## 1. Net ODA Grants and Loans and Total Tax Revenues

## Source details

**Canonical URL:** [1. Net ODA Grants and Loans and Total Tax Revenues](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12186.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12186.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12186.pdf.json)

---

### Key findings and summary results
- Net Official Development Assistance (ODA) to developing countries averaged between 3.7 and 6.7 percent of GDP during 1980–2009, amounting to around 20–40 percent of average tax revenues.
- Visual and aggregate patterns:
  - Data plot suggests a negative association between total net ODA and total tax revenues: between 1980 and 1995, rising foreign aid as a share of GDP accompanied a slight decrease in average tax revenue as a percentage of GDP; post-1995, a decline in the share of total net ODA to GDP was accompanied by higher tax revenues as percentage of GDP.
- Composition matters:
  - ODA provided in the form of grants is associated with lower total revenue.
  - ODA provided in the form of loans is not associated with lower total revenue (loans have no significant or positive effect in prior work).
- Disaggregated tax effects:
  - Net ODA (and ODA grants) is negatively related to VAT, excise, and income tax revenues.
  - Net ODA is positively associated with trade taxes, as increased aid facilitates higher imports.
- Heterogeneity:
  - Negative association between aid and tax revenues is principally observed in low-income countries.
  - Regional patterns for Africa, Asia-Pacific, and Europe are similar to total-sample results.
  - Strong negative impact of aid on revenues is found in highly corrupt countries (institutional quality matters).
- Corroboration:
  - Results corroborate Gupta et al. (2004) but with a smaller estimated association, reflecting strengthened revenue systems and reforms.

### Policy implications and context
- Domestic resource mobilization is essential for inclusive, sustainable, and resilient growth; G20 Seoul Summit (November 2010) stressed further provision of ODA while recognizing the importance of domestic revenue mobilization.
- UNDP/UN estimates:
  - Minimum level of tax revenues needed to reach the Millennium Development Goals (MDG) is 20 percent of GDP.
  - A UN report (sample of five developing countries) estimates additional domestic revenue of 4 percent of GDP is needed to fund the achievement of the MDGs.
- Investment in social and physical infrastructure requires substantial and sustained financial resources that can only be provided by additional domestic revenues; scope exists to raise revenue by broadening tax bases, improving compliance, and strengthening administrative systems.

### Theoretical and empirical motivation
- Fiscal response model (Heller, 1975): aid enters government budget constraint and can influence domestic revenue–aid trade-offs, potentially substituting away from domestic taxation.
- Political-economy arguments: aid may discourage taxation by undermining development of domestic institutions that support tax administration and good governance.
- Prior empirical evidence mixed: Ghura (1998), Remmer (2004), and Gupta et al. (2004) find negative relationships; later critiques emphasize sensitivity to sample composition, endogeneity, and tax-type heterogeneity.

### Methodology
- Dependent variable: TAX/GDP modeled in logs (aggregate and by tax components).
- Key regressors: Total net ODA; disaggregated into ODA_GRANTS and ODA_LOANS (both relative to GDP); squared terms included for non-linear effects.
- Controls X include:
  - GDP per capita (expected positive), share of agriculture in value-added (expected negative), industry share in value-added (expected positive), trade openness (ambiguous), inflation, external indebtedness, ICRG corruption index (0 to 6), oil-exporter dummy.
- Estimators and identification:
  - OLS with country (α_i) and time (μ_t) fixed effects.
  - Difference-GMM and system-GMM to address endogeneity and serial correlation; system-GMM shown to have better finite-sample properties.
  - Hansen and Sargan statistics reported for GMM; Hausman tests favor fixed effects.

### Data
- Sample: 118 countries for 1980–2009.
- Tax revenue data: IMF FAD Revenue Mobilization database, GFS, IMF country reports.
- Aid data: OECD/DAC.
- Other controls: WDI, IFS, ICRG.
- Descriptive statistics (selected):
  - Total Tax Revenue, percent of GDP: Observations 2728; Mean 16.08; Maximum 61.50; Minimum 0.10; Std. Dev. 7.55.
  - VAT Tax Revenue, percent of GDP: Observations 764; Mean 4.51; Maximum 16.00; Minimum 0.00; Std. Dev. 2.72.
  - Total ODA, percent of GDP: Observations 3322; Mean 4.72; Maximum 96.30; Minimum -2.96; Std. Dev. 6.27.
  - ODA Grants, percent of GDP: Observations 3322; Mean 4.47; Maximum 130.84; Minimum 0.00; Std. Dev. 6.30.
  - ODA Loans, percent of GDP: Observations 3322; Mean 0.25; Maximum 26.38; Minimum -34.54; Std. Dev. 1.49.
  - ICRG Corruption Score: Observations 2066; Mean 2.49; Maximum 6.00; Minimum 0.00; Std. Dev. 1.02.

