## _wp05199

## Source details

**Canonical URL:** [_wp05199](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05199.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05199.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05199.pdf.json)

---

### Context and policy background
- G-8 (June 2005) commitments and related figures:
  - Cancel at least $40 billion in debt owed by the world’s poorest nations.
  - 18 nations as a group will be spared $1 billion to $2 billion per year in debt service for loans from lenders such as the World Bank, the IMF, and the African Development Bank.
  - 20 other countries could be eligible for debt relief if they meet targets for good governance and tackling corruption.
  - Pledged to double aid to Africa and envisaged $50 billion in additional aid by 2010, with half of the increase going to Africa.
- Framing of poverty and capital flight:
  - Sachs et al. (2004): poverty traps driven by high transport costs, low agricultural productivity, high disease burdens, unfavorable geopolitical factors, slow diffusion of technology, low savings rates, and high population growth among the rural poor.
  - Boyce and Ndikumana (2001): Africa is a net creditor to the rest of the world in the sense that private assets held abroad as measured by accumulated capital flight exceed the total stock of external debt.

### Research objectives and contributions
- Primary objective: understand the relationship between capital flight and debt, and between capital flight and foreign aid.
- Contribution: first panel data estimates on capital flight for a large set of developing countries including macro policy and institutional variables as regressors.
- Specific aims:
  - Test whether distorted macroeconomic policies and/or weak institutions cause capital flight and high external indebtedness.
  - Investigate capital flight as a mediating channel between weak institutions and poor macroeconomic outcomes.
  - Assess implications for effectiveness of debt relief and aid in increasing domestic savings and reducing poverty.

### Data and methodology
- Sample and scope:
  - 134 developing countries over a 32-year time period from 1970 to 2001 (panel descriptive work spans 1970–2001; econometric samples vary).
- Core empirical specifications:
  - (1) Capital flight = f(institutional quality, macro policies and conditions, foreign financing)
  - (2) Debt accumulation or other foreign financing = g(capital flight, institutional quality, other macro policies and conditions)
- Estimation strategy:
  - Country fixed effects and period fixed effects (Hausman tests reject random effects).
  - 2SLS to address endogeneity in debt–flight relationship; Arellano-Bond and Arellano-Bover GMM used as robustness.
  - Use of predetermined (lagged) variables to mitigate endogeneity.
- Capital flight measurement:
  - Residual definition using balance of payments identity:
    - KFit = ∆Debtit + DFIit + CASit + CHORit
  - Exchange-rate adjustment for long-term debt yields:
    - KFit = ∆Debtadjit + DFIit + CASit + CHORit
  - Misinvoicing treated differently and not adjusted in residual measure (footnote 5 discussion).
- Institutional measures and macro variables:
  - Institutional quality: constraints on executive power (Polity EXCONST), ICRG confidence indicators, World Bank governance indices, Fraser Institute, Milken Institute, Heritage Foundation where available.
  - Macroeconomic variables: interest rate differentials, inflation, growth rates, budget balances, investment, domestic credit growth, foreign reserves, misaligned exchange rates, financial crises, trade openness, stock of debt, short-term and long-term debt, aid, and FDI.
- Instruments for 2SLS (debt and flight equations):
  - Instruments for flight: lagged KF2GDP, budget balance to GDP, growth rates.
  - Instruments for change in debt: lagged trade openness, lagged debt ratio (TOTDEBT2GDP), lagged government expenditure ratio (GOVEXP2GDP).
  - Reported first-stage F-statistics: 58 for debt instruments; 25 for capital flight instruments. J-test fails to reject overidentifying restrictions.

### Timeline and descriptive patterns around peak flight and repatriation
- Alignment: episodes in the top quartile of capital flight and top quartile of repatriation aligned at “time zero” (year of maximum flight/repatriation).
- Median deviations (top-quartile vs. middle half):
  - Growth rate: declines precipitously in the two years prior to maximum capital flight; growth slightly above normal during repatriation episodes.
  - Inflation: downward trend for both episodes, but higher level during flight than during repatriation.
  - Budget balances: slump prior to and during flight.
  - Government foreign borrowing: surges around flight episodes.
- Confidence indicators (ICRG: political, financial, economic, composite):
  - Confidence declines dramatically before, during, and after capital flight; nadir in year of maximum flight.
  - For repatriation, sentiment lower than normal three years before but rises steadily, peaking two years after maximum repatriation.
- Crisis probabilities:
  - Banking crisis probability rises prior to capital flight (from Caprio and Klingebiel (2003) dates).
  - Currency crisis probability (top quartile of exchange market pressure index by construction = 25 percent mean) considerably higher in years before and during capital flight; similar to mean for repatriation episodes.

