT rate = -0.0818*** (standard error 0.0396) → interpreted as a one percentage point increase in a country’s CIT rate reduces its CIT base by 0.08 percent of GDP.
  - Short-run semi-elasticity evaluated at mean CIT base of 8.59 percent of GDP: calculated short-run semi-elasticity of -0.9 (calculated as (0.08/8.79)×100 in source).
- Own-rate long-run effects:
  - θ(φ) in column (1): -1.1608 (standard error 1.0649) — large but imprecise; null of no long run effect not rejected in column (1).
  - Haven-weighted long-run effect significant at 5 percent in column (2); inverse-distance long-run effect insignificant in column (3).
- Cross-country (base) spillovers:
  - GDP-weighted (column 1): CIT rate j, weighted GDP = 0.1763* (standard error 0.0982). Interpretation: a one percentage point reduction abroad reduces the typical country’s CIT base in short run by 0.18 percent of GDP.
  - Haven-weighted (column 2): CIT rate j, weighted tax havens = 0.3544** (standard error 0.1698).
  - Inverse-distance (column 3): CIT rate j, weighted inverse-distance = 0.3317* (standard error 0.1998).
- Equality restriction test (γ = −φ):
  - Null that φ = −γ not rejected for any weight structure (p values: column (1) 0.329; column (2) 0.120; column (3) 0.204).
  - Imposing the restriction yields restricted coefficients around 0.0885** (column 1), 0.0986*** (column 2), 0.0832** (column 3).
- Diagnostics (Table 2 summary):
  - M1 (p value): 0.001, 0.001, 0.000 for columns (1)-(3).
  - M2 (p value): 0.780, 0.859, 0.789.
  - Hansen (p value): 0.452, 0.523, 0.385.
  - Observations: 1540, 1694, 1687; Number of countries: 100, 105, 103.

### Heterogeneity by income/OECD membership (Table 3 and Table 4)
- All (haven-weighted, column 1): CIT rate i = -0.0991*** (0.0413); CIT rate j, haven weighted = 0.3544** (0.1698).
- OECD: CIT rate i = -0.0596* (0.0371); CIT rate j, haven weighted = 0.3423** (0.1767).
- Non-OECD: CIT rate i = -0.1376** (0.0657); CIT rate j, haven weighted = 0.4421* (0.2718).
- Long-run θ(φ): All -0.8165** (0.4739); OECD -0.2571 (0.1695); Non-OECD -0.9398* (0.5169).
- Imposed equality (restricted coefficient): All 0.0986*** (0.0413); OECD 0.0527 (0.0416) [insignificant]; Non-OECD 0.1482** (0.0650).
- Non-OECD detailed results:
  - Own-rate coefficients across weights: (1) GDP-weighted CIT rate i = -0.1206* (0.0727); (2) haven-weighted CIT rate i = -0.1376** (0.0657); (3) inverse-distance CIT rate i = -0.1126* (0.0624).
  - Spillovers in non-OECD:
    - GDP-weighted: CIT rate j, weighted GDP = 0.1324 (0.1410) — insignificant.
    - Haven-weighted: CIT rate j, weighted tax havens = 0.4421* (0.2718) — significant at 10 percent.
    - Inverse-distance: CIT rate j, weighted inverse distance = 0.0525 (0.3296) — insignificant.
  - Restricted coefficients for non-OECD: 0.1224* (0.0691), 0.1482** (0.0650), 0.1126* (0.0624).
  - Observations: 916, 956, 949; Number of countries: 72, 74, 73.

### Joint inclusion of GDP- and haven-weighted averages (Table 5)
- Full sample:
  - CIT rate i = -0.0911** (0.0426); CIT rate j, weighted GDP = -0.0260 (0.0802); CIT rate j, weighted tax havens = 0.3640* (0.2103).
- OECD:
  - CIT rate i = -0.0582 (0.0439); CIT rate j, weighted GDP = 0.1533* (0.0835) — GDP-weighted spillover dominates for OECD; CIT rate j, weighted tax havens = 0.0575 (0.1228).
- Non-OECD:
  - CIT rate i = -0.1375** (0.0656); CIT rate j, weighted GDP = 0.0056 (0.1311); CIT rate j, weighted tax havens = 0.4417* (0.2713) — haven-weighted (profit-shifting) dominates for non-OECD.
- Interpretation: OECD base spillovers appear driven by GDP-weighted (real effects); non-OECD by haven-weighted (profit-shifting) effects with an implied very large semi-elasticity (reported as 5.4 in text, noting imprecision).

