## 1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors

## Source details

**Canonical URL:** [1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1405.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1405.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1405.pdf.json)

---

### Introduction and motivation
- Question: How do government revenues from natural resources affect domestic non-resource tax revenues and their components (taxes on goods and services, VAT, corporate income tax (CIT), personal income tax (PIT), and trade taxes)?
- Rationale and concerns:
  - Resource-rich countries may have incentives to keep a relatively low domestic (non-resource) tax burden to promote private sector activity and growth.
  - Higher domestic tax effort can improve public service quality and trust in government via an implicit fiscal contract between taxpayers and the state.
  - Resource revenues can be volatile; absent stabilizing fiscal frameworks, volatility transmits to the budget.
  - Heavy reliance on resource revenues is associated with weakened tax administration capacity, ad hoc tax exemptions, and discretionary application of tax laws.
- Empirical gap: No prior study had assessed the impact of resource revenues on different types of non-resource domestic taxes; this paper constructs a disaggregated database for 35 resource-rich countries (1992–2009).

### Data and sample
- Sample:
  - Unbalanced panel of 35 natural resource-rich countries for the period 1992–2009.
- Revenue definitions:
  - Dependent variables: domestic non-resource tax revenues (relative to GDP) disaggregated into TAX (total non-resource tax revenues), taxes on goods and services (G&S), VAT, CIT, PIT, and TRADE. Non-resource CIT excludes tax on profits paid by resource companies.
  - Resource revenues: government revenues from taxing extraction companies, taxing exports of those natural resources, royalties, and production sharing agreements.
- Data sources:
  - Government Financial Statistics (GFS) supplemented with IMF country desk detailed revenue information; additional databases used in appendices.
- Key aggregate statistics:
  - Resource revenues increased from 13.8 percent of GDP, on average, during the 1990s to 16.1 percent of GDP, on average, during the 2000s.
  - Total tax revenue increased from 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average, during the 2000s.
  - The share of resource revenues in total revenues has increased over time.

### Empirical specification
- Estimating equation:
  - T_it = α_i + β_1 R_it + β_2 X_it + μ_t + ε_it
    - T = domestic non-resource tax revenues (relative to GDP) — estimated separately for TAX, G&S, VAT, CIT, PIT, TRADE.
    - R = government revenues from natural resources (relative to GDP).
    - α_i = country fixed effect; μ_t = time fixed effect; ε_it = error term.
    - β_1 measures the marginal effect of an additional percentage point of government resource revenues on non-resource revenue effort.
- Control variables X_it:
  - log of GDP per capita
  - share of agriculture in value-added
  - non-resource trade openness (non-resource exports + imports relative to GDP)
  - inflation
  - external indebtedness (level)
  - ICRG corruption index (0 to 6; 0 = high corruption, 6 = low corruption)
  - foreign aid (ODA) relative to GDP

### Main empirical findings (Panel OLS with country and time fixed effects)
- Overall offset effect:
  - Estimated offset of about 30 percent between natural resource revenues and total domestic (non-resource) revenue in broad summary.
  - Table 1 exact contemporaneous coefficient estimates for Resource Revenue (standard errors in parenthesis):
    - Tax (1): -0.4286*** (0.0799)
    - G&S (2): -0.1238*** (0.0359)
    - VAT (3): -0.1154*** (0.0379)
    - CIT (4): -0.0392*** (0.0075)
    - PIT (5): -0.0604*** (0.0225)
    - Trade (6): -0.0284*** (0.0108)
- Quantitative interpretations reported:
  - For each additional percentage point of revenue from natural resources, there is a decline in total non-resource tax revenues by 0.43 percentage points of GDP (Column 1).
  - An increase in resource revenues by one standard deviation would imply a non-resource revenue loss of 1.6 percentage points of GDP (one standard deviation of resource revenues is 3.93; standard deviations calculated for each country separately).
  - Average annual fluctuation in resource revenues is associated with an estimated annual fluctuation in non-resource revenue of about 0.8 percentage point of GDP.
- Component-specific offsets:
  - Taxes on goods and services (G&S): offset of 0.12 percentage points of GDP per percentage point of resource revenues.
  - VAT: offset of 0.11 percentage points of GDP per percentage point of resource revenues.
  - CIT: offset of about 0.04 percentage points of GDP.
  - PIT: offset of about 0.06 percentage points of GDP.
  - Trade taxes: offset of about 0.03 percentage points of GDP.
- Control variable relationships (selected, from Table 1):
  - Agriculture (share of GDP) negative when significant; e.g., Tax: -0.2240** (0.1131).
  - Log(GDP per capita) positive for some taxes; e.g., VAT: 2.9318*** (0.9944).
  - Log(Inflation) positive for Tax: 1.8038* (0.9431).
  - ODA (share of GDP) negative association with tax effort for some specifications; e.g., CIT: -0.1192*** (0.0326).
  - No evidence of an association between level of corruption and tax revenue effort in the baseline OLS.

### Endogeneity and dynamic concerns; System-GMM estimates
- Concerns:
  - Potential endogeneity of natural resource revenues and persistence in tax revenue data.
- Estimation approach:
  - System-GMM used to address endogeneity and dynamic effects (lagged dependent variable included).
  - Diagnostics: Hansen and Sargan statistics reported; Arellano-Bond tests (M1 and M2) suggest first-order serial correlation present but second-order serial correlation absent in most specifications.
- System-GMM contemporaneous Resource Revenue coefficients (standard errors in parenthesis):
  - Tax (1): -0.2917*** (0.1215)
  - G&S (2): -0.0497** (0.0227)
  - VAT (3): -0.0620*** (0.0202)
  - CIT (4): -0.0172* (0.0098)
  - PIT (5): -0.0007 (0.0099) [statistically insignificant]
  - Trade (6): -0.0091* (0.0059)
- Long-run (ultimate) offset effects θ (Table 2):
  - Tax: θ = -0.4008*** (0.1437)
  - G&S: θ = -0.4131*** (0.1658)
  - VAT: θ = -0.2102*** (0.0572)
  - CIT: θ = -0.0550* (0.0335)
  - PIT: θ = -0.0102 (0.1457)
  - Trade: θ = -0.0801 (0.0761)
- Interpretation:
  - Negative effect of natural resource revenues on total domestic tax effort and most main tax sources remains significant under system-GMM.
  - Considering the long-run θ estimates, an increase in resource revenues by one standard deviation would imply an ultimate non-resource revenue loss of 1.8 percentage points of GDP.