### Results — total tax revenues (overview and magnitudes)
- Baseline control set: agriculture share, industry share, GDP per capita, trade openness, year dummies.
- Two ODA specifications:
  - Model 1: Total net ODA.
  - Model 2: Decomposition into ODA loans and grants.
- Main empirical findings:
  - Fixed effects (Model 1, Column 1): increase in net ODA associated with statistically significant decline in total tax revenue.
    - Reported interpretation: “For each additional dollar of total net ODA, there is an offset of 10 cents through lower tax revenues.”
  - GMM estimates larger negative impacts:
    - System-GMM (Model 1): “For each additional dollar of total net ODA, there is an offset of about 20 cents through lower tax revenues (with system-GMM).”
- Selected coefficient estimates (Model 1, Table 1):
  - Fixed Effects (Column 1):
    - Tax, lagged: -0.0589
    - Total ODA: -0.0066*
    - Total ODA, squared: 0.0001
    - Agriculture share in Value-Added: -0.0080*** (0.0015)
    - Trade Openness: -0.0020** (0.0009)
    - Observations: 2589; Number of countries: 118.
  - Difference GMM (Column 2):
    - Tax, lagged: 0.5882*** (0.0862)
    - Total ODA: -0.0216* (0.0115)
    - Total ODA, squared: 0.0002* (0.0001)
    - M1 (p value): 0.782; M2 (p value): 0.254; Observations: 2363; Number of countries: 116; Number of instruments: 54.
  - System GMM (Column 3):
    - Tax, lagged: 0.1248 (0.02804)
    - Total ODA: -0.0117* (0.0071)
    - Total ODA, squared: 0.0001 (0.0001)
    - M1 (p value): 0.028; M2 (p value): 0.793; Over-identification Hansen (p value): 0.71; Observations: 2376; Number of countries: 117; Number of instruments: 57.

### Model 2 — composition of ODA matters (grants vs loans)
- Key results:
  - ODA Grants: significantly associated with lower total tax revenues in all specifications.
  - ODA Loans: not significantly associated with lower tax revenues in most specifications.
- Magnitudes (interpretations from regressions):
  - Fixed effects (Model 2): “Each additional dollar in grants is offset by about 9 cents of tax revenues.”
  - System-GMM (Model 2): offset of about 24 cents per additional dollar in grants.
- Selected coefficient estimates (Model 2, Table 1):
  - Fixed Effects (Column 4):
    - Tax, lagged: 0.6655*** (0.1476)
    - ODA Loans: 0.0001 (0.0040)
    - ODA Grants: -0.0055* (0.0032)
    - Observations: 2589; Number of countries: 118.
  - Difference GMM (Column 5):
    - ODA Loans: 0.0049 (0.0108)
    - ODA Grants: -0.0203*** (0.0067)
    - ODA Grants, squared: 0.0002*** (0.0001)
    - M1 (p value): 0.105; M2 (p value): 0.327; Over-identification Hansen (p value): 0.87; Observations: 2363; Number of instruments: 81.
  - System GMM (Column 6):
    - ODA Loans: 0.0142 (0.0176)
    - ODA Grants: -0.0151*** (0.0010)
    - ODA Grants, squared: 0.0002** (0.0001)
    - Trade Openness: -0.0065** (0.0032)
    - M1 (p value): 0.012; M2 (p value): 0.772; Over-identification Hansen (p value): 0.60; Observations: 2376; Number of instruments: 84.

### Non-linearities and interpretation
- Squared ODA terms often have opposite sign to linear terms, indicating diminishing negative impact of grants once grants exceed certain thresholds.
- Comparison to prior literature:
  - Gupta et al. (2004) reported offset of about 28 cents per additional dollar in grants.
  - This paper’s fixed-effects estimate: offset of about 9 cents per additional dollar in grants (using comparable methods).
  - GMM estimates produce higher offsets than fixed effects but generally lower than Gupta et al. (2004).

### Disaggregation by tax type (Appendix II)
- Net ODA and ODA Grants:
  - Negatively associated with VAT, excises, and income taxes in almost all specifications.
  - ODA Loans: coefficients either positive or not significant for VAT, excises, and income taxes.
- Trade taxes:
  - Total ODA is positively associated with trade tax revenue.
  - Both ODA Loans and ODA Grants are positively associated with trade taxes (interpretation: aid facilitates higher imports).
- Magnitude of offsets (system-GMM based, Appendix III Table A3.1):
  - Total Tax Revenue
    - Sample Average (Percent of GDP): 16.08
    - Estimated Coefficient (ODA Grants): -0.0151
    - Estimated Coefficient (ODA Grants, Squared): 0.0002
    - Impact: -23.96
  - VAT
    - Sample Average (Percent of GDP): 4.30
    - Estimated Coefficient (ODA Grants): -0.1139
    - Estimated Coefficient (ODA Grants, Squared): 0.0045
    - Impact: -47.04
  - Trade Taxes
    - Sample Average (Percent of GDP): 4.01
    - Estimated Coefficient (ODA Grants): 0
    - Estimated Coefficient (ODA Grants, Squared): 0.0002
    - Impact: 0.08
  - Excises
    - Sample Average (Percent of GDP): 1.90
    - Estimated Coefficient (ODA Grants): -0.0788
    - Estimated Coefficient (ODA Grants, Squared): 0.0035
    - Impact: -14.31
  - Income Taxes
    - Sample Average (Percent of GDP): 4.68
    - Estimated Coefficient (ODA Grants): 0
    - Estimated Coefficient (ODA Grants, Squared): -0.0003
    - Impact: -0.14
- Interpretation:
  - High VAT offset could reflect donor-supported outlays that are VAT-exempt or reduce VAT efficiency.
  - Donor exemptions may apply to import duties, but higher private-sector imports facilitated by ODA can dominate and raise trade-tax revenue.

### Disaggregation by income group, region, and institutional quality
- Income groups (low-, lower-middle-, upper-middle):
  - Total ODA: significantly negative relation to tax revenue for all income groups.
  - When separating loans and grants, statistically significant coefficients appear only for low-income countries.
- Regions:
  - Africa and Asia & Pacific: Total ODA and ODA Grants show negative relationship with tax revenues in both difference- and system-GMM estimators.
  - Western Hemisphere: non-linear effect on grants indicates large aid inflows can affect incentives to mobilize taxes.
  - ODA Loans: significantly negative relationship with tax revenues found only in Africa in some specifications.
- Institutional strength (ICRG corruption quartiles):
  - For countries in the bottom two quartiles (weaker institutions), ODA Grants coefficients are significantly negative.
  - Bottom quartile (25th percentile) system-GMM reported coefficients:
    - ODA Loans coefficient = -0.0417*** (standard error 0.0061).
    - ODA Grants coefficient = -0.1081** (standard error 0.0554).
  - Interpretation: in the weakest-institution countries, external grants (and in system-GMM loans) are associated with large reductions in tax revenue-to-GDP ratios — in some cases almost complete revenue offset per additional dollar of net ODA.

