## _wp07277 - 1. List of Variables, Description, and Sources

## Source details

**Canonical URL:** [_wp07277 - 1. List of Variables, Description, and Sources](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07277.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07277.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07277.pdf.json)

---

### Introduction: scope and research question
- Focus: impact of changes in the international financial and aid architecture on bilateral aid allocation behavior.
- Key research questions:
  - Have donors become more selective over time with respect to recipient “need” (income) and “deservingness” (policy/institutional quality)?
  - Have donor sensitivities to country size, debt burden, and colonial linkages changed?
  - Do recipient-level institutional actions (PRSP adoption, HIPC eligibility) affect bilateral aid flows and selectivity?
- Data coverage: bilateral ODA flows for 147 recipient countries from 22 bilateral donors for the period 1970-2004 (three-dimensional panel).
- Dependent variable definition: net aid transfer per capita = total (bilateral) ODA grants + total (bilateral) ODA loans extended to recipients – ODA loan amortization by recipients – interest paid by recipient, scaled by recipient population.
- Main empirical approach: panel regressions (preferred estimator: Hausman-Taylor), robustness checks with fixed effects and random effects; sample used in regressions: some 48,000 observations.

### Data and main variables (definitions and sources)
- Net aid transfer pc: net aid transfer per capita. Source: DAC.
- Lagged GDP/capita: GDP per capita at PPP rates and in 2000 prices, lagged 1 year. Source: WDI.
- Log (population): the log of population. Source: WDI.
- CPIA: Country Policy and Institutional Assessment score. Source: World Bank.
- PV debt: present value of debt as a ratio to exports of goods and services. Source: World Bank.
- Net aid others: net aid per capita provided by all other donors. Source: DAC.
- Donor net aid sum: the sum of net aid provided by the specific donor. Source: DAC.
- Lagged bilateral trade: sum of bilateral donor–recipient exports and imports scaled by recipient GDP, lagged 1 year. Source: IMF DOT.
- Multilateral debt share / Bilateral debt share: shares of multilateral and bilateral debt in total debt. Source: GDF/WDI.
- HIPC: dummy if country passed the HIPC decision point. Source: Own.
- PRSP: dummy if the country adopted a PRSP. Source: Own.
- Colony: dummy = 1 if donor–recipient pair has colonial history. Source: Own.

### Methodology and estimation choices
- Panel structure: donor–recipient–time three-dimensional panel with bilateral interaction (two-way model with time and bilateral effects).
- Preferred estimator: Hausman-Taylor (HT) to estimate time-invariant effects while addressing endogeneity.
  - Variables treated as endogenous in HT: Lagged GDP/capita; Bilateral debt share; Multilateral debt share; Net aid others; Lagged bilateral trade; PV debt; CPIA; and HIPC in some specifications.
  - Example of time-invariant exogenous: Colony dummy.
- Treatment of zeros: drop donor-recipient pairs with zero bilateral flows for the whole period; retain zero observations for pairs with non-zero flows in some years.
- Robustness checks: fixed effects, random effects, Hausman pre-tests; Tobit/Heckman considerations discussed. Main results robust across estimators.

### Descriptive statistics and stylized facts (selected exact values)
- Aggregate aid dynamics (1970–2004): increase in 1980s, drop mid-1990s, recovery after about 1998; total aid volume in 2004 close to early 1990s peak. Grants have replaced loans over time; net loan transfers became negative in recent years; debt relief explains spikes.
- Per capita net aid transfer ranged between 6 and 8 dollar per person (constant) over full period, with mid-1990s outlier.
- Summary statistics (raw):
  - Net aid transfer per capita (All observations): Number of observations = 90,516; Mean = 2.39; Standard Deviation = 42.89; Minimum = -137.56; Maximum = 9,052.
  - Net aid transfer per capita (Non-zeros): Number of observations = 56,264; Mean = 3.84; Standard Deviation = 54.35; Minimum = -137.56; Maximum = 9,052.
  - Net aid transfer per capita (Positive observations): Number of observations = 53,649; Mean = 4.08; Standard Deviation = 55.64; Minimum = 0.00; Maximum = 9,052.
  - Lagged GDP/capita ($): Number of observations = 67,980; Mean = 3,900; Standard Deviation = 3,363; Minimum = 466; Maximum = 23,266.
  - Population: Number of observations = 103,972; Mean = 2.8mn; Standard Deviation = 11.8mn; Minimum = 19,700; Maximum = 1.3bn.
  - CPIA: Number of observations = 65,252; Mean = 3.46; Standard Deviation = 0.88; Minimum = 0.72; Maximum = 6.00.
  - PV debt (percent of goods and services exports): Number of observations = 75,592; Mean = 181.0; Standard Deviation = 329.0; Minimum = 0.00; Maximum = 6,510.
  - Net aid others ($): Number of observations = 90,516; Mean = 32.5; Standard Deviation = 108.0; Minimum = -129.2; Maximum = 9,567.
  - Donor net aid sum ($): Number of observations = 90,516; Mean = 308.6; Standard Deviation = 648.6; Minimum = -18.6; Maximum = 10,399.
  - Lagged bilateral trade (percent of GDP): Number of observations = 70,621; Mean = 2.20; Standard Deviation = 12.2; Minimum = 0.00; Maximum = 1,543.
  - Multilateral debt share (percent): Number of observations = 81,114; Mean = 32.6; Standard Deviation = 23.6; Minimum = 0.00; Maximum = 100.
  - Bilateral debt share (percent): Number of observations = 81,114; Mean = 38.3; Standard Deviation = 23.1; Minimum = 0.00; Maximum = 100.
- By end-2004: 27 percent of countries had a PRSP; 18 percent had passed the HIPC decision point.