### Robustness: hydrocarbon producers subset
- Subsample: hydrocarbon (oil and gas) producer countries only.
- Table 3 contemporaneous Resource Revenue coefficients (standard errors in parenthesis):
  - Tax (1): -0.1509*** (0.0582)
  - G&S (2): -0.0312* (0.0182)
  - VAT (3): -0.0526*** (0.0186)
  - CIT (4): -0.0209* (0.0112)
  - PIT (5): -0.0009 (0.0076) [statistically insignificant]
  - Trade (6): -0.0091** (0.0045)
- Long-run θ estimates for hydrocarbon producers (Table 3):
  - Tax: θ = -0.4359** (0.1976)
  - G&S: θ = -0.3929* (0.2141)
  - VAT: θ = -0.1818*** (0.0536)
  - CIT: θ = -0.0695** (0.0311)
  - PIT: θ = -0.0198 (0.1638)
  - Trade: θ = -0.0861 (0.5412)
- Interpretation:
  - Results qualitatively similar to full sample; contemporaneous offset effect on total non-resource revenues appears smaller in hydrocarbon subset, but long-run offsets remain economically significant.

### Robustness tests and heterogeneity
- Outliers:
  - Outliers identified using Hadi (1994): Russia and Norway; Sierra Leone and Senegal also considered outliers for reasons noted.
  - Estimates without outliers: "not significantly different from the full sample estimates."
- Low Resource Dependence subsample (Table 4) — Resource Revenue coefficients (standard errors in parenthesis):
  - Tax: -0.2690** (0.1355)
  - G&S: -0.0535* (0.0306)
  - VAT: -0.0731* (0.0464)
  - CIT: -0.0360* (0.0217)
  - PIT: -0.0297* (0.0174)
  - Trade: -0.0259*** (0.0097)
- Heterogeneity by domestic tax effort (sample split around average ~15 percent of GDP) (Table 5):
  - Low Tax Effort (selected Resource Revenue coefficients, standard errors in parenthesis):
    - Tax: -0.1734*** (0.0654)
    - G&S: -0.0551*** (0.0211)
    - VAT: -0.0267** (0.0119)
    - CIT: -0.0093* (0.0056)
    - PIT: 0.0176 (0.0172)
    - Trade: -0.0081 (0.0066)
  - High Tax Effort (selected Resource Revenue coefficients, standard errors in parenthesis):
    - Tax: -0.3581*** (0.1373)
    - G&S: -0.0708*** (0.0239)
    - VAT: -0.0410*** (0.0134)
    - CIT: -0.0314 (0.0213)
    - PIT: -0.0177** (0.0088)
    - Trade: -0.0068 (0.0189)
  - Interpretation: countries with relatively high non-resource domestic tax revenue effort show larger contemporaneous offsets; those with lower tax effort show lower contemporaneous offsets and lower computed ultimate effects.

### Normalization by resource- and non-resource GDP (Table 6)
- When dependent variables and revenue variables are normalized by non-resource GDP and resource GDP respectively:
  - Key result: "for each additional percentage point of natural resource revenues, there is a decline in total non resource tax revenue by 0.0042 percentage points of non resource GDP."
  - Table 6 Resource Revenue coefficients (standard errors in parenthesis):
    - Tax: -0.0042* (0.0016)
    - G&S: -0.0021* (0.0006)
    - VAT: -0.0012*** (0.0004)
    - CIT: -0.0014* (0.0006)
    - PIT: -0.0002 (0.0003)
    - Trade: -0.0142 (0.0121)
  - Quantitative interpretations in text:
    - "An increase in resource revenues (in percent of resource GDP) by one standard deviation implies a contemporaneous non resource revenue loss of about 1 percentage points of non resource GDP."
    - "One standard deviation of resource revenues is now about its average change for a period of 18 months. That is the average annual fluctuation in resource revenues is associated with an estimated annual fluctuation in non resource revenue of about 0.7 percentage points of non resource GDP."
    - "Its ultimate offset effect is estimated at about 1.6 percentage points of non resource GDP."

### Additional robustness checks (Appendix II)
- Nonlinear effects (resource revenue squared) (Table A2.1):
  - Resource Revenue coefficients remain negative and significant for many specifications; Resource Revenue (squared) coefficients not statistically significant in regressions.
  - Example: Tax: Resource Revenue -0.2814** (0.1133); Resource Revenue (squared) 0.0273 (0.3698).
- Alternative proxy: total natural resource rents (World Bank WDI) (Table A2.2):
  - Resource Rents coefficients (standard errors in parenthesis):
    - Tax: -0.1822** (0.0814)
    - G&S: -0.0230* (0.0144)
    - VAT: -0.0173** (0.0079)
    - CIT: -0.0076* (0.0047)
    - PIT: -0.0007 (0.0046)
    - Trade: -0.0027 (0.0062)
  - Findings qualitatively similar: significant negative effect of resource rents on total domestic tax revenue and main tax sources; largest offset on taxes on goods and services and VAT; smaller offset on CIT; offset on trade taxes statistically insignificant.

### Tax revenue trends and tax design evidence
- Tax revenue trends and composition:
  - Total tax revenue increased from 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average, during the 2000s.
  - The largest contributors to this increase were taxes on goods and services and income taxes.
  - Tax revenue from taxes on international trade declined as a result of trade liberalization and a reduction in trade tariffs.
- VAT and CIT design across regions:
  - VAT rates are lower in resource-rich countries in some regions (notably Middle-East); evidence on reduced VAT rates is mixed.
  - The number of VAT rates is larger in resource-rich countries.
  - Average corporate income tax rates on non-resource activities tend to be lower in resource-rich countries.