### Robustness checks and limitations
- Robustness checks:
  - Splits by income groups (three groups), five regional categories, and controls for institutional quality.
  - Additional controls: inflation, oil-exporter dummy, external indebtedness, corruption index — results qualitatively similar.
  - GMM diagnostics: Arellano-Bond M1 and M2, Hansen/Sargan reported; instrument counts and p values provided.
- Limitations:
  - Data limitations prevent fully modeling aid and taxation as a dynamic forward-looking process using dynamic heterogeneous panel techniques (insufficiently long time series with missing observations).
  - Difficulty separating resource-related revenues from non-resource tax revenues due to limited data; resource revenue disaggregation not pursued.

### Policy prescriptions and mechanisms to mitigate negative effects
- Strengthen domestic revenue mobilization capacity to ensure ODA yields net additional resources.
- Use of revenue benchmarks in reform programs can counteract potential negative consequences of ODA/grants on revenues:
  - Evidence: Brun et al. (2011) find IMF-supported programs have positive impact on revenue mobilization; greater reliance on structural benchmarks improves revenue performance.
  - Examples of structural benchmarks: introduction of a tax identification number, establishing a large taxpayer unit, increasing the VAT threshold.
  - Data coverage for revenue benchmarks available only for 2002–11, limiting statistical testing within the study.
- Policy priorities for donors and policymakers:
  - Prioritize tax policy design, reduce excessive exemptions and investment incentives that erode revenue (tax holidays, preferential regimes).
  - Strengthen tax administration, especially in countries with weak institutions.
  - Focus on VAT efficiency improvements and rationalizing preferential treatments (noting budgetary costs in some regions range between 0.5 to 6 percent of GDP).

### Concluding synthesis
- Main conclusions:
  - Negative relationship between ODA grants and tax revenue is robust across specifications; magnitude appears smaller than in some earlier studies, suggesting strengthened revenue mobilization in many countries.
  - Effects are stronger in low-income and weak-institution countries; composition of ODA (grants vs loans) matters.
  - Policy action — stronger domestic revenue institutions, better tax policy design, and use of revenue benchmarks — can mitigate negative offsets and improve additionality of ODA.

*Source: IMF working paper content unit "1. Net ODA Grants and Loans and Total Tax Revenues" from the provided PDF (_wp12186).*

### 1. Net ODA Grants and Loans and Total Tax Revenues .........................................................13

### 1. Net ODA Grants and Loans and Total Tax Revenues

### Key findings and summary results
- Net Official Development Assistance (ODA) to developing countries averaged between 3.7 and 6.7 percent of GDP during 1980–2009, amounting to around 20–40 percent of average tax revenues.
- A data plot suggests a negative association between total net ODA and total tax revenues: between 1980 and 1995, rising foreign aid as a share of GDP accompanied a slight decrease in average tax revenue as a percentage of GDP; post-1995, a decline in the share of total net ODA to GDP was accompanied by higher tax revenues as percentage of GDP (Figure 1).
- Composition of net ODA matters:
  - ODA provided in the form of grants is associated with lower total revenue.
  - ODA provided in the form of loans is not associated with lower total revenue (loans have no significant or positive effect in prior work).
- Disaggregated tax effects:
  - Net ODA (and ODA grants) is negatively related to VAT, excise, and income tax revenues.
  - Net ODA is positively associated with trade taxes, as increased aid facilitates higher imports.
- Heterogeneity of effects:
  - The negative association between aid and tax revenues is principally observed in low-income countries.
  - Regional results for Africa, Asia-Pacific, and Europe are similar to those found for the total sample.
  - A strong negative impact of aid on revenues is found in highly corrupt countries (institutional quality matters).
- Robustness and corroboration:
  - Overall results corroborate Gupta et al. (2004) but with a smaller estimated association, reflecting efforts to strengthen revenue systems.
  - Several robustness checks were performed: income-level splits (three groups), five regional categories, and controls for quality of institutions.

### Policy implications and context
- Domestic resource mobilization is emphasized as essential for inclusive, sustainable, and resilient growth; G20 Seoul Summit (November 2010) stressed further provision of ODA while recognizing the importance of domestic revenue mobilization.
- UNDP assessment: the minimum level of tax revenues needed to reach the Millennium Development Goals (MDG) is 20 percent of GDP.
- A UN report (sample of five developing countries) estimates additional domestic revenue of 4 percent of GDP is needed to fund the achievement of the MDGs.
- Investment in social and physical infrastructure requires substantial and sustained financial resources that can only be provided by additional domestic revenues; there is scope for raising additional revenue by broadening tax bases, improving compliance, and strengthening administrative systems (IMF, 2011).

### Theoretical and empirical motivation
- The empirical question is framed by the fiscal response model (Heller, 1975): aid enters the government budget constraint and can influence the trade-off between domestic revenue and aid, possibly leading to substitution away from domestic taxation.
- Political-economy arguments posit that aid may discourage taxation by undermining development of domestic institutions that support tax administration and good governance (Knack, 2000; Heller and Gupta, 2002; Brautigam and Knack, 2004; Moss et al., 2008).
- Prior empirical evidence is mixed:
  - Ghura (1998), Remmer (2004), and Gupta et al. (2004) find negative relationships between aid and tax ratios.
  - More recent critiques argue sensitivity to sample composition, endogeneity of aid, and failure to model differential impacts across tax types and forward-looking dynamics (Carter, 2011; Clist and Morrissey, 2011).