### The revenue cost of BEPS (illustrative calculations)
- Conceptual approach: “Turning off” the haven-weighted spillover channel (setting profit shifting cost parameters δ(i,j) to infinity) and evaluating implied changes in tax bases at unchanged tax rates; use restricted coefficients from Table 3 and 2013 statutory tax rates and estimated CIT bases.
- Illustrated results (Figure 3, text summary):
  - In nominal terms, revenue at stake is much larger for OECD members (OECD: USD 950.2 billion reported in figure caption area).
  - Relative to GDP:
    - OECD implied long run revenue losses: in the order of 1 percent of GDP.
    - Non-OECD implied long run revenue losses: around 1.3 percent of GDP.
  - Short-run effects summarized in the source figure (specific short-run numbers preserved in figure area of source).
- Contextual comparisons:
  - Gravelle (2013) estimate for U.S. loss from selected avoidance techniques: around 25 percent of corporate tax revenues (order of 0.6 percent of GDP).
  - Median ratio of tax revenue to GDP: low income countries around 15 percent versus about 35 percent in the OECD.
- Caveats emphasized:
  - Estimates are highly speculative and illustrative.
  - Identification of haven spillovers is difficult; haven classification does not fully capture features making jurisdictions attractive for profit shifting.
  - Firm-level data would be preferable but are scarce for developing countries.

### Strategic rate spillovers (estimation of equation (15)) — main findings
- Estimation: system GMM with instruments based on lags of own and weighted CIT rates.
- Main result: strategic complementarity in tax-setting — countries cut their own rates in response to rate reductions elsewhere.
- Estimates (Table 6 summary):
  - Full sample:
    - CIT rate j, weighted GDP = 0.8015*** (0.1740)
    - CIT rate j, weighted tax havens = 0.7678* (0.4393) in alternate specification
  - OECD:
    - CIT rate j, weighted GDP = 1.0881*** (0.2104)
    - CIT rate j, weighted tax havens = 1.8420* (1.006)
    - For OECD, one point cut abroad elicits approximately a one point cut domestically.
  - Non-OECD:
    - CIT rate j, weighted GDP = 0.6128*** (0.1255)
    - CIT rate j, weighted tax havens = 0.7106* (0.4127)
    - For non-OECD, a one point cut abroad elicits about a two-thirds point domestic cut.
- Diagnostics: M1 (p values between 0.015 and 0.044 across specifications), M2 (p values generally not rejecting second-order), Hansen (p values not rejecting over-identification).
- Interpretation:
  - Responsiveness is higher and more statistically significant to GDP-weighted rates than to haven-weighted rates, especially for OECD members.
  - Possible interpretation: changes in large economies’ rates are more salient for policymaking even if profit-shifting channels matter more substantively for non-OECD countries.

### Appendix 2 — Results Using Average Effective Tax Rates (AETRs)
- Sample and data limitations:
  - Analysis focuses only on non-OECD members.
  - Sample using AETRs contains 43 developing countries.
  - Data limitations prevent construction of a haven-weighted average of AETRs; an unweighted average with a time trend is used instead.
- Key findings:
  - Base spillover results differ from those using statutory rates; own rate effects are insignificantly estimated in this AETR-based specification.
  - Insignificance of haven-weighted rates for AETRs is less surprising because avoidance opportunities are expected to be associated with differences in statutory rates, not in AETRs.
  - Short-term base spillover effects when weighting AETRs abroad using GDP (column (1)) or by inverse-distance (column (3)) are larger and more significant than in the statutory-rate specification.
  - For strategic rate spillovers using AETRs:
    - Column (6) shows no significant effect from inverse-distance weighted rates.
    - GDP-weighted (column (4)) and the unweighted average AETR (column (5)) effects are significant, though less so than in the statutory-rate specification.
- Selected estimation results (Appendix Table A1 highlights):
  - Base Spillover (columns (1)-(3)):
    - CIT Base, lagged: (1) 0.9248*** (0.0710); (2) 0.7471*** (0.1872); (3) 0.7842*** (0.1003)
    - EATR i: (1) 0.0083 (0.0041); (2) -0.1268* (0.0747); (3) -0.1423 (0.1326)
    - EATR j, weighted GDP: (1) 0.2788** (0.1461)
    - θ(휑): (1) 0.0011 (0.0493); (2) -0.5018 (0.3826); (3) -0.6596 (0.7114)
    - θ(γ): (1) 3.7095 (4.1756); (2) 0.2644 (0.4392); (3) 1.3529* (0.8029)
    - γ = −휑 (p value): (1) 0.060; (2) 0.502; (3) 0.242
    - Restricted Coefficient: (1) 0.013** (0.0061); (2) 0.1007* (0.0637); (3) 0.256*** (0.0895)
    - Observations: (1) 326; (2) 307; (3) 307
  - Strategic Spillover (columns (4)-(6)):
    - EATR j, weighted GDP: (4) 1.8971* (1.0860)
    - EATR j, simple average: (5) 0.3784* (0.2123)
    - EATR j, weighted inverse-distance: (6) 0.1700 (0.3319)
    - Observations: (4)-(6): 397; Number of countries: 41.

### Conclusions (as stated in source)
- Empirical evidence indicates that base erosion, profit shifting and international tax competition matter materially for developing countries and may matter at least as much as for advanced economies.
- Key findings:
  - Base spillovers from others’ tax rates can be stronger for non-OECD countries than for OECD countries and may operate more through profit shifting than through real investment effects.
  - Spillovers associated with tax havens are identified as important, particularly for non-OECD members.
  - Illustrative long-run revenue loss estimates from haven-related profit shifting are in the order of something over one percentage point of GDP for non-OECD members and around 1 percent of GDP for OECD members.
- Limitations and recommendations:
  - Results are tentative due to identification challenges and limitations of the haven classification.
  - Firm-level data would enable stronger inference but remain scarce for developing countries.
  - Developing countries have a considerable stake in ongoing international tax policy debates and potential reconfiguration of international tax architecture.