### Data diagnostics and descriptive statistics (Table A1.1)
- Descriptive statistics (preserve figures exactly):
  - Total Tax Revenue, percent of GDP — Obs. 527; Mean 14.8; Max. 54.4; Min. 0.4; Std. Dev. 8.0
  - Tax Revenue from Goods and Services, percent of GDP — Obs. 552; Mean 5.0; Max. 15.8; Min. 0.0; Std. Dev. 3.6
  - VAT Tax Revenue, percent of GDP — Obs. 269; Mean 4.3; Max. 9.9; Min. 0.0; Std. Dev. 2.1
  - Corporate Income Tax Revenue, percent of GDP — Obs. 361; Mean 1.8; Max. 9.2; Min. 0.0; Std. Dev. 1.4
  - Personal Income Tax Revenue, percent of GDP — Obs. 440; Mean 1.9; Max. 11.5; Min. 0.0; Std. Dev. 2.1
  - Trade Tax Revenue, percent of GDP — Obs. 467; Mean 1.9; Max. 7.0; Min. 0.1; Std. Dev. 1.3
  - Resource Revenue, percent of GDP — Obs. 551; Mean 15.2; Max. 68.7; Min. 0.0; Std. Dev. 14.1
  - Natural Resource Rents, percent of GDP — Obs. 665; Mean 21.6; Max. 80.7; Min. 0.1; Std. Dev. 19.3
  - Agriculture Value-added, percent of GDP — Obs. 627; Mean 15.6; Max. 62.4; Min. 0.3; Std. Dev. 14.8
  - Non-resource Trade Openness, percent of GDP — Obs. 655; Mean 41.9; Max. 875.6; Min. 0.0; Std. Dev. 68.2
  - GDP per capita, 2000 USD — Obs. 637; Mean 11453.7; Max. 70928.2; Min. 411.0; Std. Dev. 15398.3
  - Inflation, in percent — Obs. 618; Mean 0.1; Max. 3.9; Min. -0.2; Std. Dev. 0.3
  - Foreign Debt, percent of GDP — Obs. 629; Mean 57.8; Max. 281.4; Min. 0.3; Std. Dev. 50.4
  - ICRG Corruption Score — Obs. 642; Mean 2.5; Max. 6.0; Min. 1.0; Std. Dev. 0.9
  - Total ODA, percent of GDP — Obs. 633; Mean 2.7; Max. 41.7; Min. 0.0; Std. Dev. 4.5

### Main conclusions and policy implications
- Core empirical conclusion:
  - There is a statistically significant negative relationship between resource revenues and domestic (non-resource) revenues across a range of specifications.
  - Representative quantified results:
    - OLS contemporaneous effect: total non-resource tax revenues decline by 0.43 percentage points of GDP per percentage point increase in resource revenues.
    - System-GMM long-run effect θ for total tax revenue: -0.4008*** (0.1437).
    - One standard deviation increase in resource revenues implies a non-resource revenue loss of 1.6 percentage points of GDP (OLS contemporaneous) and 1.8 percentage points of GDP (GMM long-run).
- Disaggregated impacts:
  - Largest offsets concentrated in taxes on goods and services and VAT.
  - Smaller offsets for CIT and PIT (PIT less robust in some specifications).
  - Smallest offsets for trade taxes.
- Policy priorities suggested:
  - Strengthen tax policy and tax administration to preserve or rebuild non-resource tax bases (especially VAT and consumption taxes).
  - Design and implement fiscal frameworks and revenue-stabilizing rules to mitigate resource revenue volatility transmission to budgets.
  - Diversify tax bases and increase non-resource tax effort to enhance inter-temporal fiscal sustainability, especially where resource endowments may be depleted.
  - Improve institutional quality to reduce rent-seeking and discretionary application of tax law.
- Risks and considerations:
  - Greater dependence on volatile resource revenues increases procyclicality in fiscal policy and may retard institutional development.
  - The "optimal" offset between resource and non-resource revenue depends on trade-offs between efficiency and equity and is beyond the scope of the paper.

*Source: _wp1405 - 1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors (excerpts from the supplied IMF PDF content).*

### 1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors ..........................13

### 1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors

### Introduction and motivation
- Question: How do government revenues from natural resources affect domestic non-resource tax revenues and their components (taxes on goods and services, VAT, corporate income tax (CIT), personal income tax (PIT), and trade taxes)?
- Rationale:
  - Resource-rich countries may face incentives to keep a relatively low domestic (non-resource) tax burden to promote private sector activity and growth.
  - Countervailing literature argues higher domestic tax effort improves public service quality and trust in government through an implicit fiscal contract between taxpayers and the state (references in source).
  - Resource revenues can be volatile and, absent appropriate fiscal frameworks and revenue-stabilizing rules, volatility transmits to the budget.
  - Heavy reliance on resource revenues is associated with weakened tax administration capacity, ad hoc tax exemptions, and discretionary application of tax laws.
- Empirical gap addressed: No prior study had assessed the impact of resource revenues on different types of non-resource domestic taxes; this paper constructs a disaggregated database for 35 resource-rich countries (1992–2009) to fill this gap.

### Data and sample
- Sample size and period:
  - Unbalanced panel of 35 natural resource-rich countries for the period 1992–2009.
- Revenue definitions and construction:
  - Dependent variables: domestic non-resource tax revenues (relative to GDP) disaggregated into TAX (total non-resource tax revenues), taxes on goods and services (G&S), VAT, corporate income tax (CIT), personal income tax (PIT), and taxes on international transactions (TRADE). Non-resource CIT excludes tax on profits paid by resource companies.
  - Resource revenues: government revenues from taxing extraction companies, taxing exports of those natural resources, royalties, and production sharing agreements.
  - Data sources: Government Financial Statistics (GFS) supplemented with IMF country desk detailed revenue information.
- Key aggregate statistics highlighted:
  - Resource revenues increased from 13.8 percent of GDP, on average, during the 1990s to 16.1 percent of GDP, on average, during the 2000s.
  - The share of resource revenues in total revenues has increased over time.