### Methodology
- Estimating equation (1):
  - TAX/GDP is modeled in logs (aggregate and by tax components) as a function of ODA, disaggregated into ODA_GRANTS and ODA_LOANS (both expressed relative to GDP), including squared terms to capture non-linear effects and a large set of controls X.
  - The coefficient on ODA measures the semi-elasticity of the tax revenue ratio in response to a one percentage point change in the ODA ratio.
  - Country (α_i) and time (μ_t) fixed effects are included; i = 1,...,N and t = 1,...,L.
- Control variables included (drawn from prior tax-determinant studies):
  - GDP per capita (expected positive correlation with revenue).
  - Share of agriculture in value-added (expected negative association).
  - Industry share in value-added (expected positive association).
  - Trade openness (sum of imports and exports as share of GDP) — ambiguous sign: may be positive due to ease of trade tax collection or negative due to tariff reduction from liberalization.
  - Inflation (revenue effects via unindexed tax systems and seigniorage).
  - External indebtedness (reflects need to generate revenue to service debt).
  - Quality of institutions proxied by the ICRG corruption index, which takes values from 0 (high corruption) to 6 (low corruption).
  - Dummy variable for oil exporter countries to capture the potential negative impact of natural resource revenues on domestic tax effort.
- Estimation strategies to address endogeneity and persistence:
  - Ordinary least squares with country and time fixed-effects (OLS-FE) reported.
  - Difference-GMM and system-GMM estimators employed to address endogeneity of aid and serial correlation.
  - System-GMM estimates differenced and levels equations as a system, using lagged changes as instruments in the levels equation; system-GMM shown to have better finite sample properties than difference-GMM.
  - Hansen and Sargan statistics reported for GMM estimators; Sargan used where instrument proliferation is a concern.
  - Hausman tests favor fixed effects over random effects; poolability tests support panel specification with homogeneous slope coefficients.

### Data
- Sample: 118 countries for the period 1980–2009.
- Tax revenue data: Government Financial Statistics (GFS) and IMF annual consultation reports (used to fill gaps and ensure consistency); data further disaggregated into components of tax revenues.
- Aid data: OECD/DAC.
- Figure references in source:
  - Figure 1: Average Net ODA and Tax Revenue in Low- and Middle-Income Countries, 1980–2009 (tax revenue left axis, total ODA right axis).
  - Figure 2: Average Taxes and Total Net ODA, 1980–2009 (shows negative association of total ODA with income taxes, VAT, and excise taxes; positive association with trade taxes).

### Robustness checks and limitations
- Robustness checks performed include:
  - Division of countries into three income groups.
  - Five regional categories.
  - Controls for institutional quality.
- Limitations noted:
  - Data limitations prevent fully modeling aid and taxation as a dynamic forward-looking process using dynamic heterogeneous panel techniques (insufficiently long time series with missing observations).
  - Difficulty in separating resource-related revenues from non-resource tax revenues due to limited data; hence resource revenue disaggregation was not pursued.

*Source: IMF working paper content unit "1. Net ODA Grants and Loans and Total Tax Revenues" from the provided PDF.*

### Appendix I shows pairwise correlations between tax revenue and ODA grants for all

### _wp12186 - Appendix I shows pairwise correlations between tax revenue and ODA grants for all

### Appendix I: pairwise correlations
- For about 70 percent of the countries, the correlation between tax revenue and ODA grants is negative.
- Literature notes heterogeneity by tax type (Gambaro et al., 2007; Carter, 2011).
  - Two explanations for a positive ODA–trade tax relationship:
    - If ODA facilitates higher imports and those imports are not tax-exempt, increases in ODA associate positively with trade taxes.
    - Positive ODA coefficients may reflect omitted variables, notably weak tax administration capacity in countries relying heavily on trade taxes.

### Figure 2 — Average taxes and total net ODA, 1980–2009 (percent of GDP)
- Source datasets: IMF’s FAD Revenue Mobilization database and OECD DAC database.
- The figure presents time series for:
  - Income tax, VAT, Excises, and Trade taxes (percent of GDP) alongside Total ODA (net).
- (No numeric series beyond axes labels and years are provided in the text excerpt.)

### IV. Results — Total tax revenues: overview
- Baseline model controls: share of agriculture in value-added, share of industry in value-added, GDP per capita, trade openness, and a full set of year dummies.
- Two main ODA specifications:
  - Model 1: Total net ODA (Columns 1–3).
  - Model 2: Decomposition into ODA loans and grants (Columns 4–6).
- Control variable relationships (summary):
  - Agriculture share in GDP: negatively related to revenues.
  - Industry share in value-added: positively related to revenues.
  - GDP per capita (when significant): positively related to tax revenues.
  - Trade openness: negatively related to tax revenues.

### Main empirical findings on aid and tax revenue
- Fixed effects specification (Model 1, Column 1):
  - An increase in net ODA is associated with a statistically significant decline in total tax revenue.
  - “For each additional dollar of total net ODA, there is an offset of 10 cents through lower tax revenues.”13
- Difference- and system-GMM estimators (Model 1, Columns 2–3):
  - Diagnostics: Arellano-Bond M1 and M2 tests indicate first-order serial correlation may be present while second-order is not; Hansen statistic acceptable.
  - The negative impact of ODA on tax revenue is larger under GMM:
    - “For each additional dollar of total net ODA, there is an offset of about 20 cents through lower tax revenues (with system-GMM).”