### Main empirical findings — aggregate and time variation (exact coefficients and patterns)
- General conclusion: donors have shifted toward greater selectivity on “need” and “policy” and away from historical/geopolitical and defensive-lending motives; defensive lending influence declined.
- Income (poverty) selectivity:
  - Income responsiveness increased over time. Table 7 averages by period:
    - 1970-1989: Lagged GDP/capita = -0.376
    - 1990-1998: Lagged GDP/capita = -0.515
    - 1999-2004: Lagged GDP/capita = -0.545
  - Hausman-Taylor base estimate: Lagged GDP/capita = -0.598** (0.043) (Table 3 and Table 4 HT base).
- Small-country bias (population):
  - Decline in favoritism toward small countries. Table 7 averages by period:
    - 1970-1989: Log (population) = -2.396
    - 1990-1998: Log (population) = -2.031
    - 1999-2004: Log (population) = -1.765
  - Hausman-Taylor base estimate: Log (population) = -1.013** (0.097) (Table 4 HT base).
- Policy/institutional selectivity (CPIA):
  - Sharp increase in policy sensitivity. Table 7 averages:
    - 1970-1989: CPIA = -0.067
    - 1990-1998: CPIA = 0.185
    - 1999-2004: CPIA = 0.899
  - Hausman-Taylor base: CPIA = 0.0898* (0.044) (Table 4 HT base); expanded specifications and time interactions show larger positive CPIA coefficients in later periods.
- Debt and defensive lending:
  - PV debt coefficient: negative and significant in early period, statistically insignificant thereafter in aggregated patterns.
  - Debt composition:
    - Multilateral debt share: positive and significant in first period (10 percent level), turns negative (not significant) in later periods.
    - Bilateral debt share: positive and significant in first period (5 percent), negative and statistically significant in the third period.
  - Appendix interaction estimates (selected):
    - Time-period interactions: *dummy1970-1989 multilateral interaction = 0.684+ (0.40); *dummy1970-1989 bilateral interaction = 1.067** (0.31).
    - *dummy1999-2004 bilateral debt share interacted = -0.881* (0.36).
- Bilateral linkages and control variables (key HT coefficients, Table 3 / Table 4 HT base):
  - Net aid others = 6.602** (1.10).
  - Donor net aid sum = -0.252** (0.043).
  - Lagged bilateral trade = 13.22** (1.09).
  - Colony (donor–recipient colonial history) = 7.543** (0.90).
- Time-split dynamics:
  - Yearly and period interactions indicate rising responsiveness to poverty and policy, decreasing influence of population and debt across 1980–2005 figures and time-split regressions.

### Recipient-level institutional actions: PRSP and HIPC impacts (exact findings)
- PRSP and HIPC dummies: both positive and statistically significant in many specifications; implied increase in bilateral aid of about $0.40–$0.50 per capita (noting average bilateral aid ≈ $4 per capita).
- Interaction effects:
  - PRSP adoption: generally no statistically significant increased responsiveness to income, CPIA, debt, or population in most interactions.
  - HIPC eligibility: interactions generally not significant; one positive HIPC × CPIA interaction observed but ambiguous due to other coefficient patterns.
- Debt composition interactions:
  - Evidence that PRSP adoption and HIPC eligibility reduce the importance of debt composition (shares of multilateral and bilateral debt) for aid allocation, consistent with reduced defensive lending.
  - Effect stronger for bilateral debt share with PRSP; HIPC effect on reducing defensive lending is strong for both bilateral and multilateral shares (see Appendix Table 1 interaction coefficients).

### Donor-level heterogeneity and time-varying sensitivities (selected exact values)
- Donor sensitivities (Hausman-Taylor donor-specific averages; Table 6 sample entries):
  - Italy: Lagged GDP/capita = 0.215 | CPIA = -0.278 | Log (population) = -1.604 | PV Debt = 0.107
  - Luxembourg: -0.710 | 0.047 | -1.885 | -0.068
  - United States: -2.357 | 0.050 | -1.657 | 0.171
  - Finland: -0.175 | 0.135 | -1.952 | 0.006
  - Norway: -0.599 | 0.156 | -2.033 | 0.022
  - Japan: 0.863 | 0.702 | -1.945 | 0.056
  - Germany: -0.747 | 0.737 | -2.445 | -0.201
  - United Kingdom: -2.100 | 0.969 | -3.157 | -0.160
- Donor debt sensitivity range: 0.307 for France to 0.171 for the United States (text note), indicating cross-donor variation in how debt affects aid allocation.
- Average sensitivities by period (Table 7 repeated):
  - 1970-1989: Lagged GDP/capita = -0.376; CPIA = -0.067; Log (population) = -2.396; PV Debt = -0.089.
  - 1990-1998: Lagged GDP/capita = -0.515; CPIA = 0.185; Log (population) = -2.031; PV Debt = -0.001.
  - 1999-2004: Lagged GDP/capita = -0.545; CPIA = 0.899; Log (population) = -1.765; PV Debt = -0.017.
- Implication: aggregate improvement in selectivity but persistent donor heterogeneity; convergence stronger in CPIA dimension than GDP per capita dimension.

### Robustness, diagnostics, and model performance
- Preferred estimator selection: Hausman pre-tests and economic intuition; HT preferred to fixed effects, which in turn preferred to random effects in pre-test ordering.
  - Hausman test comparisons: fixed vs. random: χ2(36)=496.89 (p=0.000). Fixed vs. Hausman-Taylor: χ2(36)=46.55 (p=0.112).
- Comparable signs and statistical significance across fixed effects, random effects, and HT; magnitudes sometimes differ but main conclusions stable.
- Observations in regression tables: 47,883 in main models; number of bilateral effects = 2,349.
- Treatment of zeros and sample handling tested; results robust to alternative treatments and sample restrictions.
- Additional robustness: donor-specific regressions, grouping donors and recipients by characteristics; most checks confirmed panel results though statistical significance can be reduced in subgroup analyses.

### Policy-relevant conclusions and recommendations (inferred from empirical evidence)
- Aggregate outcomes:
  - Donors have become more selective over time—allocating more aid to poorer countries and to countries with better policy and institutional environments.
  - Small-country bias and the importance of colonial ties have declined.
  - Defensive lending distortions associated with high debt stocks and debt composition have diminished, consistent with effects of debt relief (HIPC, MDRI) and PRSP processes.
- Country-level policy implications:
  - PRSP adoption and HIPC eligibility are associated with increased bilateral aid (about $0.40–$0.50 per capita) and with reductions in the importance of debt composition for aid allocation.
- Donor-internal reforms:
  - Significant variation across donors in selectivity suggests donor domestic institutional environments matter. Correlations (2006) between the CGD Commitment to Development Index (CDI) and KKM governance indexes indicate:
    - CDI correlation with voice and accountability index: 0.81.
    - CDI correlation with government effectiveness index: 0.75.
    - CDI correlation with control of corruption index: 0.75.
    - CDI correlation with political stability index: 0.26.
    - Note: correlation between the CDI specific-quality-of-aid index and KKM voice and accountability = 0.67.
  - Implication: multifaceted donor reforms—including political economy and accountability mechanisms—may be required to achieve more uniform improvements in selectivity across donors.
- Recommended areas for further research and monitoring:
  - Document and quantify over time donor- and international-level institutional changes (transparency, coordination, accountability) to identify drivers of behavioral change.
  - Continued monitoring of debt relief impacts on aid composition and fiscal outcomes given concerns about “headroom” and substitution effects.