### Empirical specification
- Estimating equation (as presented in the source):
  - T_it = α_i + β_1 R_it + β_2 X_it + μ_t + ε_it
    - T = domestic non-resource tax revenues (relative to GDP) — estimated separately for TAX, G&S, VAT, CIT, PIT, TRADE.
    - R = government revenues from natural resources (relative to GDP).
    - α_i = country fixed effect; μ_t = time fixed effect; ε_it = error term.
    - β_1 measures the marginal effect of an additional percentage point of government resource revenues on non-resource revenue effort.
- Control variables (X_it) included, drawn from previous tax effort and tax ratio literature:
  - log of GDP per capita (proxy for development)
  - share of agriculture in value-added
  - non-resource trade openness (non-resource exports + imports relative to GDP)
  - inflation
  - external indebtedness (level)
  - quality of institutions proxied by the ICRG corruption index (values from 0 (high corruption) to 6 (low corruption))
  - foreign aid (ODA) relative to GDP

### Main empirical findings (summary)
- Overall offset effect:
  - Estimated offset of about 30 percent between natural resource revenues and total domestic (non-resource) revenue (i.e., an additional percentage point of resource revenues is associated with roughly a 0.30 percentage point decline in non-resource tax revenues).
- Differential effects by tax type:
  - Large and robust negative impact of natural resource revenues on taxes on goods and services — in particular on the VAT.
  - More modest, though still negative and significant, impacts on corporate income tax (CIT) and trade taxes.
  - Results for PIT are estimated but not summarized in the supplied excerpt as being large or small relative to others.
- Robustness:
  - Results are robust to inclusion of control variables, exclusion of outliers, alternative estimation methodologies, and approaches addressing endogeneity of resource revenue.
- Supporting descriptive evidence:
  - Data plots and pairwise correlations indicate a negative association between resource and non-resource domestic revenues; for 27 countries the correlation between the two variables is negative.
  - Prior studies cited: Bornhorst, Gupta and Thornton (2009) found an offset of about 20 percent for 30 hydrocarbon-producing countries; this paper finds a somewhat larger offset (~30 percent) for the constructed 35-country panel.

### Methodological notes and data challenges
- Challenges in disaggregation:
  - Disaggregating resource from non-resource revenues and further disaggregating non-resource taxes into components is challenging due to how taxes are compiled and reported (e.g., corporate income tax receipts include resource company contributions; trade taxes may include commodity export revenues).
  - The authors adjusted revenue series (for example, excluding resource company profits from CIT) using GFS and IMF country desk information to construct non-resource series.
- Fiscal and policy context considerations emphasized:
  - Importance of well-diversified tax base as resource endowments may be exhausted in the foreseeable future; non-resource taxation can provide sustainable revenue streams.
  - Need for revenue-stabilizing fiscal rules and stronger institutions to mitigate volatility and avoid deterioration of tax administration capacity.

### Policy relevance and implications (as framed in the source)
- Implication: Heavy reliance on resource revenues is associated with lower non-resource tax effort overall, with the most pronounced reductions in VAT and other taxes on goods and services.
- Policy priorities suggested by the analysis:
  - Strengthen tax policy and tax administration to preserve or rebuild non-resource tax bases (especially VAT and consumption taxes).
  - Design and implement fiscal frameworks and revenue-stabilizing rules to mitigate resource revenue volatility transmission to budgets.
  - Diversify tax bases and increase non-resource tax effort to enhance inter-temporal fiscal sustainability, especially where resource endowments may be depleted.
  - Improve institutional quality to reduce rent-seeking and discretionary application of tax law that can accompany resource dependence.

*Source: _wp1405 - 1. Panel OLS Results with Fixed Country and Time Effects, Robust Errors (excerpts from the supplied IMF PDF content).*

### 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average,

### _wp1405 - 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average,

### Tax revenue trends and composition
- Total tax revenue increased from 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average, during the 2000s.
- The largest contributors to this increase were taxes on goods and services and income taxes.
- Tax revenue from taxes on international trade declined as a result of trade liberalization and a reduction in trade tariffs.
- Figures reported (authors’ calculations; IMF Fiscal Affairs Department Revenue Database) indicate:
  - Total revenue and resource revenue by period (1992-99 vs 2000-09) and tax-component shares (Other, Trade, PIT, CIT, G&S) shown in Figures 4 and 5 (visuals referenced in source).

### VAT and corporate income tax design across regions
- Evidence on VAT:
  - Panel A (Figure 6) reports average standard VAT rates by region; Panel B (Figure 6) reports average number of reduced VAT rates by region.
  - VAT rates are lower in resource-rich countries in some regions (notably Middle-East), but evidence on reduced VAT rates is mixed.
  - The number of VAT rates is larger in resource-rich countries.
- Evidence on corporate income tax (Figure 7):
  - Average corporate income tax rates on non-resource activities tend to be lower in resource-rich countries.

### Main empirical results (Panel OLS with country and time fixed effects)
- Estimation: Eq. (1) estimated by ordinary least squares with country and time fixed-effects for total tax revenue (Tax) and components: G&S, VAT, CIT, PIT, Trade. Full control set and constant included.
- Key finding on resource revenue offsets:
  - For each additional percentage point of revenue from natural resources, there is a decline in total non-resource tax revenues by 0.43 percentage points of GDP (Column 1).
  - An increase in resource revenues by one standard deviation would imply a non-resource revenue loss of 1.6 percentage points of GDP (one standard deviation of resource revenues is 3.93; standard deviations calculated for each country separately).
  - Average annual fluctuation in resource revenues is associated with an estimated annual fluctuation in non-resource revenue of about 0.8 percentage point of GDP.
- Component-specific OLS offsets (Table 1):
  - Taxes on goods and services (G&S): offset of 0.12 percentage points of GDP per percentage point of resource revenues (Column 2).
  - VAT: offset of 0.11 percentage points of GDP per percentage point of resource revenues (Column 3).
  - Corporate income tax (CIT): offset of about 0.04 percentage points of GDP (Column 4).
  - Personal income tax (PIT): offset of about 0.06 percentage points of GDP (Column 5).
  - Trade taxes: offset of about 0.03 percentage points of GDP (Column 6).
- Table 1 exact coefficient estimates for Resource Revenue (standard errors in parenthesis):
  - Tax (1): -0.4286*** (0.0799)
  - G&S (2): -0.1238*** (0.0359)
  - VAT (3): -0.1154*** (0.0379)
  - CIT (4): -0.0392*** (0.0075)
  - PIT (5): -0.0604*** (0.0225)
  - Trade (6): -0.0284*** (0.0108)
- Additional control variable relationships (selected, from Table 1):
  - Agriculture (share of GDP) negative when significant; e.g., Tax: -0.2240** (0.1131).
  - Log(GDP per capita) positive for some taxes; e.g., VAT: 2.9318*** (0.9944).
  - Log(Inflation) positive for Tax: 1.8038* (0.9431).
  - ODA (share of GDP) negative association with tax effort for some specifications; e.g., CIT: -0.1192*** (0.0326).
  - No evidence of an association between level of corruption and tax revenue effort in the baseline OLS.