### Table 1 (Selected coefficient estimates and diagnostics)
- Model 1 (Total ODA)
  - Fixed Effects (Column 1):
    - Tax, lagged: -0.0589
    - Total ODA: -0.0066*
    - Total ODA, squared: 0.0001
    - Agriculture share in Value-Added: -0.0080*** (standard error 0.0015)
    - Trade Openness: -0.0020** (standard error 0.0009)
    - Constant: -0.6849*** (standard error 0.1661)
    - Observations: 2589
    - Number of countries: 118
  - Difference GMM (Column 2):
    - Tax, lagged: 0.5882*** (standard error 0.0862)
    - Total ODA: -0.0216* (standard error 0.0115)
    - Total ODA, squared: 0.0002* (standard error 0.0001)
    - M1 (p value): 0.782
    - M2 (p value): 0.254
    - Observations: 2363
    - Number of countries: 116
    - Number of instruments: 54
  - System GMM (Column 3):
    - Tax, lagged: 0.1248 (standard error 0.02804)
    - Total ODA: -0.0117* (standard error 0.0071)
    - Total ODA, squared: 0.0001 (standard error 0.0001)
    - M1 (p value): 0.028
    - M2 (p value): 0.793
    - Over-identification Hansen (p value): 0.71
    - Observations: 2376
    - Number of countries: 117
    - Number of instruments: 57
- Model 2 (ODA Loans and Grants)
  - Fixed Effects (Column 4):
    - Tax, lagged: 0.6655*** (standard error 0.1476)
    - ODA Loans: 0.0001 (standard error 0.0040)
    - ODA Loans, squared: -0.0000 (standard error 0.0002)
    - ODA Grants: -0.0055* (standard error 0.0032)
    - ODA Grants, squared: 0.0000 (standard error 0.0000)
    - Agriculture share in Value-Added: -0.0026 (standard error 0.0028)
    - Trade Openness: -0.0019** (standard error 0.0009)
    - Constant: -0.6997*** (standard error 0.1662)
    - Observations: 2589
    - Number of countries: 118
  - Difference GMM (Column 5):
    - ODA Loans: 0.0049 (standard error 0.0108)
    - ODA Loans, squared: -0.0018* (standard error 0.0009)
    - ODA Grants: -0.0203*** (standard error 0.0067)
    - ODA Grants, squared: 0.0002*** (standard error 0.0001)
    - M1 (p value): 0.105
    - M2 (p value): 0.327
    - Over-identification Hansen (p value): 0.87
    - Observations: 2363
    - Number of countries: 116
    - Number of instruments: 81
  - System GMM (Column 6):
    - ODA Loans: 0.0142 (standard error 0.0176)
    - ODA Loans, squared: -0.0011 (standard error 0.0010)
    - ODA Grants: -0.0151*** (standard error 0.0010)
    - ODA Grants, squared: 0.0002** (standard error 0.0001)
    - Trade Openness: -0.0065** (standard error 0.0032)
    - M1 (p value): 0.012
    - M2 (p value): 0.772
    - Over-identification Hansen (p value): 0.60
    - Observations: 2376
    - Number of countries: 117
    - Number of instruments: 84
- Note: Dependent variable is total tax revenue to GDP. Robust standard errors in parenthesis; ***(**,*) indicate significance at 1(5, 10) percent.

### Model 2 — composition of ODA matters
- Grants vs. loans:
  - ODA provided as grants is significantly associated with lower total tax revenues in all specifications.
  - ODA provided as loans is not significantly associated with lower tax revenues.
- Magnitude of offsets:
  - Fixed effects (Model 2): “Each additional dollar in grants is offset by about 9 cents of tax revenues.”
  - System-GMM (Model 2): offset of about 24 cents per additional dollar in grants.
- Non-linear effects:
  - The squared ODA variables often have the opposite sign to the linear term, indicating that as grants exceed a certain threshold, their negative impact on tax collection diminishes.

### Comparison to prior literature and interpretation
- Gupta et al. (2004) reported an offsetting effect on total revenue of about 28 cents per additional dollar in grants.
- This study’s fixed-effects estimate: offset of about 9 cents per additional dollar in grants (using similar econometric techniques), consistent with strengthened revenue mobilization efforts (e.g., tax policy and tax administration reforms).
- GMM estimates produce higher tax offsets than fixed effects, but still generally lower than Gupta et al. (2004).

### Policy implications and mechanisms to mitigate negative effects
- Establishing revenue benchmarks in reform programs may help counteract the potential negative consequences of ODA/grants on revenues.
  - Evidence cited: Brun et al. (2011) find IMF-supported programs have a positive impact on revenue mobilization; greater reliance on structural benchmarks improves revenue performance.
- Structural benchmarks in IMF-supported programs increased since 2002–03.
  - Examples of structural benchmarks: introduction of a tax identification number, establishing a large taxpayer unit, increasing the VAT threshold.
  - Data coverage limitation: available data covers 2002–11, so the statistical significance of revenue benchmarks on the tax ratio cannot be empirically tested within this study.

### Robustness checks (brief)
- Additional controls added to Models 1 and 2: inflation, dummy for oil exporters, level of external indebtedness, and corruption index.
- The results are reported as qualitatively similar to those in Table 1 (detailed coefficients not included in the provided excerpt).