### Selected exact regression coefficients and diagnostics (representative HT estimates)
- Hausman-Taylor base (Table 4 HT, Observations = 47,883; bilateral effects = 2,349):
  - Lagged GDP/capita = -0.598** (0.043).
  - Log (population) = -1.013** (0.097).
  - CPIA = 0.0898* (0.044).
  - Net aid others = 6.602** (1.10).
  - Donor net aid sum = -0.252** (0.043).
  - Lagged bilateral trade = 13.22** (1.09).
  - Colony = 7.543** (0.90).
- Fixed effects and random effects (Table 3 selected):
  - Lagged GDP/capita: Fixed effects = -0.651** (0.046); Random effects = -0.323** (0.035).
  - Log (population): Fixed effects = -2.055** (0.45); Random effects = -0.808** (0.063).
  - CPIA: Fixed effects = 0.0795+ (0.045); Random effects = 0.109* (0.044).
- Appendix interaction highlights (Appendix Table 1):
  - Multilateral debt share interacted: *dummy1970-1989 = 0.684+ (0.40).
  - Bilateral debt share interacted: *dummy1970-1989 = 1.067** (0.31); *dummy1999-2004 = -0.881* (0.36).

*Source: _wp07277 - 1. List of Variables, Description, and Sources (IMF Working Paper content unit provided).*

### 1. List of Variables, Description, and Sources.........................................................................3

### _wp07277 - 1. List of Variables, Description, and Sources.........................................................................3

### Major sections
- 1. List of Variables, Description, and Sources.........................................................................35
- 2. Descriptive Statistics of Variables Used..............................................................................36
- 3. Fixed Effects, Random Effects, and Hausman Taylor Estimations.....................................37
- 4. Basic Regression Results .....................................................................................................38
- 5. Expanded Regression Results ..............................................................................................39
- 6. Donor Specific Sensitivities With Respect to Country Variables .......................................40
- 7. Average Sensitivity ..............................................................................................................40

### Figures (listed)
- 1. Bilateral Net ODA Transfers (1970–2004; millions of USD at year 2000 constant ...........41
- 2. Recipient Country Per Capita Bilateral Net ODA Transfers ...............................................41
- 3. Evolution of Responsiveness of Aid to Countries’ GDP per Capita ...................................42
- 4. Evolution of Responsiveness of Aid to Countries’ Population ...........................................43
- 5. Evolution of Responsiveness of Aid to Countries’ Policy...................................................44
- 6. Evolution of Responsiveness of Aid to Countries’ Debt .....................................................45
- 7. Evolution of Responsiveness of Aid to Countries’ Colonial Linkages .................................................46
- 8a. Time-Varying, Donor-Specific Sensitivities for CPIA ......................................................47
- 8b. Time-Varying, Donor-Specific Sensitivities for GDP per Capita .....................................48
- 8c. Time-Varying, Donor-Specific Sensitivities for (log) Population .....................................49
- 8d. Time-Varying, Donor-Specific Sensitivities for PV Debt .................................................50
- 9. Relationship between KKM Voice and Accountability.......................................................51

### Appendix
- Additional Regression Results .................................................................................................52

*Source: _wp07277 - 1. List of Variables, Description, and Sources (PDF).*

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

### _wp07277 - References

### Introduction: scope and research question
- Focus: impact of changes in the international financial and aid architecture on bilateral aid allocation behavior.
- Key research questions:
  - Have donors become more selective over time with respect to recipient “need” (income) and “deservingness” (policy/institutional quality)?
  - Have donor sensitivities to country size, debt burden, and colonial linkages changed?
  - Do recipient-level institutional actions (PRSP adoption, HIPC eligibility) affect bilateral aid flows and selectivity?
- Data coverage: bilateral ODA flows for 147 recipient countries from 22 bilateral donors for the period 1970-2004 (three-dimensional panel).
- Dependent variable: net aid transfer per capita = total (bilateral) ODA grants + total (bilateral) ODA loans extended to recipients – ODA loan amortization by recipients – interest paid by recipient, scaled by recipient population.
- Main empirical approach: panel regressions (preferred estimator: Hausman-Taylor), robustness checks with fixed effects and random effects; sample used in regressions: some 48,000 observations.

### Related literature: context and prior findings
- Earlier literature (pre-1990s): political and strategic interests often dominated economic or developmental objectives in aid allocation.
- Key influences on recent research and policy: Alesina and Dollar (2000), Burnside and Dollar (2000), World Bank study ‘Assessing Aid’ (1998).
- Recent empirical studies report improved donor selectivity over time (e.g., Dollar and Levin 2006; Berthélemy and Tichit 2004; Roodman 2005; Sundberg and Gelb 2006), though dissenting evidence exists (Easterly 2007).
- Debt relief (HIPC, MDRI) and institutional reforms (PRSPs, Paris Declaration) constitute major changes in the international aid architecture that can plausibly affect allocation patterns.
- Defensive lending/granting hypothesis: high debt stocks and certain debt compositions can have distorted impacts on new aid flows; debt relief should reduce defensive lending and restore selectivity.

### Data and main variables
- Aid data source: OECD/DAC On-line CRS system (DAC Table 2a), disaggregated by grants, loans and debt relief (1970–2004).
- Recipient sample: developing countries (DAC Part I list as of January 1, 2001); Part II recipients excluded.
- Macroeconomic data: World Bank WDI; debt data: World Bank Global Development Finance and a World Bank time series of present value of debt (used in Chauvin and Kraay, 2005, 2006).
- Key explanatory variables:
  - Poverty/need: recipient GDP per capita (constant U.S. dollars), lagged one period.
  - Policy selectivity: World Bank Country Policy and Institutional Assessment (CPIA) index.
  - Size effect: population (log used in regressions).
  - Defensive lending proxies: present value of debt relative to exports; shares of bilateral and multilateral claims in total debt.
  - Bilateral linkages/control variables: colonial linkage dummies, bilateral trade (share of GDP), net aid provided by other donors, donor total aid generosity.
- Country-level institutional events:
  - Post-PRSP dummy: date of full PRSP as published by IMF/World Bank.
  - Post-HIPC dummy: enhanced HIPC decision point as break year.
- Time splits for structural analysis: 1970–89, 1990–98, 1999–2004 (also considered 1999–2001 and 2002–04 to check post-9/11 effects).

### Methodology and estimation choices
- Panel model structure: three-dimensional donor-recipient-time panel with bilateral interaction (two-way model with time and bilateral effects).
- Preferred estimator: Hausman-Taylor (HT) to allow estimation of time-invariant effects while addressing endogeneity of selected regressors.
  - Variables treated as endogenous in HT specification (instrumented): lagged GDP per capita, share of bilaterals and multilaterals in total debt, aid of other donors to same country, lagged trade as share of GDP, and CPIA.
  - Time-invariant exogenous example: colonial ties dummy.
- Treatment of zeros: drop donor-recipient pairs that have zero bilateral flows for the whole period; keep zero observations for pairs that have non-zero flows in some years to avoid bias from permanently excluded pairs.
- Robustness checks: fixed effects, random effects, Hausman pre-tests, Tobit/Heckman considerations discussed; main results robust across estimators.