### Endogeneity and dynamic concerns; System-GMM estimation
- Concerns:
  - Potential endogeneity of natural resource revenues (e.g., fiscal deficits and policy responses).
  - Persistence in tax revenue data raises possibility of serial correlation.
- Estimation approach:
  - System generalized method of moments (system-GMM) used to address endogeneity and dynamic effects (lagged dependent variable included).
  - Diagnostics: Hansen and Sargan statistics reported; Arellano-Bond tests (M1 and M2) suggest first-order serial correlation present but second-order serial correlation absent in most specifications.
- System-GMM results (Table 2) — contemporaneous Resource Revenue coefficients (standard errors in parenthesis):
  - Tax (1): -0.2917*** (0.1215)
  - G&S (2): -0.0497** (0.0227)
  - VAT (3): -0.0620*** (0.0202)
  - CIT (4): -0.0172* (0.0098)
  - PIT (5): -0.0007 (0.0099) [statistically insignificant]
  - Trade (6): -0.0091* (0.0059)
- Long-run (ultimate) offset effects θ reported in Table 2:
  - Tax: θ = -0.4008*** (0.1437)
  - G&S: θ = -0.4131*** (0.1658)
  - VAT: θ = -0.2102*** (0.0572)
  - CIT: θ = -0.0550* (0.0335)
  - PIT: θ = -0.0102 (0.1457)
  - Trade: θ = -0.0801 (0.0761)
- Interpretation:
  - With system-GMM, the negative effect of natural resource revenues on total domestic tax effort and most main tax sources remains significant and of similar order.
  - Considering the long-run θ estimates, an increase in resource revenues by one standard deviation would imply an ultimate non-resource revenue loss of 1.8 percentage points of GDP.

### Robustness: hydrocarbon producers subset
- Subsample: estimation of Eq.(1) for hydrocarbon (oil and gas) producer countries only.
- Table 3 key contemporaneous Resource Revenue coefficients (standard errors in parenthesis):
  - Tax (1): -0.1509*** (0.0582)
  - G&S (2): -0.0312* (0.0182)
  - VAT (3): -0.0526*** (0.0186)
  - CIT (4): -0.0209* (0.0112)
  - PIT (5): -0.0009 (0.0076) [statistically insignificant]
  - Trade (6): -0.0091** (0.0045)
- Long-run θ estimates for hydrocarbon producers (Table 3):
  - Tax: θ = -0.4359** (0.1976)
  - G&S: θ = -0.3929* (0.2141)
  - VAT: θ = -0.1818*** (0.0536)
  - CIT: θ = -0.0695** (0.0311)
  - PIT: θ = -0.0198 (0.1638)
  - Trade: θ = -0.0861 (0.5412)
- Interpretation:
  - Results qualitatively similar to full sample; contemporaneous offset effect on total non-resource revenues appears smaller (e.g., -0.1509***), consistent with earlier literature for hydrocarbon producers, but long-run offsets remain economically significant (e.g., θ = -0.4359**).

### Analytical conclusions and implications
- Robust empirical evidence that higher resource revenue shares are associated with significant offsets in domestic non-resource tax revenues across a range of specifications (OLS fixed-effects and system-GMM).
- The largest offsets are concentrated in taxes on goods and services and VAT, with smaller but significant offsets for CIT and PIT (PIT less robust in some specifications), and the smallest offsets for trade taxes.
- The magnitude of revenue offsets is economically meaningful:
  - OLS contemporaneous effect: total non-resource tax revenues decline by 0.43 percentage points of GDP per percentage point increase in resource revenues.
  - System-GMM long-run effect θ for total tax revenue: -0.4008***.
  - One standard deviation increase in resource revenues implies a non-resource revenue loss of 1.6 percentage points of GDP (OLS contemporaneous) and 1.8 percentage points of GDP (GMM long-run).
- Control variables confirm expected relationships: agriculture share tends to reduce tax revenues; higher GDP per capita tends to increase tax revenues; foreign aid (ODA) negatively associated with tax effort in line with previous literature.

*Italic: Source: _wp1405 - 13.6 percent of GDP, on average, during the 1990s to 15.3 percent of GDP, on average,*