*Source: _wp12186 - Appendix I shows pairwise correlations between tax revenue and ODA grants for all*

### Appendix II.

### Appendix II.

### Disaggregation by tax type
- Purpose: Decompose tax revenues into VAT, excises, income, and trade taxes to test differential relationship between ODA and taxes. Only difference- and system-GMM estimates reported; full set of control variables and year dummies included. Diagnostics: Hansen/Sargan tests tolerable; Arellano-Bond (M1, M2) consistent with no second-order serial correlation.
- Key empirical findings:
  - Net ODA has a negative relationship with VAT, excises, and income tax revenues.
  - The composition of ODA matters:
    - ODA Grants: negatively associated with VAT, excises, and income taxes in almost all specifications.
    - ODA Loans: coefficients either positive or not significant for VAT, excises, and income taxes.
  - Trade taxes behave differently:
    - Total ODA is positively associated with trade tax revenue.
    - ODA Loans and ODA Grants are both positively associated with trade taxes.
    - Interpretation: increased aid facilitates higher imports, contributing to higher trade tax revenues.
- Magnitudes and offsets:
  - Estimated coefficients for specific taxes are larger in magnitude than for total taxes.
  - Using average values for ODA grants and the relevant tax type, revenue offsets for an additional dollar of ODA grants are:
    - 47 cents for the VAT.
    - 14 cents for excises.
    - Offsets for income taxes are very small (still negative).
    - Trade taxes show a small positive relationship with aid.
  - Possible mechanisms:
    - High VAT offset could reflect exemptions from tax for donor-supported outlays.
    - Donor exemptions may also apply to import duties, but higher private-sector imports facilitated by ODA can overwhelm exemption effects.
- Context on tax policy and administration:
  - Recent estimates indicate a potential for raising additional VAT revenue in developing countries by about 2 percent of GDP, on average.
  - For excise taxes, IMF (2011) estimates potential increases in revenue by between 0.5 and 1.3 percent of GDP in developing countries, on average.
  - Low VAT efficiency and proliferation of investment incentives (tax holidays) noted; e.g., in the 1980s ~40 percent of low-income countries in sub-Saharan Africa offered tax holidays and this number doubled by 2005.
  - Budgetary cost of preferential treatments in Latin America estimated to range between 0.5 to 6 percent of GDP.

### Disaggregation by income groups
- Sample split: low-, lower-middle-, and upper-middle-income country groups (World Bank 2009 GNI per capita thresholds used for grouping).
- Key findings:
  - For total ODA, a significantly negative relation to tax revenue for all income groups.
  - When separating aid into loans and grants, statistically significant coefficients appear only for low-income countries.
  - Results are robust across difference- and system-GMM estimators (diagnostics satisfactory).

### Disaggregation by geographical regions
- Regions: Africa; Asia and the Pacific; Europe; Middle East and Central Asia; Western Hemisphere.
- Regional findings (Table 4 summary):
  - Africa and Asia & Pacific: Total ODA and ODA Grants show a negative relationship with tax revenues in both difference- and system-GMM estimators.
  - Western Hemisphere: Non-linear effect on grants indicates that relatively large aid inflows can affect incentives to mobilize taxes (non-linear/graduated effect).
  - ODA Loans: a significantly negative relationship with tax revenues is found only in Africa.
- Diagnostics: GMM estimators satisfactory; Arellano-Bond tests show no presence of serial correlation (details omitted in table to preserve space).

### Strength of country’s institutions
- Method: Split sample by ICRG ranking of corruption as proxy for institutional strength; countries grouped into quartiles. Regress total tax revenues for countries in the 50th and 25th percentiles (weakest institutions).
- Findings (Table 5 summary):
  - For countries in the bottom two quartiles (weaker institutions), ODA Grants coefficients are significantly negative.
  - Bottom quartile (25th percentile) findings:
    - System GMM: ODA Loans coefficient = -0.0417*** (standard error 0.0061).
    - System GMM: ODA Grants coefficient = -0.1081** (standard error 0.0554).
    - Interpretation: For the bottom quartile of corrupt countries, the effect of external grants is particularly strong — almost complete revenue offset on each additional dollar of net ODA. Both ODA Grants and ODA Loans associated with reductions in tax revenue-to-GDP ratio.
  - Policy implication: Policymakers and donors need to pay particular attention to strengthening revenue-raising capacity in countries with weak institutions to ensure ODA yields net additional resources.

### Concluding remarks and policy prescriptions
- Main synthesis:
  - Findings are consistent with earlier literature: a negative relationship between ODA grants and tax revenue.
  - Results robust to different model specifications and to consideration of endogeneity and serial correlation.
  - Effects appear stronger in low-income countries, though the negative impact of ODA grants appears to be weakening over time.
  - Comparable historical benchmarks:
    - Gupta et al. (2004) found an offset of total revenue of about 28 cents for each additional dollar in grants.
    - This paper (using comparable estimation method) finds tax revenue declining by 9 cents for each grant dollar.
- Policy recommendations:
  - Negative impact of grants can be managed if policymakers focus on strengthening domestic revenue mobilization capacity.
  - Evidence presented of increased use of revenue benchmarks in many low-income countries to strengthen revenue performance.
  - Donors and policymakers should prioritize tax policy design, reducing excessive exemptions and investment incentives that erode revenue, and strengthening tax administration, especially in countries with weak institutions.
- Overall implication: To ensure ODA has net additionality, attention to domestic tax capacity and governance is required.