### Descriptive statistics and stylized facts
- Aggregate aid dynamics (1970–2004):
  - Aid volumes: increase in eighties, drop from mid-1990s, recovery after about 1998; total aid volume in 2004 close to early 1990s peak.
  - Composition shift: grants have replaced loans over time; net loan transfers became negative in recent years; debt relief explains spikes (e.g., 1991) and much of recent increases.
  - Per capita net aid transfer ranged between 6 and 8 (constant) dollar per person over the full period, with mid-1990s outlier.
- Summary statistics (raw):
  - Average net aid transfer per donor-recipient: $2.4 per capita per year; minimum -$138; maximum $9,052 per capita per year.
  - If excluding zero observations: average aid per capita per donor = $3.80 ($4 excluding negative entries).
  - Component averages per donor (including zeros): grants $2.20 per capita per year; net loans $0.16 per capita per year; debt relief $0.04 per capita per year.
  - Recipient GDP per capita (2000 prices) average: $3,900 (range < $500 to $23,266).
  - Population average: 2.8 million; standard deviation: 11.8 million; smallest about 20,000; largest 1.3 billion.
  - Total aid provided by other donors to same recipient average: $32 per capita per year; donors on average provided $309 net aid transfer per capita to all other countries in same year.
  - Debt burden (present value/expo rts) average: 181 percent (range <1 percent to 6,500 percent).
  - Multilateral share of total debt average: 33 percent; bilateral share average: 38 percent.
  - By end-2004: 27 percent of countries had a PRSP; 18 percent had passed the HIPC decision point.

### Main empirical findings (aggregate and time variation)
- Overall finding: significant structural changes in aid allocation over time—donors allocate more on “need” and “policy” grounds and less on historical/(geo-)political grounds; defensive lending influence has declined.
- Robustness: results consistent across fixed effects, random effects and Hausman-Taylor specifications; coefficients’ signs and statistical significance largely stable.
- Income (poverty) selectivity:
  - Income responsiveness increased over time (coefficients for log GDP per capita): from -0.434 to -0.604 (all highly statistically significant).
  - Yearly interactions show coefficient decreasing from -0.4 to -0.6 over the sample (income sensitivity always statistically different from zero).
- Small-country bias:
  - Coefficient for log population became less negative over time: from -1.394 to -0.733 (both statistically significant), indicating a decline in small-country favoritism.
  - Yearly trend: population coefficient moves from about -1.5 to -0.7 (statistically significant different from zero through time).
- Policy/institutional selectivity (CPIA):
  - Policy sensitivity increased markedly: CPIA coefficient evolves from essentially 0 in early period to 0.228 (second period), then to 0.954 (most recent period).
  - Annual evolution indicates CPIA coefficient rising up to about 1.2 with sharpest rise starting mid-1990s (policy sensitivity not statistically different from zero until mid-1990s).
- Debt and defensive lending:
  - Present value of debt relative to exports: coefficient negative and significant in early period, becomes statistically insignificant thereafter—suggests declining role of debt burden in reducing aid flows.
  - Debt composition dynamics:
    - Multilateral debt share: positive and statistically significant in first period (10 percent level), turns negative (not significant) in later periods.
    - Bilateral debt share: positive and significant (5 percent) in first period, negative but not significant in second, and statistically significant negative in third period.
  - Interpretation: early defensive lending to protect creditor claims diminished over time; debt relief (HIPC, MDRI) and grant substitution likely contributed to restoration of selectivity, with bilateral donors able to substitute grants faster than multilaterals.
- Colonial linkage and geopolitical factors:
  - Colonial linkages have declined in importance over time; graphical trends show declining role of colonial ties—aid allocated more according to economic criteria over time.
- Period-split and post-9/11 checks:
  - Further split of last period into 1999–2001 and 2002–04 shows progress in selectivity accelerated in the later subperiods; income and CPIA coefficients further increase in magnitude.
- Visual/yearly interactions:
  - Figures (3–7, described in text) show consistent year-by-year increases in responsiveness to poverty and policy, and decreases in influence of population and debt.

### Recipient-level institutional actions: PRSP and HIPC impacts
- Net aid flows increase following PRSP adoption and on HIPC eligibility:
  - PRSP and HIPC dummies both positive and statistically significant; implied increase in bilateral aid of about $0.40 - $0.50 per capita (noting average bilateral aid ≈ $4 per capita).
- Interaction effects (PRSP/HIPC × key determinants):
  - PRSP adoption: no statistically significant increased responsiveness to income, policy (CPIA), debt, or population following PRSP in most interactions.
  - HIPC eligibility: interaction results generally not significant; one positive interaction (HIPC × CPIA) observed but interpretation ambiguous because HIPC dummy itself can be negative and significant in some specifications.
- Debt composition interactions:
  - Evidence that both PRSP adoption and HIPC eligibility reduce the importance of debt composition (shares of multilateral and bilateral debt) in determining aid flows—i.e., reduced defensive lending after country-level debt relief and PRSP adoption.
  - Effect stronger for bilateral debt share with PRSP; HIPC effect on reducing defensive lending is strong for both bilateral and multilateral shares.

### Donor-level heterogeneity and changes over time
- Donors differ substantially in selectivity dimensions; differences have persisted though many donors improved selectivity over time.
- Donor-by-donor elasticities (Table 6 summary):
  - CPIA sensitivity ranges across donors from -0.278 (Italy) to 0.969 (United Kingdom).
  - GDP per capita sensitivity ranges from -2.357 (United States) to 0.863 (Japan).
  - Population sensitivity ranges from -3.157 (United Kingdom) to -1.481 (Canada).
  - Debt burden sensitivity varies across donors (detailed donor-level numbers described in Table 6 in text).
- Implication: reforms to the international aid architecture improved aggregate selectivity, but remaining donor heterogeneity suggests donor domestic institutional environments matter for revealed selectivity; further multifaceted reforms in donor political economy and accountability systems may be necessary.

### Robustness and model diagnostics
- Preferred estimator chosen via Hausman pre-test and economic intuition; HT preferred to fixed effects, which is preferred to random effects in pre-test ordering.
- Fixed effects, random effects and HT yield similar signs and significance; coefficients sometimes differ in magnitude but main conclusions unchanged.
- Treatment of zeros: excluded permanently zero donor-recipient pairs; kept zero observations for pairs with non-zero flows in some years—results robust to different treatments as shown by comparison with alternatives and past literature.
- Additional robustness: reduction of outliers and alternative data samples tested with similar qualitative outcomes.