### Appendix I describes the countries included in the sample.

### _wp1405 - Appendix I describes the countries included in the sample.

### Robustness tests and outliers
- Outliers identified using Hadi (1994): Russia and Norway; Sierra Leone and Senegal also considered outliers (resources contributed only marginally to GDP during the sample period due to civil war and lack of investment in resource sectors).  
- Estimates without outliers: "not significantly different from the full sample estimates."  
- Additional robustness: excluding countries with share of resource revenues to GDP below 20 percent (Table 4 sample labelled "Low Resource Dependence").  
  - Resource Revenue coefficients (Low Resource Dependence, Table 4):  
    - Tax: -0.2690** (std. error 0.1355)  
    - G&S: -0.0535* (0.0306)  
    - VAT: -0.0731* (0.0464)  
    - CIT: -0.0360* (0.0217)  
    - PIT: -0.0297* (0.0174)  
    - Trade: -0.0259*** (0.0097)  
  - Dependent variable (lagged) coefficients (Low Resource Dependence, Table 4):  
    - Tax: 0.3082** (0.1581)  
    - G&S: 0.8477*** (0.0591)  
    - VAT: 0.8115*** (0.1156)  
    - CIT: 0.6738*** (0.0778)  
    - PIT: 0.9159*** (0.0502)  
    - Trade: 0.9242*** (0.0836)  
  - θ coefficients (Low Resource Dependence, Table 4):  
    - Tax: -0.3888** (0.1574)  
    - G&S: -0.3512** (0.1820)  
    - VAT: -0.3879** (0.2056)  
    - CIT: -0.1103* (0.0785)  
    - PIT: -0.3541*** (0.1109)  
    - Trade: -0.3424** (0.2213)  
  - Model diagnostics (Low Resource Dependence, Table 4):  
    - Observations: 307, 318, 75, 176, 269, 251 (by column)  
    - No. of countries: 25, 25, 12, 17, 23, 20 (by column)

### Heterogeneity by domestic (non‑resource) tax effort
- Sample split around average domestic non-resource tax revenue effort (~15 percent of GDP) (Table 5).  
- Low Tax Effort (Table 5, selected coefficients):  
  - Resource Revenue: -0.1734*** (0.0654) for Tax; -0.0551*** (0.0211) for G&S; -0.0267** (0.0119) for VAT; -0.0093* (0.0056) for CIT; 0.0176 (0.0172) for PIT; -0.0081 (0.0066) for Trade.  
  - Dependent variable (lagged): 0.2269*** (0.0940) to 0.9390*** (0.0894) across series.  
  - θ: -0.2242*** (0.0675) to -0.1330 (0.2326) across series.  
  - Observations: 228, 211, 88, 131, 149, 181; No. of countries: 20, 17, 10, 14, 13, 16.  
- High Tax Effort (Table 5, selected coefficients):  
  - Resource Revenue: -0.3581*** (0.1373) for Tax; -0.0708*** (0.0239) for G&S; -0.0410*** (0.0134) for VAT; -0.0314 (0.0213) for CIT; -0.0177** (0.0088) for PIT; -0.0068 (0.0189) for Trade.  
  - Dependent variable (lagged): 0.2508* (0.1618) to 0.9047*** (0.0873) across series.  
  - θ: -0.4781*** (0.1203) to -0.0421 (0.1321) across series.  
  - Observations: 174, 180, 112, 99, 169, 134; No. of countries: 15, 15, 11, 10, 15, 12.  
- Interpretation: countries with relatively high non-resource domestic tax revenue effort show larger contemporaneous offsets; those with lower tax effort show lower contemporaneous offsets and lower computed ultimate effects.

### Normalization by resource- and non‑resource GDP (Table 6)
- Dependent variables and revenue variables normalized by non-resource GDP and resource GDP respectively.  
- Key result: "for each additional percentage point of natural resource revenues, there is a decline in total non resource tax revenue by 0.0042 percentage points of non resource GDP."  
- Table 6 coefficients (Resource Revenue normalized):  
  - Tax: -0.0042* (0.0016)  
  - G&S: -0.0021* (0.0006)  
  - VAT: -0.0012*** (0.0004)  
  - CIT: -0.0014* (0.0006)  
  - PIT: -0.0002 (0.0003)  
  - Trade: -0.0142 (0.0121)  
- Dependent variable (lagged) coefficients (Table 6): 0.6943*** (0.1330) to 0.9617*** (0.0386).  
- θ coefficients (Table 6): -0.0137** (0.0089) to -0.0576 (0.0651) across series.  
- Quantitative interpretation provided in text:  
  - "An increase in resource revenues (in percent of resource GDP) by one standard deviation implies a contemporaneous non resource revenue loss of about 1 percentage points of non resource GDP."  
  - "One standard deviation of resource revenues is now about its average change for a period of 18 months. That is the average annual fluctuation in resource revenues is associated with an estimated annual fluctuation in non resource revenue of about 0.7 percentage points of non resource GDP."  
  - "Its ultimate offset effect is estimated at about 1.6 percentage points of non resource GDP."  
- Table 6 model diagnostics: Observations 402, 391, 200, 230, 318, 315; No. of countries 35, 32, 21, 24, 28, 28.

### Additional robustness checks (Appendix II)
- Nonlinear effects (resource revenue squared): Table A2.1.  
  - Resource Revenue coefficients remain negative and significant for many specifications (e.g., Tax: -0.2814** (0.1133)).  
  - Resource Revenue (squared) coefficients not statistically significant in regressions (e.g., Tax: 0.0273 (0.3698); VAT: 0.0009 (0.0016)).  
  - Estimated ultimate effects reported as "very high," indicating larger long-run offsets for larger shares of resource revenues despite insignificant squared term contemporaneously.  
  - Table A2.1 diagnostics: Observations 401, 391, 200, 230, 318, 315; No. of countries 35, 32, 21, 24, 28, 28.  
- Alternative proxy: total natural resource rents (World Bank WDI) used as resource revenue proxy (Table A2.2).  
  - Resource Rents coefficients (Table A2.2):  
    - Tax: -0.1822** (0.0814)  
    - G&S: -0.0230* (0.0144)  
    - VAT: -0.0173** (0.0079)  
    - CIT: -0.0076* (0.0047)  
    - PIT: -0.0007 (0.0046)  
    - Trade: -0.0027 (0.0062)  
  - Findings qualitatively similar to Section III: significant negative effect of resource rents on total domestic tax revenue and main tax sources; largest offset on taxes on goods and services and VAT; smaller offset on CIT; offset on trade taxes statistically insignificant.  
  - Table A2.2 diagnostics: Observations 397, 429, 222, 255, 360, 366; No. of countries 34, 32, 23, 24, 28, 28.