*Source: _wp15118 - References, IMF PDF content provided.*

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

### _wp15118 - References

### Major themes and objectives
- Focus: provide the first systematic empirical evidence on whether BEPS and tax competition matter for developing countries, and quantify possible effects.
- Two distinct cross-border fiscal externalities analyzed:
  - Base spillovers: impact of one country’s tax policy on the tax bases of other countries (via real activities and/or profit shifting).
  - Strategic rate spillovers: impact on a country’s policy choices of tax changes abroad (tax competition).
- Data exploited: aggregate data on corporate tax bases for a panel of 120 countries (and tax rates for 173 countries), over the period 1980–2013.
- Emphasis: concentrate primarily on base spillovers because of direct relevance to BEPS and to enable an estimate—acknowledged as very speculative—of revenue impact of international corporate tax avoidance involving tax ‘havens’.

### Theoretical framework (key constructs and implications)
- World of n countries; country i has population share ℎi; world population normalized to unity (∑ℎj = 1).
- Multinational with one affiliate per country; real activity in i described by fi(ki) with fi′>0 and fi′′<0.
- Base in country i: bi = hi ki + ∑sij (equation (1)).
- Multinational’s after-tax profit Π given by expression (2), with tax rates ti and base-shifting costs cij(sij).
- First-order condition for real capital allocation: fi′(ki) = ρ + ti (equation (3)); ki implicitly a function ki(t1,..,tn).
- For quadratic production functions, tax changes imply:
  - ∂ki/∂ti = −(1 − hi) < 0 (equation (4)).
  - ∂ki/∂tj = hj > 0 (equation (5)).
  - Size (hi) matters critically for real-investment channel: effects from j on i larger the larger is country j.
- Base-shifting first-order condition (for sij): cij′(sij) + cji′(sij) = tj − ti when tj > ti (equation (6)); with quadratic costs cij(sij) = (1/2) Δij sij^2 this yields responsiveness ∂sij/∂tj = 1/δ(i,j) > 0 (equation (7)), where δ(i,j) ≡ Δij + Δji.
- Profit-shifting channel is independent of country sizes and depends on ease of shifting (δ(i,j)); even small jurisdictions can induce shifting from large ones.
- Combined small change impact on per capita base:
  - dbi/hi = βi (dti − ∑ωij dtj) (equation (8)), where βi and ωij defined in (9) and (10).
  - If effects operate only through real investment, weights become ωij = hj/(1 − hi) (equation (11)) — “GDP-weighted”.
  - If effects operate only through profit shifting, weights become ωij = (1/δ(i,j)) / ∑p≠i (1/δ(i,p)) (equation (12)) — “haven/shiftability-weighted”.

### Empirical specification and identification
- Base spillover regressions estimated with dynamic specification:
  - bit = λ bi,t−1 + φ τit + γ W−i τ−i,t + ζ′ Xit + αi + μt + εit (equation (13)),
    - bit: CIT base for country i at time t; τit: domestic CIT rate; W−i τ−i,t: weighted average of foreign statutory CIT rates (∑ωij τjt, ∑ωij = 1); Xit: controls; αi, μt: fixed effects.
  - Short-run own-rate effect φ expected negative; long-run impact θ(φ) ≡ φ/(1 − λ).
  - Spillover coefficient γ expected positive; long-run θ(γ) ≡ γ/(1 − λ) (equation (14)).
- Weighting matrices to distinguish channels:
  - GDP-weighted rates: tax rate in j weighted by j’s GDP as share of total GDP of all countries other than i (captures real investment channel; inspired by equation (11)).
  - Haven-weighted rates: unweighted average of rates in jurisdictions on a commonly-used list of ‘tax havens’ (captures profit-shifting channel; inspired by equation (12) and the absence of direct data on δ(i,j)).
  - Inverse-distance weighting also considered to capture geographic proximity effects.
- Identification issues and solutions:
  - Haven-weight variable does not vary across non-haven countries and is collinear with time effects—resolved by imposing a common linear time trend (as in Devereux et al. (2008)) or by imposing φ = −γ (testable restriction derived from theory).
  - Use of statutory rates versus Average Effective Tax Rates (AETR): AETRs more relevant for investment but data limited; AETRs pursued where available (appendix).

### Empirical strategy for strategic rate spillovers
- Rate-setting (strategic) regressions follow:
  - τit = b W−i τ−i,t + ζ′ Xit + ai + ct + εit (equation (15)),
    - same three weighted-average constructs used (GDP-weighted, haven-weighted, inverse-distance).
  - Specification includes country effects and a common time trend; lagged dependent variable omitted (as in prior literature).