*Source: Appendix II, _wp12186 - Appendix II.*

### Appendix I. Data

### Appendix I. Data

### Sample of Countries
- Low-income countries: Bangladesh, Benin, Burkina Faso, Burundi, Cambodia, Central African Rep., Chad, Comoros, Congo, Dem. Rep. of, Eritrea, Ethiopia, The Gambia, Ghana, Guinea, Guinea-Bissau, Haiti, Kenya, Kyrgyz Republic, Lao People’s Democratic Republic, Madagascar, Mali, Mauritania, Mozambique, Myanmar, Nepal, Niger, Rwanda, Sierra Leone, Solomon Islands, Tajikistan, Tanzania, Togo, Uganda, Zambia, Zimbabwe
- Lower middle-income countries: Angola, Armenia, Belize, Bhutan, Bolivia, Cameroon, Cape Verde, China, P.R.: Mainland, Republic of Congo, Côte d’Ivoire, Djibouti, Egypt, Georgia, Guatemala, Guyana, Honduras, India, Indonesia, Jordan, Lesotho, Maldives, Moldova, Mongolia, Morocco, Nicaragua, Nigeria, Pakistan, Papua New Guinea, Paraguay, Philippines, Samoa, Senegal, Sri Lanka, Sudan, Swaziland, Syrian Arab Republic, São Tomé and Príncipe, Thailand, Tunisia, Ukraine, Uzbekistan, Vanuatu, Vietnam, Republic of Yemen
- Upper middle-income countries: Albania, Algeria, Argentina, Republic of Azerbaijan, Belarus, Bosnia and Herzegovina, Botswana, Brazil, Bulgaria, Chile, Colombia, Dominica, Dominican Republic, Fiji, Gabon, Grenada, I.R. of Iran, Jamaica, Kazakhstan, Lebanon, Libya, Lithuania, Macedonia FYR, Malaysia, Mauritius, Mexico, Namibia, Panama, Peru, Russian Federation, Seychelles, South Africa, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Suriname, Turkey, Uruguay, Rep. Bol. Venezuela

### Data Sources and Variable Definitions
- Total tax revenue, VAT, excises, income tax revenue, and trade tax revenue: IMF’s Fiscal Affairs Department Database on Revenue Mobilization, Government Financial Statistics (GFS) database, and IMF country documents; expressed relative to GDP.
- Net ODA, ODA_Grants, and ODA_Loans (relative to GDP): OCDE (Development Co-operation Directorate) database.
- Share of agriculture and industry in aggregate value added: World Bank’s World Development Indicators (WDI) database.
- Trade Openness: imports plus exports in percent of GDP, from IMF’s International Financial Statistics (IFS) database.
- GDP per capita: constant (2000) U.S. dollars, from WDI, expressed in logs.
- Inflation: annual change in the CPI, from IFS.
- Foreign debt (relative to GDP): WDI database.
- ICRG corruption scores: Political Risk Services Group; scores range from 0 to 6 where 0 indicates the highest potential risk of corruption and 6 indicates the lowest potential risk.
- Oil exporter dummy: equals 1 when the country is a net oil exporter and 0 otherwise.
- Table A1.1 summarizes the data.

### Descriptive Statistics (Table A1.1)
- Total Tax Revenue, percent of GDP
  - Observations: 2728
  - Mean: 16.08
  - Maximum: 61.50
  - Minimum: 0.10
  - Std. Dev.: 7.55
- VAT Tax Revenue, percent of GDP
  - Observations: 764
  - Mean: 4.51
  - Maximum: 16.00
  - Minimum: 0.00
  - Std. Dev.: 2.72
- Income Tax Revenue, percent of GDP
  - Observations: 2650
  - Mean: 4.68
  - Maximum: 50.60
  - Minimum: 0.00
  - Std. Dev.: 4.01
- Excise Tax Revenue, percent of GDP
  - Observations: 1613
  - Mean: 2.00
  - Maximum: 21.90
  - Minimum: 0.00
  - Std. Dev.: 1.59
- Trade Tax Revenue, percent of GDP
  - Observations: 2615
  - Mean: 4.01
  - Maximum: 41.80
  - Minimum: 0.00
  - Std. Dev.: 4.09
- Total ODA, percent of GDP
  - Observations: 3322
  - Mean: 4.72
  - Maximum: 96.30
  - Minimum: -2.96
  - Std. Dev.: 6.27
- ODA Grants, percent of GDP
  - Observations: 3322
  - Mean: 4.47
  - Maximum: 130.84
  - Minimum: 0.00
  - Std. Dev.: 6.30
- ODA Loans, percent of GDP
  - Observations: 3322
  - Mean: 0.25
  - Maximum: 26.38
  - Minimum: -34.54
  - Std. Dev.: 1.49
- Agriculture Value-added, percent of GDP
  - Observations: 3140
  - Mean: 23.09
  - Maximum: 93.98
  - Minimum: 0.80
  - Std. Dev.: 14.63
- Industry Value-added, percent of GDP
  - Observations: 3138
  - Mean: 27.99
  - Maximum: 78.52
  - Minimum: 1.88
  - Std. Dev.: 11.76
- Openness, percent of GDP
  - Observations: 3516
  - Mean: 18.75
  - Maximum: 643.10
  - Minimum: 0.02
  - Std. Dev.: 50.49
- GDP per capita, 2000 USD (log)
  - Observations: 3161
  - Mean: 3.92
  - Maximum: 29.94
  - Minimum: 0.19
  - Std. Dev.: 7.39
- Foreign Debt, percent of GDP
  - Observations: 3127
  - Mean: 54.31
  - Maximum: 2,079.72
  - Minimum: 0.11
  - Std. Dev.: 61.75
- Inflation, in percent
  - Observations: 3314
  - Mean: 44.68
  - Maximum: 1,220.00
  - Minimum: -71.43
  - Std. Dev.: 354.99
- Oil Exporter Dummy
  - Observations: 378
  - Mean: 0.31
  - Maximum: 1.00
  - Minimum: 0.00
  - Std. Dev.: 0.46
- ICRG Corruption Score
  - Observations: 2066
  - Mean: 2.49
  - Maximum: 6.00
  - Minimum: 0.00
  - Std. Dev.: 1.02

### Pairwise Correlations
- Figure A1.1 presents pairwise correlations between tax revenue and ODA Grants across low-income, lower-middle-income, and upper-middle-income country samples (figure labels and country points shown in the source).