### Main empirical findings and interpretation (Concluding Remarks)
- Sample: newly constructed revenue database for 35 resource-rich countries.  
- Core result: "there is a statistically significant negative relationship between resource revenues and domestic (non resource) revenues."  
  - Quantified: "For each additional percentage point of GDP in resource revenue, there is a corresponding reduction in domestic (non resource) revenues of about 0.3 percentage points of GDP."  
- Disaggregated results: largest impacts on taxes on goods and services (in particular VAT); smaller impacts on corporate and trade taxes.  
- Robustness: results robust to different model specifications, endogeneity and serial correlation controls, alternative definitions of resource revenues, and a subsample of hydrocarbon producing countries.  
- Policy implications highlighted:  
  - Monitor evolution of non-resource revenues; if offsets are large, examine tax design and strengthen revenue administration.  
  - Assess VAT design and administration: resource-rich countries tend to have lower VAT rates, higher-than-usual exemptions, and a larger number of rates — all potentially lowering VAT efficiency.  
  - Assess income taxation, particularly if the tax base narrows with expanding resource revenue dependence.  
- Risks noted: greater dependence on volatile resource revenues increases procyclicality in fiscal policy, may retard institutional development, and could affect fiscal sustainability if public spending is scaled up without a sound fiscal framework.  
- Caveat: the "optimal" offset between resource and non-resource revenue depends on trade-offs between efficiency and equity and is beyond the scope of the paper.

### Data and sample (Appendix I)
- Countries in the sample (hydrocarbon producing countries identified with an asterisk): Algeria*, Angola*, Argentina*, Bahrain*, Botswana*, Brunei*, Cameroon*, Colombia*, Congo Republic*, Equatorial Guinea*, Gabon*, Gambia, Ghana, Guinea, Guinea-Bissau, Guyana, Indonesia*, Iran*, Ivory Coast*, Kazakhstan*, Mali, Mexico*, Nigeria*, Norway*, Oman*, Peru, Russia*, Saudi Arabia*, Senegal, Sierra Leone, Syria*, Trinidad and Tobago*, United Arab Emirates*, Vietnam*, and Yemen*.  
- Data sources: IMF Government Financial Statistics (GFS) database; IMF country documents; IMF Fiscal Affairs Department Database on Natural Resources Taxation; World Bank World Development Indicators (WDI); IMF International Financial Statistics (IFS); OCDE (Development Co-operation Directorate) for Net total ODA.  
- Variable definitions summarized in text (e.g., ICRG corruption scores range from 0 to 6; 0 = highest potential risk of corruption, 6 = lowest potential risk).  
- Table A1.1 Descriptive statistics (preserve figures exactly):  
  - Total Tax Revenue, percent of GDP — Obs. 527; Mean 14.8; Max. 54.4; Min. 0.4; Std. Dev. 8.0  
  - Tax Revenue from Goods and Services, percent of GDP — Obs. 552; Mean 5.0; Max. 15.8; Min. 0.0; Std. Dev. 3.6  
  - VAT Tax Revenue, percent of GDP — Obs. 269; Mean 4.3; Max. 9.9; Min. 0.0; Std. Dev. 2.1  
  - Corporate Income Tax Revenue, percent of GDP — Obs. 361; Mean 1.8; Max. 9.2; Min. 0.0; Std. Dev. 1.4  
  - Personal Income Tax Revenue, percent of GDP — Obs. 440; Mean 1.9; Max. 11.5; Min. 0.0; Std. Dev. 2.1  
  - Trade Tax Revenue, percent of GDP — Obs. 467; Mean 1.9; Max. 7.0; Min. 0.1; Std. Dev. 1.3  
  - Resource Revenue, percent of GDP — Obs. 551; Mean 15.2; Max. 68.7; Min. 0.0; Std. Dev. 14.1  
  - Natural Resource Rents, percent of GDP — Obs. 665; Mean 21.6; Max. 80.7; Min. 0.1; Std. Dev. 19.3  
  - Agriculture Value-added, percent of GDP — Obs. 627; Mean 15.6; Max. 62.4; Min. 0.3; Std. Dev. 14.8  
  - Non-resource Trade Openness, percent of GDP — Obs. 655; Mean 41.9; Max. 875.6; Min. 0.0; Std. Dev. 68.2  
  - GDP per capita, 2000 USD — Obs. 637; Mean 11453.7; Max. 70928.2; Min. 411.0; Std. Dev. 15398.3  
  - Inflation, in percent — Obs. 618; Mean 0.1; Max. 3.9; Min. -0.2; Std. Dev. 0.3  
  - Foreign Debt, percent of GDP — Obs. 629; Mean 57.8; Max. 281.4; Min. 0.3; Std. Dev. 50.4  
  - ICRG Corruption Score — Obs. 642; Mean 2.5; Max. 6.0; Min. 1.0; Std. Dev. 0.9  
  - Total ODA, percent of GDP — Obs. 633; Mean 2.7; Max. 41.7; Min. 0.0; Std. Dev. 4.5