### Policy-relevant conclusions (from empirical evidence)
- Aggregate improvements:
  - Donors have become more selective over time—allocating more aid to poorer countries and to countries with better policy and institutional environments.
  - The small-country bias and the importance of colonial ties have declined.
  - Defensive lending distortions associated with high debt stocks and debt composition have diminished, consistent with the intended effects of debt relief (HIPC, MDRI) and PRSP processes.
- Country-level reforms:
  - PRSP adoption and HIPC eligibility are associated with increased bilateral aid to those countries (about $0.40–$0.50 per capita increase), and with reductions in the importance of debt composition for aid allocation.
- Donor heterogeneity:
  - Significant variation exists across donors in selectivity and responsiveness to policy, poverty, size and debt; donor-internal institutional environments appear to influence aid allocation behavior.
- Research and policy recommendations (inferred from empirical limitations and findings):
  - Further work needed to document and quantify over time donor- and international-level institutional changes (e.g., transparency, coordination, accountability) to identify which reforms have driven behavioral changes.
  - Multifaceted donor reforms—including political economy and accountability mechanisms—may be required to achieve more uniform improvements in selectivity across donors.
  - Continued monitoring of debt relief impacts on aid composition and fiscal outcomes is warranted given concerns about “headroom” and substitution effects.

*Source: _wp07277 - References (IMF Working Paper content unit) — content as provided.*

### 0.307 for France to 0.171 for the United States, suggesting that for France debt is more a

### _wp07277 - 0.307 for France to 0.171 for the United States, suggesting that for France debt is more a

### Improvement in donor selectivity over time
- Coefficients on debt sensitivity vary across donors, ranging from 0.307 for France to 0.171 for the United States, suggesting that for France debt is more a detriment to aid flows than it is for the United States.
- Average sensitivities across the 22 donors for three periods (reported in Table 7) show:
  - an increase in sensitivity with respect to income;
  - a sharp increase in sensitivity with respect to CPIA;
  - a lowering of the bias towards smaller countries;
  - a reduced concern over debt burdens.
- Donors have become more homogeneous in their aid allocations over time in the CPIA dimension, as measured by a decline in the coefficient of variation in the sensitivities.
- Figures (referenced) indicate:
  - CPIA dimension: for all donors (except the U.S.) sensitivities were much higher in the late 1990s than before;
  - Population dimension: sensitivities are uniformly higher (i.e., less negative);
  - Debt dimension: coefficients generally increase towards zero (not uniformly);
  - GDP per capita dimension: progress is less consistent across donors and significant differences remain.
- Overall conclusion: clear improvements and convergence in reducing small-country bias, increasing policy selectivity, lowering concerns over debt burdens, and somewhat increasing income sensitivities, with most changes occurring in the 1990s and intensified more recently.

### Robustness checks and heterogeneity analysis
- Panel regression results depend on degree of homogeneity over time, across donors, and across recipients.
- Heterogeneity tests conducted:
  - Individual aid allocation regressions for each donor separately;
  - Aid allocation regressions by groups of similar-like donors;
  - Aid allocation regressions by similar-like recipients grouped by income level, CPIA index score, and country size.
- Most robustness checks confirmed the general panel regression results but with generally reduced statistical significance.
- Notable exception: when grouping recipients by level of income, income is no longer as statistically significant.
- For space reasons, the detailed robustness results are not reported in the source.

### Conclusions on behavioral changes in aid allocation
- Over time, bilateral aid flows have shifted toward more optimal allocations:
  - roles of poverty and countries’ policy and institutional environment have increased;
  - small-country effect has reduced;
  - role of debt burden in deterring aid flows has declined;
  - no evidence of defensive lending driving overall flows in recent periods.
- Recipient-country actions found to matter:
  - Debt relief, especially through the HIPC Initiative, and adoption of PRSPs reduced the importance of the share of official debt in determining aid flows.
- These changes are related to reforms in the international aid architecture, though the specific institutional drivers remain unresolved and warrant further analysis.

### Donor institutional environment and commitment to development
- Correlations computed for year 2006 between the CGD Commitment to Development Index (CDI) and Kaufmann, Kraay, and Mastruzzi (KKM, 2004) governance indexes for 21 donor countries (CDI not available for Luxembourg) show:
  - CDI correlation with voice and accountability index: 0.81;
  - CDI correlation with government effectiveness index: 0.75;
  - CDI correlation with control of corruption index: 0.75;
  - CDI correlation with political stability index: 0.26.
- Note: correlations are somewhat less strong for the specific quality of aid CDI index; correlation with the KKM voice and accountability index in that case is 0.67.
- Interpretation:
  - Donor countries with better governance tend to be more committed to development and provide aid in a more development-friendly manner.
  - Preferences of citizens in donor countries, greater government effectiveness, and lower presence of corruption in donor countries are associated with more selective, development-focused aid allocation.
- Implication: future analysis of the international aid architecture should account for donor-country institutional environments in addition to recipient-country policy and institutional factors.

*Source: _wp07277 (content unit provided).*

### References

### References

### Bibliography (selected authors and works)
- Alesina, Alberto, and David Dollar, 2000, “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth, Vol. 5, No.1, pp. 33–63.
- Alesina, Alberto, and Beatrice Weder, 2002, “Do Corrupt Governments Receive Less Foreign Aid?” American Economic Review, Vol. 92, No. 4, pp. 1126–37.
- Amprou, Jacky, Patrick Guillaumont, and Sylviane Guillaumont-Jeanneney, 2007, “Aid Selectivity According to Augmented Criteria,” The World Economy, Vol. 30, No. 5, pp. 733–63.
- Anderson, Edward, and Hugh Waggington, 2006, Aid and the MDG Poverty Target: How much is required, how much can be absorbed, and how it should be allocated, unpublished paper, (London: ODI).
- Baltagi, Badi H., 2001. Econometric Analysis of Panel Data (Chichester: John Wiley & Sons).
- Baltagi, Badi, Georges Bresson, and Alain Pirotte, 2003, “Fixed Effects, Random Effects, or Hausman–Taylor? A Pretest Estimator,” Economics Letters, Vol. 79, No. 3, pp. 361–69.
- Berthélemy, Jean-Claude, 2006, “Bilateral Donors Interest vs. Recipients Development Motives in Aid Allocation: Do All Donors Behave the Same?” Review of Development Economics, Vol. 10, No. 2, pp. 179–94.
- Birdsall, Nancy, Stijn Claessens, and Ishac Diwan, 2003, “Policy Selectivity Foregone: Debt and Donor Behavior in Africa,” World Bank Economic Review, Vol. 17, No. 3, pp. 409–35.
- Burnside, Craig, and David Dollar, 2000, “Aid, Policies, and Growth,” American Economic Review, Vol. 90, No. 4, pp. 847–68.
- Collier, Paul, 2006, “Is Aid Oil? An Analysis of Whether Africa Can Absorb More Aid,” World Development, Vol. 34, No. 9, pp. 1482–97.
- Dollar, David, and Victoria Levin, 2004, “The Increasing Selectivity of Foreign Aid, 1984–2002,” World Bank Policy Working Paper No. 3299 (Washington: World Bank).
- Easterly, William, 2003, “Can Foreign Aid Buy Growth?” Journal of Economic Perspectives, Vol. 17, No. 3, pp. 23–48.
- Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi, 2004, “Governance Matters III: Governance Indicators for 1996, 1998, 2000, and 2002,” World Bank Economic Review, Vol. 18, No. 2, pp. 253–87 (www.govindicators.org was used for the data).
- Rajan, Raghuram, and Arvind Subramanian, 2005, “Aid and Growth: What Does the Cross-Country Evidence Really Show?” IMF Working Paper No. 05/127 (Washington: International Monetary Fund).
- World Bank, 1998, Assessing Aid: What Works, What Doesn’t, and Why (Oxford: Oxford University Press).
- (The document contains a comprehensive list of additional references on aid allocation, debt relief, selectivity, econometric methods, and donor behavior.)