### Contextual and literature-based quantitative benchmarks preserved from source
- Data scope: panel of 120 countries for tax bases; tax rates for 173 countries; period 1980–2013.
- Prior findings cited:
  - De Mooij and Ederveen (2008): a 10 percentage point reduction in a country’s average effective tax rate increases its stock of FDI, on average and in the long run, by over 30 percent.
  - Heckemeyer and Overesch (2013): consensus semi-elasticity of −0.8, implying that a 10 percentage point higher tax rate will reduce reported profit in an affiliate by 8 percent.
  - Devereux et al. (2008): among OECD countries, a 10 percentage point decrease in statutory CIT rates in other countries generates, on average, a cut of 7 percentage points in response.
  - Klemm and Van Parys (2012): among Sub-Saharan African and Caribbean countries, estimated strategic interactions of 2.5 to 3 points in response to a 10 point tax cut abroad.
- Figure 1 source and note retained: Source: IMF Staff estimates; data from IMF’s Fiscal Affairs Department database. Note: Total tax revenue including grants and excluding social contributions; resource-rich countries excluded.

### Key methodological caveats highlighted in the source
- Imperfections and limitations emphasized: difficulty disentangling real investment and profit-shifting channels precisely; many identification challenges due to data limitations.
- AETR data limitations: not available for as many countries or as long a period as statutory rates; calculated AETRs depend on tax base elements producing endogeneity concerns.
- Measured GDP may be affected by profit shifting (mispricing of exports and imports), making GDP an imperfect size indicator for weighting, but likely second-order relative to GNP effects.
- The approach for revenue impact of BEPS and havens is acknowledged as “very speculative”.

*Source: _wp15118 - References, IMF PDF content provided.*

### Appendix 2.

### Appendix 2.

### Identification and estimation issues
- Equations (13) and (15) are estimated by system generalized method of moments (GMM), using only internal instruments.
- While the panel is sufficiently long that Nickell bias may not be a significant concern, other endogeneity issues arise.
  - In the base spillover regression, shocks that affect a country’s domestic tax base may also affect its contemporaneous tax rate choice.
  - The estimation of the CIT base by simply dividing revenues by the main statutory rate can give rise to measurement error when, as is quite often the case, more than one CIT rate is applied.
  - In the strategic rate spillover equation, tax rates are evidently jointly determined across countries.
- A further possibility is to take W−i to be the simple average rate over all countries other than i; this though raises still more sharply the identification challenge mentioned below, and for brevity the results are not reported here.
- Using country-specific trends instead of a common trend gives very similar results, not reported here.

### Tax-haven effects and data limitations
- Perhaps more troubling data limitations relate to the estimation of effects operating through tax havens.
  - The difficulty is that the attractions of tax ‘havens’ do not solely, or even mainly, derive from low statutory CIT rates, but from special regimes and arrangements for which descriptive data are unavailable.
  - The identification of haven effects thus depends on a plausible but (on our data) untestable correlation between movements in their statutory rates and special regimes. The results, for this reason, can be no more than indicative.
- Over the full sample period, the average CIT rate in the ‘havens’ is around 17 percent, compared to 32 percent for the full sample (Table 1): see also Figure 2.
- Many havens are small, however, and a low rate is common among smaller countries more generally.
  - Regressing the CIT rate on country size (which enters with a significant positive coefficient, as models of tax competition would predict) and a dummy for tax haven status (and the using other controls being used), it emerges that tax-havens actually have, on average, a significantly higher CIT rate than otherwise similar countries.

### Data, sample, and construction
- The sample is an unbalanced panel comprising 173 countries over 1980–2013.
- The countries in the sample, identifying those labeled, following Gravelle (2013), as ‘havens’, classified by income group and as between OECD members and non-members (at the end of the sample period), are listed in Appendix 1.
- The latter group comprises a wide range of countries, of course, but for brevity we sometimes refer to this as the group of developing countries, lower income countries indeed being heavily represented in the sample.
- Data on CIT revenues and statutory tax rates are from the IMF’s Fiscal Affairs Department database.
  - The country coverage of CIT rates is full, though unbalanced in the time dimension.
- To eliminate artificial variation in the weighted average tax rates as a result of missing observations for certain country-year pairs, we linearly interpolated the tax rate series for years with missing tax rates.
  - The balanced panel of tax rates this creates is used only for calculating the weighted average tax rates; own tax rates in all regressions are actual values, not interpolations.
- Resource-rich countries are excluded from the exercise in the sense that their tax bases are not treated as dependent variables, since they will likely have distinct drivers and reflect a variety of distinct tax design choices; the tax rates set by these countries are, however, included in constructing the various average tax rates used as explanatory variables.
- As mentioned above, the CIT base in percent of GDP, bi, is calculated by dividing CIT revenue in recent of GDP by the standard CIT rate; lack of revenue data means that this is possible for only 121 countries.
- The far more limited data on average effective tax rates (AETR) used in