---

### Appendix II. Full Set of Controls

### Table A2.1 — Selected Coefficients (Dependent variable: total tax revenue to GDP)
- Columns: (1) Difference GMM 1/, (2) System GMM 2/, (3) Difference GMM 1/, (4) System GMM 2/
- Tax, lagged
  - (1): -0.3539*  (0.2748)
  - (2): 0.3599*   (0.2223)
  - (3): 0.1157    (0.1536)
  - (4): 0.6433*** (0.0725)
- Total ODA
  - (1): -0.0332** (0.0146)
  - (2): -0.0293*  (0.0175)
  - (3): (not reported)
  - (4): (not reported)
- Total ODA, squared
  - (1): 0.0002** (0.0001)
  - (2): 0.0002  (0.0001)
- ODA Loans
  - (3): 0.0021  (0.0118)
  - (4): 0.0076  (0.0134)
- ODA Loans, squared
  - (3): -0.0021* (0.0011)
  - (4): -0.0023** (0.0011)
- ODA Grants
  - (3): -0.0190** (0.0087)
  - (4): -0.0219** (0.0098)
- ODA Grants, squared
  - (3): 0.0002** (0.0001)
  - (4): 0.0003** (0.0001)
- Inflation
  - (1): 0.0002  (0.0002)
  - (2): 0.0001  (0.0003)
  - (3): 0.0001  (0.0001)
  - (4): 0.0001  (0.0002)
- Agriculture share in Value-Added
  - (1): -0.0296* (0.0181)
  - (2): -0.0259  (0.0272)
  - (3): -0.0036  (0.0091)
  - (4): -0.0115  (0.0086)
- Industry share in Value-Added
  - (1): -0.0162  (0.0262)
  - (2): -0.0113  (0.0296)
  - (3): 0.0022   (0.0096)
  - (4): 0.0007   (0.0120)
- GDP Per Capita (log)
  - (1): 0.4322  (1.2780)
  - (2): 0.0572  (1.4431)
  - (3): 0.0214  (0.3363)
  - (4): -0.2305 (0.1740)
- Trade Openness
  - (1): -0.0050  (0.0048)
  - (2): -0.0058  (0.0096)
  - (3): -0.0057** (0.0024)
  - (4): -0.0065** (0.0027)
- Oil Exporter
  - (1): -0.4558  (0.5238)
  - (2): -0.4458  (0.5423)
  - (3): -0.0162  (0.2146)
  - (4): -0.3343  (0.3530)
- External Debt to GDP
  - (1): -0.0087  (0.0138)
  - (2): -0.0105  (0.0152)
  - (3): -0.0058  (0.0096)
  - (4): 0.0126   (0.0141)
- Corruption
  - (1): 0.0499   (0.1597)
  - (2): 0.0272   (0.1640)
  - (3): 0.0451   (0.1421)
  - (4): 0.0828   (0.0659)
- Diagnostic statistics
  - M1 (p value): (1) 0.574, (2) 0.573, (3) 0.140, (4) 0.025
  - M2 (p value): (1) 0.206, (2) 0.198, (3) 0.158, (4) 0.923
  - Over-identification Hansen (p value): (1) 0.756, (2) 0.700, (3) 0.702, (4) 0.297
- Sample and instruments
  - Observations: (1) 1395, (2) 1455, (3) 1395, (4) 1455
  - Number of instruments: (1) 46, (2) 48, (3) 69, (4) 72
  - Number of countries: 76 (in all regressions)
- Notes:
  - Dependent variable is total tax revenue to GDP. Full set of year dummies in all regressions.
  - Robust standard errors in parenthesis; ***(**,*) indicate significance at 1(5, 10) percent.
  - 1/ Two step, robust, instruments based on second lags of tax and ODA.
  - 2/ Two step, robust, with instruments based on first lag of differences in tax and ODA in levels equation, and second lags of their levels in the differenced equation.

---

### Appendix III. Revenue Impact of ODA Grants

### Table A3.1 — Estimated Revenue Offset from an Additional Dollar of ODA Grants
- Method: From Tables 1-2, based on system-GMM estimates. For non-statistically significant coefficients, a coefficient of zero is assumed.
- Sample Average (Percent of GDP) and Estimated Coefficients:
  - Total Tax Revenue
    - Sample Average (Percent of GDP): 16.08
    - Estimated Coefficient (ODA Grants): -0.0151
    - Estimated Coefficient (ODA Grants, Squared): 0.0002
    - Impact: -23.96
  - VAT
    - Sample Average (Percent of GDP): 4.30
    - Estimated Coefficient (ODA Grants): -0.1139
    - Estimated Coefficient (ODA Grants, Squared): 0.0045
    - Impact: -47.04
  - Trade Taxes
    - Sample Average (Percent of GDP): 4.01
    - Estimated Coefficient (ODA Grants): 0
    - Estimated Coefficient (ODA Grants, Squared): 0.0002
    - Impact: 0.08
  - Excises
    - Sample Average (Percent of GDP): 1.90
    - Estimated Coefficient (ODA Grants): -0.0788
    - Estimated Coefficient (ODA Grants, Squared): 0.0035
    - Impact: -14.31
  - Income Taxes
    - Sample Average (Percent of GDP): 4.68
    - Estimated Coefficient (ODA Grants): 0
    - Estimated Coefficient (ODA Grants, Squared): -0.0003
    - Impact: -0.14

*Source: _wp12186 - Appendix I. Data*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12186.pdf_