*Reference list continues in source PDF.*

---

### Key Data Definitions and Sources

### Table 1 — List of Variables, Description, and Sources
- Net aid transfer pc = net aid transfer per capita. Source: DAC.
- Lagged GDP/capita = GDP per capita at PPP rates and in 2000 prices, lagged 1 year. Source: WDI.
- Log (population) = the log of population. Source: WDI.
- CPIA = Country Policy and Institutional Assessment score. Source: World Bank.
- PV debt = the present value of debt as a ratio to exports of goods and services. Source: World Bank.
- Net aid others = net aid per capita provided by all other donor. Source: DAC.
- Donor net aid sum = the sum of net aid provided by the specific donor. Source: DAC.
- Lagged bilateral trade = the sum of bilateral donor–recipient country exports and imports scaled by recipient country GDP, lagged 1 year. Source: IMF DOT.
- Multilateral debt share = the share of multilateral debt in total debt. Source: GDF/WDI.
- Bilateral debt share = the share of bilateral debt in total debt. Source: GDF/WDI.
- HIPC = a dummy if the country passed the HIPC decision point. Source: Own.
- PRSP = a dummy if the country adopted a PRSP. Source: Own.
- Colony = a dummy taking the value of 1 if the donor country pair has colonial history. Source: Own.

---

### Descriptive Statistics

### Table 2 — Descriptive Statistics of Variables Used (USD, unless indicated otherwise)
- Net aid transfer per capita:
  - All observations: Number of observations = 90516; Mean = 2.39; Standard Deviation = 42.89; Minimum = -137.56; Maximum = 9052.
  - Non-zeros observations: Number of observations = 56264; Mean = 3.84; Standard Deviation = 54.35; Minimum = -137.56; Maximum = 9052.
  - Positive observations: Number of observations = 53649; Mean = 4.08; Standard Deviation = 55.64; Minimum = 0.00; Maximum = 9052.
- Lagged GDP/capita ($): Number of observations = 67980; Mean = 3900; Standard Deviation = 3363; Minimum = 466; Maximum = 23266.
- Population: Number of observations = 103972; Mean = 2.8mn; Standard Deviation = 11.8mn; Minimum = 19700; Maximum = 1.3bn.
- CPIA: Number of observations = 65252; Mean = 3.46; Standard Deviation = 0.88; Minimum = 0.72; Maximum = 6.00.
- PV debt (percent of goods and services exports): Number of observations = 75592; Mean = 181.0; Standard Deviation = 329.0; Minimum = 0.00; Maximum = 6510.
- Net aid others ($): Number of observations = 90516; Mean = 32.5; Standard Deviation = 108.0; Minimum = -129.2; Maximum = 9567.
- Donor net aid sum ($): Number of observations = 90516; Mean = 308.6; Standard Deviation = 648.6; Minimum = -18.6; Maximum = 10399.
- Lagged bilateral trade (percent of GDP): Number of observations = 70621; Mean = 2.20; Standard Deviation = 12.2; Minimum = 0.00; Maximum = 1543.
- Multilateral debt share (percent): Number of observations = 81114; Mean = 32.6; Standard Deviation = 23.6; Minimum = 0.00; Maximum = 100.
- Bilateral debt share (percent): Number of observations = 81114; Mean = 38.3; Standard Deviation = 23.1; Minimum = 0.00; Maximum = 100.

---

### Regression Results — Main Findings

### Table 3 — Fixed Effects, Random Effects, and Hausman-Taylor Estimations (selected coefficients)
- Model observation counts and effects:
  - Observations: 47883 in each model.
  - Number of bilateral effects: 2349.
- Key coefficient estimates (with robust standard errors in parentheses; ** p<0.01, * p<0.05, + p<0.1):
  - Lagged GDP/capita: Fixed effects = -0.651** (0.046); Random effects = -0.323** (0.035); Hausman-Taylor = -0.598** (0.043).
  - Log (population): Fixed effects = -2.055** (0.45); Random effects = -0.808** (0.063); Hausman-Taylor = -1.013** (0.097).
  - CPIA: Fixed effects = 0.0795+ (0.045); Random effects = 0.109* (0.044); Hausman-Taylor = 0.0898* (0.044).
  - Net aid others: Fixed effects = 6.080** (1.12); Random effects = 9.262** (1.06); Hausman-Taylor = 6.602** (1.10).
  - Donor net aid sum: Fixed effects = -0.297** (0.044); Random effects = -0.205** (0.043); Hausman-Taylor = -0.252** (0.043).
  - Lagged bilateral trade: Fixed effects = 12.22** (1.11); Random effects = 19.83** (1.02); Hausman-Taylor = 13.22** (1.09).
  - Colony (donor–recipient colonial history): Hausman-Taylor = 7.543** (0.90).
- Endogeneity in Hausman-Taylor: Lagged GDP/capita, Bilateral debt share, Multilateral debt share, Net aid others, Lagged bilateral trade, PV debt, and CPIA are treated as endogenous.
- Hausman specification tests:
  - Comparing fixed effects and random effects: χ2(36)=496.89 (p=0.000).
  - Comparing fixed effects with Hausman-Taylor: χ2(36)=46.55, p=0.112.

### Table 4 — Basic Regression Results (Hausman-Taylor base regression and time-interacted specifications)
- Base Hausman-Taylor estimates (column (1), HT):
  - Lagged GDP/capita = -0.598** (0.043).
  - Log (population) = -1.013** (0.097).
  - CPIA = 0.0898* (0.044).
  - Net aid others = 6.602** (1.10).
  - Donor net aid sum = -0.252** (0.043).
  - Lagged bilateral trade = 13.22** (1.09).
  - Colony = 7.543** (0.90).
  - Observations = 47883; Number of bilateral effects = 2349.
- Time-period dummy interactions show varying coefficients across periods:
  - *dummy1970-1989: coefficient examples include -0.434** (0.051) and -1.394** (0.10) in interacted specifications.
  - *dummy1990-1998 and *dummy1999-2004 have negative coefficients in several interacted models (e.g., *dummy1990-1998 = -0.567** (0.048) in one specification).
- Note: All regressions use the Hausman-Taylor model with the same set of endogenous variables as in Table 3.