*Source: Appendix 2.*

### Appendix 2, for 43 countries

### Appendix 2, for 43 countries

### Data and descriptive statistics
- Sample composition: 43 countries enumerated in the source text (list preserved in source).
- Controls used in regressions: (the log of) GDP per capita, the share of agriculture in value-added, trade openness (the sum of non-resource exports plus imports, relative to GDP), and inflation.
- Data sources and definitions:
  - Agricultural value-added: World Development Indicators (WDI).
  - Trade openness: IMF’s International Financial Statistics (IFS).
  - GDP per capita: constant (2000) U.S. dollars from WDI.
  - Inflation: annual change in the consumer price index from IFS.
- Key descriptive statistics (Table 1):
  - Statutory CIT Rate, in percent: Obs. 3037; Mean 32.15; Max. 61.80; Min. 0.00; Std. Dev. 10.85
  - GDP-weighted average tax rate, in percent: Obs. 3037; Mean 39.18; Max. 48.04; Min. 26.98; Std. Dev. 5.28
  - Haven-weighted average CIT rate, in percent: Obs. 3037; Mean 17.09; Max. 24.46; Min. 11.08; Std. Dev. 3.55
  - Inverse-distance-weighted average CIT rate, in percent: Obs. 4771; Mean 32.08; Max. 42.18; Min. 18.60; Std. Dev. 4.60
  - CIT revenue, percent of GDP: Obs. 2161; Mean 2.64; Max. 13.37; Min. 0.00; Std. Dev. 1.53
    - OECD countries: Obs. 913; Mean 2.76; Max. 8.02; Min. 0.26; Std. Dev. 1.28
    - Non-OECD countries: Obs. 2354; Mean 2.42; Max. 18.40; Min. 0.01; Std. Dev. 1.98
  - CIT base, percent of GDP: Obs. 2161; Mean 8.59; Max. 29.99; Min. 0.00; Std. Dev. 5.45
    - OECD countries: Obs. 893; Mean 8.75; Max. 29.99; Min. 1.06; Std. Dev. 4.61
    - Non-OECD countries: Obs. 1268; Mean 8.47; Max. 29.73; Min. 0.00; Std. Dev. 5.97
  - AETR, in percent: Obs. 508; Mean 22.23; Max. 40.27; Min. -11.61; Std. Dev. 9.24
  - GDP per capita, 2000 USD: Obs. 1970; Mean 13349; Max. 87716; Min. 126; Std. Dev. 15353
  - Trade openness, percent of GDP: Obs. 1974; Mean 79.04; Max. 436.95; Min. 6.32; Std. Dev. 45.66
  - Inflation, in percent: Obs. 1925; Mean 36.46; Max. 11749.64; Min. -4.47; Std. Dev. 368.39
- Stylized trend: Figure 2 shows a pronounced decline in mean statutory CIT rates by 15 to 20 percentage points over the last three decades for both OECD and non-OECD groups; haven-weighted averages interpolated are substantially lower.

### Base spillovers (estimation of equation (13))
- Estimation approach: System GMM (one step, robust) with alternative weighting matrices: GDP-weighted, haven-weighted, inverse-distance-weighted. Dependent variable: CIT base.
- Own-rate short-run effects:
  - Column (1) GDP-weighted: short-run marginal coefficient on own CIT rate = -0.0818*** (standard error 0.0396) → interpreted as a one percentage point increase in a country’s CIT rate reduces its CIT base by 0.08 percent of GDP.
  - Short-run semi-elasticity evaluated at mean CIT base of 8.59 percent of GDP: calculated short-run semi-elasticity of -0.9 (calculated as (0.08/8.79)×100 in source).
- Own-rate long-run effects:
  - θ(φ) in column (1): -1.1608 (standard error 1.0649) — large but imprecise; null of no long run effect not rejected in column (1).
  - Haven-weighted long-run effect significant at 5 percent in column (2); inverse-distance long-run effect insignificant in column (3).
- Cross-country (base) spillovers:
  - GDP-weighted (column 1): CIT rate j, weighted GDP = 0.1763* (standard error 0.0982). Interpretation: a one percentage point reduction abroad reduces the typical country’s CIT base in short run by 0.18 percent of GDP (short-run semi-elasticity of over two noted).
  - Haven-weighted (column 2): CIT rate j, weighted tax havens = 0.3544** (standard error 0.1698) — larger and more significant effects.
  - Inverse-distance (column 3): CIT rate j, weighted inverse-distance = 0.3317* (standard error 0.1998) — similar magnitude but barely significant.
- Equality restriction test (γ = −φ):
  - The null that φ = −γ is not rejected for any weight structure (p values: column (1) 0.329; column (2) 0.120; column (3) 0.204).
  - Imposing the restriction yields restricted coefficients around 0.0885** (column 1), 0.0986*** (column 2), 0.0832** (column 3).
- Diagnostics (Table 2 summary):
  - M1 (p value): 0.001, 0.001, 0.000 for columns (1)-(3).
  - M2 (p value): 0.780, 0.859, 0.789.
  - Hansen (p value): 0.452, 0.523, 0.385.
  - Observations: 1540, 1694, 1687; Number of countries: 100, 105, 103.