*Source: _wp1405 - Appendix I describes the countries included in the sample.*

### REFERENCES

### REFERENCES

### Cited works
- Acosta-Ormaechea, S., and J. Yoo, 2012, “Tax Composition and Economic Growth,” IMF Working Paper 12/257, (Washington: International Monetary Fund).
- Arellano, M. and S. Bond, 1991, “Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations,” The Review of Economic Studies, 58, pp. 277-297.
- Arnold, J., B. Brys, C. Heady, A. Johansson, C. Schwellnus, and L. Vartia, 2011, “Tax Policy for Economic Recovery and Growth,” Economic Journal, Vol.111, pp. 59-80.
- Baltagi, B., 2008, “Econometric Analysis of Panel Data,” Wiley, 4th Edition, 2008.
- Baunsgaard, T., and M. Keen, 2010, “Tax Revenue and (or?) Trade Liberalization,” Journal of Public Economics, Elsevier, Vol. 94(9–10), pp. 563–77, October.
- Benedek, D., E. Crivelli, S. Gupta, and P. Muthoora, 2013, “Foreign aid and revenue: Still a crowding out effect? FinanzArchiv (forthcoming).
- Bergman, M., 2002, “Who pays for social policy? A study on taxes and trust,” Journal of Social Policy 31 (2), pp. 289-305.
- Blundell, R., and S. Bond, 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models,” Journal of Econometrics, 87, pp. 115–43.
- Bornhost, F., S. Gupta, and J. Thornton, 2009, “Natural Resource Endowments and the Domestic Revenue Effort,” European Journal of Political Economy, Elsevier, Vol. 25(4), pp. 439–46, December.
- Chaudhry, K., 1997, “The Price of Wealth: Economies and Institutions in the Middle East,” Cornell University Press, Ithaca.
- Chow, C., 1960, “Tests of Equality Between Sets of Coefficients in Two Linear Regressions,” Econometrica, 28, pp. 591–605.
- Collier, P., 2006, Is aid oil? An analysis of whether Africa can absorb more aid,” World Development 34, pp. 1482-97.
- Crandall, W., and J. Bodin, 2005, “Revenue Administration Reform in Middle Eastern Countries, 1994-2004,” IMF Working Paper 05/203 (Washington: International Monetary Fund).
- Crivelli, E., 2013, “Fiscal impact of privatization revisited: The role of tax revenues in transition economies,” Economic Systems 37, 217-32.
- Davis, G., and J. Tilton, 2005, “The resource curse,” Natural Resources Forum 29, pp. 233-42.
- Ghura, D., 1998, “Tax Revenue in Sub-Saharan Africa: Effects of Economic Policies and Corruption,” IMF Working Paper 98/135 (Washington: International Monetary Fund).
- Gupta, S and S. Tareq, 2008, “Mobilizing Revenue,” Finance and Development, Vol. 45, Issue 3 (Washington: International Monetary Fund).
- Hadi, A., 1994, “A modification of a method for the detection of outliers in multivariate samples,” Journal of the Royal Statistical Society (Series B) 56, 393-96.
- International Monetary Fund, 2011, “Revenue Mobilization in Developing Countries,” Policy Paper, Fiscal Affairs Department, March (Washington: International Monetary Fund)
- International Monetary Fund, 2012, “Fiscal Regimes for Extractive Industries: Design and Implementation,” Board Paper, August (Washington: International Monetary Fund)
- Keen, M., and M. Mansour, 2010, “Revenue Mobilisation in Sub-Saharan Africa: Challenges from Globalisation II - Corporate Taxation,” Development Policy Review, Overseas Development Institute, Vol. 28(5), pp. 573–96, 09.
- Knack, S., 2009, “Sovereign rents and quality of tax policy and administration,” Journal of Comparative Economics 37, pp. 359-71.
- Leite, C., and J. Weidmann, 1999, “Does mother nature corrupt? Natural resources, corruption, and economic growth.” In: Abed, G. and S. Gupta (Eds), Governance, Corruption, and Economic Performance, (Washington: International Monetary Fund).
- Levi, M., 1988, “Of Rule and Revenue,” University of California Press, Berkeley.
- McGuirk, E., 2013, “The illusory leader: natural resources, taxation and accountability,” Public Choice 154, pp. 285-313.
- Moore, M., 1998, “Death without taxes: democracy, state capacity and aid dependence in the fourth world.” In: Robinson, M., White, G. (Eds.), The Democratic Development State, Oxford University Press, Oxford.
- Moore, M., 2007, “How does taxation affect the quality of governance,” IDS Working Paper 280, Brighton, UK.
- Nickell, S., 1981, “Bias in Dynamic Models with Fixed Effects,” Econometrica, 49(6), pp. 1417–26.
- Ossowski, R., and A. Gonzales, 2012, “Manna from Heaven: The Impact of Nonrenewable Resource Revenues on Other Revenues of Resource Exporters in Latin America and the Caribbean,” IDB Working Paper 337, (Washington: Inter-American Development Bank).
- Pessino, C. and R. Fenochietto, 2010, “Determining Countries’ Tax Effort,” Hacienda Pública Española/Revista de Economía Pública, Vol. 195, No. 4, pp. 65–87.
- Rodrik, D., 1998, “Why Do More Open Economies have Bigger Governments? Journal of Political Economy, Vol. 106, pp. 997–1032.
- Roodman, D., 2009, “A Note on the Theme of Too Many Instruments,” Oxford Bulletin of Economics and Statistics 71(1), pp.135–58.
- Ross, M., 1999, “The political economy of the resource curse,” World Politics 51, pp. 297-322.
- Ross, M., 2001, “Does oil hinders democracy?” World Politics 53, pp. 325-61.
- Roy, S., 1957, “Some Aspects of Multivariate Analysis,” Wiley.
- Schneider, F., and D. Enste, 2000, “Shadow Economies: Size, Causes and Consequences,” Journal of Economic Literature 38, pp. 73-110.
- Sen Gupta, A., 2007, “Determinants of Tax Revenue Efforts in Developing Countries,” IMF Working Paper 07/184 (Washington: International Monetary Fund).
- Smith, R., 2001, “Estimation and Inference with Non-stationary Panel Time-series Data,” Paper prepared for the RC33 Conference, Cologne October 2000, Department of Economics, Birkbeck College, London.
- Tanzi, V., and H. Davoodi, 1997, “Corruption, Growth, and Public Finances,” In: Abed, G. and S. Gupta (Eds), Governance, Corruption, and Economic Performance, (Washington: International Monetary Fund).
- Thomas, A., and J. Treviño, 2013, “Resource Dependence and Fiscal Effort in Sub-Saharan Africa,” IMF Working Paper 13/188 (Washington: International Monetary Fund).
- Treisman, D., 2007, “What have we learned about the causes of corruption from ten years of cross-national empirical research?” Annual Review of Political Science 10, pp. 211-44.
- World Bank, 2011, “The changing wealth of nations: Measuring sustainable development in the new millennium,” Environment and Development (Washington, D.C: The World Bank).
- Zellner, A., “An Efficient Method of Estimating Seemingly Unrelated Regression and Tests for Aggregation Bias,” Journal of the American Statistical Association, Vol. 57.

*Source: _wp1405 - REFERENCES (PDF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1405.pdf_