### Table 5 — Expanded Regression Results (columns 1–13, robustness and additional controls)
- Core patterns stable across specifications (coefficients shown are from multiple columns; robust standard errors in parentheses; ** p<0.01, * p<0.05, + p<0.1):
  - Lagged GDP/capita: range of estimates ≈ -0.598** to -0.585** (standard errors around 0.043–0.044).
  - Log (population): range ≈ -1.013** to -1.049** (standard errors around 0.097–0.10).
  - CPIA: generally positive and statistically significant in many specifications (e.g., 0.0898* (0.044), 0.0914* (0.044), with some + and * significance across columns).
  - PV debt: coefficients generally small and not consistently statistically significant (examples: -0.004 (0.010), -0.003 (0.010), 0.00196 (0.011)).
  - Net aid others: consistently positive and highly significant (e.g., 6.602** (1.10), 6.651** (1.10)).
  - Donor net aid sum: consistently negative and highly significant (e.g., -0.252** (0.043), -0.254** (0.043)).
  - Lagged bilateral trade: consistently positive and highly significant (≈ 13.16** with (1.09) standard error).
  - Multilateral debt share: generally negative but not always significant (e.g., -0.161 (0.33) to -0.468 (0.34)).
  - Bilateral debt share: generally small positive coefficients (e.g., 0.171 (0.27) to 0.274 (0.27)).
  - Colony: consistently positive and highly significant around 7.53** (standard errors around 0.90–0.94).
- Additional policy dummies included in different columns:
  - PRSP: examples include 0.469** (0.16) in one specification and 3.097* (1.46) in another.
  - HIPC: examples include 0.406** (0.15) and 4.515** (1.55) in other specifications.
- Interaction terms and donor-specific effects introduced in later columns:
  - Examples of interaction effects reported include *multilateral debt share coefficients like -2.519+ (1.52) and -4.103* (1.61); *bilateral debt share coefficients like -3.408* (1.73) and -4.729** (1.75).
- Observations and panel structure:
  - Observations = 47883 in each column.
  - Number of bilateral effects = 2349 in each column.

---

### Donor-Specific and Time-Varying Sensitivities

### Table 6 — Donor Specific Sensitivities With Respect to Country Variables (average three periods; sorted by sensitivity w.r.t. CPIA)
- Example donor sensitivities (columns: Lagged GDP/capita | CPIA | Log (population) | PV Debt):
  - Italy: 0.215 | -0.278 | -1.604 | 0.107
  - Luxembourg: -0.710 | 0.047 | -1.885 | -0.068
  - United States: -2.357 | 0.050 | -1.657 | 0.171
  - Finland: -0.175 | 0.135 | -1.952 | 0.006
  - Norway: -0.599 | 0.156 | -2.033 | 0.022
  - Japan: 0.863 | 0.702 | -1.945 | 0.056
  - Germany: -0.747 | 0.737 | -2.445 | -0.201
  - United Kingdom: -2.100 | 0.969 | -3.157 | -0.160
  - (Full donor list includes: Italy, Luxembourg, United States, Finland, Norway, New Zealand, Greece, Ireland, Switzerland, Austria, Sweden, Netherlands, Portugal, Denmark, Belgium, France, Canada, Australia, Japan, Spain, Germany, United Kingdom.)
- Note: Results from Hausman-Taylor regressions with the listed endogenous variables.

### Table 7 — Average Sensitivity by Period
- Period 1970-1989:
  - Lagged GDP/capita = -0.376
  - CPIA = -0.067
  - Log (population) = -2.396
  - PV Debt = -0.089
- Period 1990-1998:
  - Lagged GDP/capita = -0.515
  - CPIA = 0.185
  - Log (population) = -2.031
  - PV Debt = -0.001
- Period 1999-2004:
  - Lagged GDP/capita = -0.545
  - CPIA = 0.899
  - Log (population) = -1.765
  - PV Debt = -0.017

---

### Figures — Trends and Visual Findings (captions and indicated patterns)

- Figure 1. Bilateral Net ODA Transfers (1970–2004; millions of USD at year 2000 constant prices).
  - Plotted components: loans, grants, debt_relief. Time axis: 1970 to 2002 (labels shown: 1970 1974 1978 1982 1986 1990 1994 1998 2002).
- Figure 2. Recipient Country Per Capita Bilateral Net ODA Transfers (1970–2004; USD at year 2000 constant prices).
  - Plotted components: loans, grants, debt_relief. Time axis: 1970 to 2002.
- Figure 3. Evolution of Responsiveness of Aid to Countries’ GDP per Capita (1980–2005).
  - Y-axis range shown from -0.8 to -0.2; series labeled La gg ed GDP/ca p ita (Lagged GDP/capita).
- Figure 4. Evolution of Responsiveness of Aid to Countries’ Population (1980–2005).
  - Y-axis range shown from -2.0 to -0.5; series labeled log (population).
- Figure 5. Evolution of Responsiveness of Aid to Countries’ Policy (1980–2005).
  - Y-axis range shown from -0.5 to 1.5; series labeled CPIA.
- Figure 6. Evolution of Responsiveness of Aid to Countries’ Debt (1980–2005).
  - Y-axis range shown from -0.8 to 0.2; series labeled PV debt.
- Figure 7. Evolution of Responsiveness of Aid to Colonial Linkages (1980–2005).
  - Y-axis range shown from -5 to 5; series labeled Colony.
- Figures 8a–8d. Time-Varying, Donor-Specific Sensitivities (70-89, 90-98, 99-04) for:
  - CPIA (Figure 8a), GDP per Capita (Figure 8b), (log) Population (Figure 8c), PV Debt (Figure 8d).
  - Each figure lists donors and displays sensitivity ranges by period (examples visible in axis scales).
- Figure 9. Relationship between KKM Voice and Accountability and CGD Commitment to Development Index.
  - Scatter regression: R-squared = 0.66; fitted line y = 1.202 + 2.852.x.
  - Axes labeled Voice and Accountability (range ~1.0 to 1.6) and CDI 2006 (range ~3.5 to 6.5).