- Heterogeneity by income/OECD membership (Table 3):
  - All (column 1, repeating haven-weighted): CIT rate i = -0.0991*** (0.0413); CIT rate j, haven weighted = 0.3544** (0.1698).
  - OECD (column 2): CIT rate i = -0.0596* (0.0371); CIT rate j, haven weighted = 0.3423** (0.1767).
  - Non-OECD (column 3): CIT rate i = -0.1376** (0.0657); CIT rate j, haven weighted = 0.4421* (0.2718).
  - Long-run θ(φ): All -0.8165** (0.4739); OECD -0.2571 (0.1695); Non-OECD -0.9398* (0.5169).
  - Imposed equality (restricted coefficient): All 0.0986*** (0.0413); OECD 0.0527 (0.0416) [insignificant]; Non-OECD 0.1482** (0.0650).
  - Observations and sample sizes: All 1694; OECD 624; Non-OECD 956; Number of countries: All 105; OECD 28; Non-OECD 74.

- Non-OECD detailed results (Table 4):
  - Own-rate coefficients across weights: (1) GDP-weighted CIT rate i = -0.1206* (0.0727); (2) haven-weighted CIT rate i = -0.1376** (0.0657); (3) inverse-distance CIT rate i = -0.1126* (0.0624).
  - Spillovers in non-OECD:
    - GDP-weighted: CIT rate j, weighted GDP = 0.1324 (0.1410) — insignificant.
    - Haven-weighted: CIT rate j, weighted tax havens = 0.4421* (0.2718) — significant at 10 percent.
    - Inverse-distance: CIT rate j, weighted inverse distance = 0.0525 (0.3296) — insignificant.
  - Restricted coefficients for non-OECD: 0.1224* (0.0691), 0.1482** (0.0650), 0.1126* (0.0624) across the three weightings.
  - Observations: 916, 956, 949; Number of countries: 72, 74, 73.

- Joint inclusion of GDP- and haven-weighted averages (Table 5):
  - Full sample (column): CIT rate i = -0.0911** (0.0426); CIT rate j, weighted GDP = -0.0260 (0.0802); CIT rate j, weighted tax havens = 0.3640* (0.2103).
  - OECD: CIT rate i = -0.0582 (0.0439); CIT rate j, weighted GDP = 0.1533* (0.0835) — GDP-weighted spillover dominates for OECD; CIT rate j, weighted tax havens = 0.0575 (0.1228).
  - Non-OECD: CIT rate i = -0.1375** (0.0656); CIT rate j, weighted GDP = 0.0056 (0.1311); CIT rate j, weighted tax havens = 0.4417* (0.2713) — haven-weighted (profit-shifting) dominates for non-OECD.
  - Interpretation: OECD base spillovers appear driven by GDP-weighted (real effects); non-OECD by haven-weighted (profit-shifting) effects with an implied very large semi-elasticity (reported as 5.4 in text, noting imprecision).

### The revenue cost of BEPS (illustrative calculations)
- Conceptual approach: “Turning off” the haven-weighted spillover channel (setting profit shifting cost parameters δ(i,j) to infinity) and evaluating implied changes in tax bases at unchanged tax rates; use restricted coefficients from Table 3 and 2013 statutory tax rates and estimated CIT bases.
- Illustrated results (Figure 3, text summary):
  - In nominal terms, revenue at stake is much larger for OECD members (OECD: USD 950.2 billion reported in figure caption area).
  - Relative to GDP:
    - OECD implied long run revenue losses: in the order of 1 percent of GDP.
    - Non-OECD implied long run revenue losses: around 1.3 percent of GDP.
  - Short-run effects summarized in the box beneath the figure in the source (specific short-run numbers preserved in figure area of source).
- Contextual comparisons:
  - Gravelle (2013) estimate for U.S. loss from selected avoidance techniques: around 25 percent of corporate tax revenues (order of 0.6 percent of GDP).
  - Median ratio of tax revenue to GDP: low income countries around 15 percent versus about 35 percent in the OECD — underscoring that a loss of ~1+ percent of GDP is large relative to developing countries’ revenue base.
- Caveats emphasized:
  - Estimates are highly speculative and illustrative.
  - Identification of haven spillovers is difficult; haven classification here does not fully capture features making jurisdictions attractive for profit shifting.
  - Firm-level data would be preferable but are scarce for developing countries.

### Strategic rate spillovers (estimation of equation (15))
- Dependent variable: statutory CIT rate. Estimation uses system GMM with instruments based on lags of own and weighted CIT rates.
- Main result: strategic complementarity in tax-setting — countries cut their own rates in response to rate reductions elsewhere.
- Estimates (Table 6 summary):
  - Full sample:
    - CIT rate j, weighted GDP = 0.8015*** (0.1740)
    - CIT rate j, weighted tax havens = 0.7678* (0.4393) in alternate specification
  - OECD:
    - CIT rate j, weighted GDP = 1.0881*** (0.2104)
    - CIT rate j, weighted tax havens = 1.8420* (1.006)
    - For OECD, one point cut abroad elicits approximately a one point cut domestically (cannot reject null of one-for-one response).
  - Non-OECD:
    - CIT rate j, weighted GDP = 0.6128*** (0.1255)
    - CIT rate j, weighted tax havens = 0.7106* (0.4127)
    - For non-OECD, a one point cut abroad elicits about a two-thirds point domestic cut.
  - Diagnostics: M1 (p values between 0.015 and 0.044 across specifications), M2 (p values generally not rejecting second-order), Hansen (p values not rejecting over-identification).
- Interpretation:
  - Responsiveness is higher and more statistically significant to GDP-weighted rates than to haven-weighted rates, especially for OECD members.
  - Possible interpretation: changes in large economies’ rates are more salient for policymaking even if profit-shifting channels matter more substantively for non-OECD countries.

### Conclusions
- Empirical evidence indicates that base erosion, profit shifting and international tax competition matter materially for developing countries and may matter at least as much as for advanced economies.
- Key findings:
  - Base spillovers from others’ tax rates can be stronger for non-OECD countries than for OECD countries and may operate more through profit shifting than through real investment effects.
  - Spillovers associated with tax havens are identified as important, particularly for non-OECD members.
  - Illustrative long-run revenue loss estimates from haven-related profit shifting are in the order of something over one percentage point of GDP for non-OECD members and around 1 percent of GDP for OECD members — amounts large relative to developing countries’ total revenue.
- Limitations and recommendations:
  - Results are tentative due to identification challenges and limitations of the haven classification.
  - Firm-level data would enable stronger inference but remain scarce for developing countries.
  - Developing countries have a considerable stake in ongoing international tax policy debates and potential reconfiguration of international tax architecture.

*Source: Appendix 2, for 43 countries (IMF working paper appendix content as provided).*

### Appendix 1. Country Listing and Classification

### Appendix 1. Country Listing and Classification

### Low- and middle income countries (as listed)
- Afghanistan*
- Albania
- Algeria*
- Antigua and Barbuda* 1/
- Argentina
- Armenia
- Bangladesh
- Barbados 1/
- Republic of Belarus
- Belize 1/
- Benin*
- Bhutan*
- Bolivia
- Bosnia and Herzegovina
- Botswana
- Brazil
- Bulgaria
- Burkina Faso*
- Burundi*
- Cambodia
- Cameroon
- Cabo Verde*
- Central African Republic*
- Chad*
- Chile* 2/
- China
- Colombia
- Comoros*
- Republic of Congo*
- Costa Rica 1/
- Côte d’Ivoire
- Djibouti*
- Dominica* 1/
- Dominican Republic
- Ecuador
- Egypt
- El Salvador
- Republic of Equatorial Guinea*
- Eritrea*
- Ethiopia
- Fiji
- Gabon
- The Gambia
- Georgia
- Ghana
- Guatemala
- Grenada* 1/
- Guinea
- Guinea-Bissau*
- Guyana
- Haiti
- Honduras
- Hungary 2/
- India
- Indonesia
- Islamic Republic of Iran
- Iraq
- Jamaica
- Jordan 1/
- Kazakhstan
- Kenya*
- Kyrgyz Republic
- Lao P.D.R.*
- Latvia
- Lebanon* 1/
- Lesotho*
- Liberia* 1/
- Libya*
- Lithuania
- Former Yugoslav Republic of Macedonia
- Madagascar*
- Malaysia
- Malawi
- Maldives* 1/
- Mali*
- Mauritania*
- Mauritius 1/
- Mexico* 2/
- Moldova
- Montenegro*
- Mongolia*
- Montserrat* 1/
- Morocco
- Mozambique
- Myanmar*
- Namibia
- Nepal*
- Nicaragua*
- Nigeria
- Niger*
- Pakistan
- Panama 1/
- Papua New Guinea
- Paraguay
- Peru
- Philippines
- Romania
- Russian Federation
- Rwanda*
- São Tomé and Príncipe*
- Senegal
- Serbia*
- Seychelles* 1/
- Sierra Leone
- Sri Lanka
- Solomon Islands*
- South Africa
- St. Kitts and Nevis* 1/
- St. Lucia 1/
- St. Vincent and the Grenadines 1/
- Swaziland
- Syrian Arab Republic
- Tajikistan*
- Tanzania
- Thailand
- Togo*
- Tonga* 1/
- Tunisia
- Turkey 2/
- Turkmenistan
- Uganda
- Ukraine
- Uruguay
- Uzbekistan
- Vanuatu* 1/
- Venezuela
- Vietnam
- Yemen
- Zambia
- Zimbabwe

### High income countries (as listed)
- Australia 2/
- Austria 2/
- The Bahamas 1/
- Bahrain* 1/
- Belgium 2/
- Canada 2/
- Croatia
- Cyprus 1/
- Czech Republic 2/
- Denmark 2/
- Estonia 2/
- Finland 2/
- France 2/
- Germany 2/
- Greece 2/
- Hong Kong SAR 1/
- Iceland 2/
- Ireland 1/,2/
- Israel 2/
- Italy 2/
- Japan 2/
- Korea 2/
- Kuwait
- Luxemburg 1/,2/
- Malta 1/
- Netherlands 2/
- New Zealand 2/
- Norway 2/
- Oman
- Poland 2/
- Portugal 2/
- San Marino* 1/
- Saudi Arabia*
- Singapore 1/
- Slovak Republic 2/
- Slovenia 2/
- Spain 2/
- Sweden 2/
- Switzerland 1/,2/
- Trinidad and Tobago*
- United Arab Emirates
- United Kingdom 2/
- United States 2/

### Classification notes
- Classification by income group follows the World Bank.
- Data on CIT rates are available for all countries listed; * indicates that data on CIT revenue (and hence base) are not available.
- 1/ indicates countries labeled, following Gravelle (2013) as ‘havens’.
- 2/ indicates an OECD member.