---

### Appendix
- Appendix contains additional regression results and supplementary tables and figures as listed above.

*Source: _wp07277 - References (source PDF content used above).*

### Appendix Table 1: Expanded Regression Results for

### Appendix Table 1: Expanded Regression Results for Debt Shares Interacted with Time Periods

### Key estimated coefficients (Base regression results (HT); Multilateral debt share interacted; Bilateral debt share interacted)
- Lagged GDP/capita:
  - Base regression (HT): -0.598** (0.043)
  - Multilateral debt share interacted: -0.487** (0.040)
  - Bilateral debt share interacted: -0.627** (0.044)
- Log (population):
  - Base regression (HT): -1.013** (0.097)
  - Multilateral debt share interacted: -0.947** (0.095)
  - Bilateral debt share interacted: -1.002** (0.094)
- CPIA:
  - Base regression (HT): 0.0898* (0.044)
  - Multilateral debt share interacted: 0.104* (0.044)
  - Bilateral debt share interacted: 0.106* (0.044)
- PV debt:
  - Base regression (HT): -0.00366 (0.010)
  - Multilateral debt share interacted: 0.00223 (0.011)
  - Bilateral debt share interacted: -0.00198 (0.011)
- Net aid others:
  - Base regression (HT): 6.602** (1.10)
  - Multilateral debt share interacted: 7.462** (1.09)
  - Bilateral debt share interacted: 5.889** (1.11)
- Donor net aid sum:
  - Base regression (HT): -0.252** (0.043)
  - Multilateral debt share interacted: -0.251** (0.043)
  - Bilateral debt share interacted: -0.249** (0.043)
- Lagged bilateral trade:
  - Base regression (HT): 13.22** (1.09)
  - Multilateral debt share interacted: 13.07** (1.09)
  - Bilateral debt share interacted: 13.23** (1.09)
- Multilateral debt share:
  - Base regression (HT): -0.161 (0.33)
  - Multilateral debt share interacted: -0.0246 (0.34)
  - (No multilateral value listed for bilateral debt share interacted column)
- Bilateral debt share:
  - Base regression (HT): 0.171 (0.27)
  - Multilateral debt share interacted: -0.0242 (0.27)
  - (No bilateral value listed for multilateral debt share interacted column)
- Colony:
  - Base regression (HT): 7.543** (0.90)
  - Multilateral debt share interacted: 7.561** (0.89)
  - Bilateral debt share interacted: 7.541** (0.87)
- Time-period interactions (selected):
  - *dummy1970-1989:
    - Multilateral debt share interacted: 0.684+ (0.40)
    - Bilateral debt share interacted: 1.067** (0.31)
  - *dummy1990-1998:
    - Multilateral debt share interacted: -0.0841 (0.37)
    - Bilateral debt share interacted: -0.300 (0.34)
  - *dummy1999-2004:
    - Multilateral debt share interacted: -0.478 (0.36)
    - Bilateral debt share interacted: -0.881* (0.36)
- Constant:
  - Base regression (HT): 18.13** (1.58)
  - Multilateral debt share interacted: 16.60** (1.56)
  - Bilateral debt share interacted: 17.64** (1.55)
- Sample and model details:
  - Observations: 47883 (for all three specifications)
  - Number of bilateral effects: 2349 (for all three specifications)
  - Note: Robust standard errors in parentheses; ** p<0.01, * p<0.05, + p<0.1.
  - All regressions use the Hausman-Taylor model with Lagged GDP/capita, Bilateral debt share, Multilateral debt share, Net aid others, Lagged bilateral trade, PV debt, CPIA and HIPC as endogenous.

---

### Appendix Table 2: Basic Regression Results for Variables Interacted with Four Period Splits

### Variables Interacted with Time-Period Dummies — Columns (1) through (5)
- Column headings and main endogenous variables noted in the regression: Base regression (HT); Lagged GDP/capita; Log (population); CPIA; PV debt.
- Lagged GDP/capita:
  - (1) Base regression (HT): -0.598** (0.043)
  - (2): -0.528** (0.043)
  - (3): -0.741** (0.046)
  - (4): -0.588** (0.044)
  - (5): (value not separately listed beyond above columns)
- Log (population):
  - (1) Base regression (HT): -1.013** (0.097)
  - (2): -1.101** (0.11)
  - (3): -1.064** (0.097)
  - (4): -1.014** (0.097)
  - (5): -1.013** (0.097)
- CPIA:
  - (1) Base regression (HT): 0.0898* (0.044)
  - (2): 0.102* (0.044)
  - (3): 0.126** (0.044)
  - (4): 0.0662 (0.044)
  - (5): 0.0898* (0.044)
- PV debt:
  - (1) Base regression (HT): -0.00366 (0.010)
  - (2): -0.00857 (0.010)
  - (3): 0.0140 (0.010)
  - (4): 0.00181 (0.011)
  - (5): -0.00366 (0.010)
- Net aid others:
  - (1) Base regression (HT): 6.602** (1.10)
  - (2): 5.662** (1.11)
  - (3): 0.819 (1.15)
  - (4): 4.780** (1.12)
  - (5): 6.228** (1.10)
- Donor net aid sum:
  - (1): -0.252** (0.043)
  - (2): -0.263** (0.043)
  - (3): -0.251** (0.043)
  - (4): -0.251** (0.043)
  - (5): -0.252** (0.043)
- Lagged bilateral trade:
  - (1): 13.22** (1.09)
  - (2): 13.01** (1.08)
  - (3): 12.90** (1.08)
  - (4): 13.15** (1.09)
  - (5): 13.18** (1.09)
- Multilateral debt share:
  - (1): -0.161 (0.33)
  - (2): -0.570+ (0.34)
  - (3): 0.425 (0.34)
  - (4): 0.0744 (0.33)
  - (5): -0.153 (0.33)
- Bilateral debt share:
  - (1): 0.171 (0.27)
  - (2): 0.323 (0.27)
  - (3): 0.242 (0.27)
  - (4): 0.258 (0.27)
  - (5): 0.227 (0.27)
- Colony:
  - (1): 7.543** (0.90)
  - (2): 7.528** (1.03)
  - (3): 7.478** (0.95)
  - (4): 7.540** (0.90)
  - (5): 7.544** (0.90)
- Time-period interaction dummies (selected coefficients and standard errors):
  - *dummy1970-1989:
    - (1): -0.434** (0.051)
    - (2): -1.418** (0.10)
    - (3): -0.0542 (0.051)
    - (4): -0.0768** (0.020)
  - *dummy1990-1998:
    - (1): -0.567** (0.048)
    - (2): -1.040** (0.10)
    - (3): 0.228** (0.067)
    - (4): 0.00869 (0.011)
  - *dummy1998-2001:
    - (1): -0.599** (0.047)
    - (2): -0.726** (0.11)
    - (3): 0.857** (0.13)
    - (4): -0.0167 (0.035)
  - *dummy2002-2004:
    - (1): -0.610** (0.046)
    - (2): -0.794** (0.11)
    - (3): 1.072** (0.14)
    - (4): -0.0131 (0.042)
- Constant terms:
  - (1) Base regression (HT): 18.13** (1.58)
  - (2): 20.71** (1.88)
  - (3): 15.00** (1.81)
  - (4): 17.49** (1.70)
  - (5): 18.56** (1.59)
- Sample and model details:
  - Observations: 47883 (for all five specifications)
  - Number of bilateral effects: 2349 (for all five specifications)
  - Note: Robust standard errors in parentheses; ** p<0.01, * p<0.05, + p<0.1.
  - All regressions use the Hausman-Taylor model with Lagged GDP/capita, Bilateral debt share, Multilateral debt share, Net aid others, Lagged bilateral trade, PV debt, and CPIA as endogenous.

*Source: Appendix Tables 1 and 2 from the provided content unit.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07277.pdf_