---

### Appendix 2. Results Using Average Effective Tax Rates

### Sample and data limitations
- The analysis focuses only on non-OECD members.
- The sample using AETRs contains 43 developing countries.
- Data limitations prevent construction of a haven-weighted average of AETRs; an unweighted average with a time trend is used instead.

### Key findings (text summary)
- Base spillover results differ from those using statutory rates; own rate effects are insignificantly estimated in this AETR-based specification.
- The insignificance of haven-weighted rates is less surprising for AETRs because avoidance opportunities are expected to be associated with differences in statutory rates, not in AETRs.
- Short-term base spillover effects when weighting AETRs abroad using GDP (column (1)) or by inverse-distance (column (3)) are larger and more significant than in the statutory-rate specification, possibly reflecting that AETR better indicates the impact of tax considerations on the location of real investments.
- For strategic rate spillovers:
  - Column (6) shows no significant effect from inverse-distance weighted rates.
  - GDP-weighted (column (4)) and the unweighted average AETR (column (5)) effects are significant, though less so than in the statutory-rate specification.
  - Signs of strategic rate-setting interactions appear somewhat stronger with statutory tax rates than with AETRs.

### Appendix Table A1 — Selected estimation results (using AETRs)
- Dependent variable: CIT base.
- Estimation notes: Full set control variables and common time trend in all regressions. Robust standard errors in parenthesis. ***(**,*) indicate significance at 1 (5, 10) percent.
- Instrument/estimation details: One step, robust, with instruments based on first lag of differences in the CIT tax base and CIT tax rates (collapsed to avoid proliferation in the number of instruments) in levels equation, and second lags of their levels in the differenced equation.

Base Spillover (columns (1)-(3)):
- CIT Base, lagged:
  - (1): 0.9248*** (0.0710)
  - (2): 0.7471*** (0.1872)
  - (3): 0.7842*** (0.1003)
- EATR i:
  - (1): 0.0083 (0.0041)
  - (2): -0.1268* (0.0747)
  - (3): -0.1423 (0.1326)
- EATR j, weighted GDP:
  - (1): 0.2788** (0.1461)
- EATR j, simple average:
  - (2): 0.0668 (0.0813)
- EATR j, weighted inverse-distance:
  - (3): 0.2919*** (0.0944)
- θ(휑):
  - (1): 0.0011 (0.0493)
  - (2): -0.5018 (0.3826)
  - (3): -0.6596 (0.7114)
- θ(γ):
  - (1): 3.7095 (4.1756)
  - (2): 0.2644 (0.4392)
  - (3): 1.3529* (0.8029)
- γ = −휑 (p value):
  - (1): 0.060
  - (2): 0.502
  - (3): 0.242
- Restricted Coefficient:
  - (1): 0.013** (0.0061)
  - (2): 0.1007* (0.0637)
  - (3): 0.256*** (0.0895)
- M1 (p value):
  - (1): 0.049
  - (2): 0.033
  - (3): 0.038
- M2 (p value):
  - (1): 0.485
  - (2): 0.414
  - (3): 0.495
- Over-identification Hansen (p value):
  - (1): 0.460
  - (2): 0.532
  - (3): 0.504
- Observations:
  - (1): 326
  - (2): 307
  - (3): 307
- Number of instruments:
  - (1): 42
  - (2): 44
  - (3): 41
- Number of countries:
  - (1): 38
  - (2): 37
  - (3): 37

Strategic Spillover (columns (4)-(6)):
- EATR j, weighted GDP:
  - (4): 1.8971* (1.0860)
- EATR j, simple average:
  - (5): 0.3784* (0.2123)
- EATR j, weighted inverse-distance:
  - (6): 0.1700 (0.3319)
- M1 (p value):
  - (4): 0.008
  - (5): 0.008
  - (6): 0.034
- M2 (p value):
  - (4): 0.118
  - (5): 0.127
  - (6): 0.343
- Over-identification Hansen (p value):
  - (4): 0.690
  - (5): 0.556
  - (6): 0.389
- Observations:
  - (4): 397
  - (5): 397
  - (6): 397
- Number of instruments:
  - (4): 27
  - (5): 25
  - (6): 27
- Number of countries:
  - (4): 41
  - (5): 41
  - (6): 41

---

*Source: _wp15118 - Appendix 1. Country Listing and Classification*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15118.pdf_