### Empirical findings — determinants of capital flight (panel and cross-section)
- Cross-section highlights:
  - Politically connected firms (Faccio 2005 measures: Polcon1, Polcon2, Polcon3) positively associated with average capital flight (1992–2001); percentage politically connected explains close to 50 percent of variation in these regressions (high R-squared reported).
  - Fraser Institute’s control of irregular payments indicator negative and significant when controlling for political connections and income per capita.
  - Milken Institute’s banking governance reduces capital flight significantly in countries with politically connected firms.
  - Settlers’ mortality (Acemoglu, Johnson, Robinson 2001) positively associated with average capital flight over 1971–2001 when controlling for per capita GDP (but lower R-squared than political connection regressions).
  - Cross-section samples small in overlap (e.g., 21 countries for Faccio overlap) — motivates panel analysis to control for omitted country effects.
- Panel fixed-effects results:
  - Constraints on executive power (Polity EXCONST) highly significant determinant of capital flight to GDP (KF2GDP).
  - Macroeconomic variables significant even after controlling for institutional quality:
    - Low growth (GRRT), high fiscal deficits (BUDBALGDP), currency crises (CURRCRISIS) → higher capital flight.
    - Domestic credit growth (DOMCRGR) significantly higher two years prior to capital flight → credit booms provide resources for flight.
  - Persistence: lagged KF2GDP significant.
  - Insignificant once fixed effects included: INFLATION; TRADE2GDP; FOREX2GDP; INTRT-USINTRT.
- Interactions and dynamics:
  - Lagged DOMCRGR × country-average governance indicators (control of corruption, political stability, government effectiveness, rule of law, Fraser indices) → interaction terms significantly negative: better institutions attenuate capital flight induced by domestic credit booms.
  - Hausman-Taylor style regressions with time-invariant governance:
    - World Bank governance indices negatively related to capital flight.
    - Fraser Institute variables insignificant in these specifications.
    - Percent politically connected firms positively related to capital flight (consistent with cross-section).

### Empirical findings — debt and the “revolving door”
- Strong contemporaneous positive relationship between increases in external debt and capital flight:
  - OLS/2SLS evidence: 21 cents of each dollar of additional total debt flows back out as capital flight (units in percent of GDP; Table 5 statement).
  - Conversely, one dollar of capital flight draws in 69 cents of new borrowing (Table 5); in 2SLS this increases to 81 cents.
- Short-term debt effects (Table 6):
  - One dollar increase in short-term debt associated with 84 cents of capital flight (OLS, Table 6).
  - In 2SLS, one dollar increase in short-term debt generates 92 cents of capital flight, versus 13 cents for total debt in 2SLS (Table 6 and Table 5 comparisons).
- Channels and interpretation:
  - Debt-fueled capital flight: external borrowing provides liquidity elites siphon abroad or raises expectations of devaluation/taxation.
  - Financing-need channel: capital flight reduces domestic savings and prompts foreign borrowing to fill the gap.
  - Interaction with institutions:
    - Weak institutions increase propensity to accumulate debt via capital flight; constraint on executive power reduces debt accumulation only when capital flight excluded from debt equations — when capital flight included, institution coefficient disappears → suggests capital flight is conduit through which weak institutions raise debt.
    - Using predetermined (lagged) debt, debt fuels subsequent capital flight more in countries with weak institutions and high income inequality.

### Empirical findings — aid and FDI
- Aid:
  - Aid inflows reduce contemporaneous capital flight (Table 8).
  - Aid does not influence capital flight in subsequent year (AID2GNI(-1) insignificant in Table 8).
  - Aid appears exogenous to capital flight after controlling for fixed effects (Hausman test fails to reject exogeneity).
  - Nonlinear relationship: high levels of aid tend to stem capital flight; interaction AID2GNI × EXCONST(-1) strongly negative — good institutions help absorb aid without flight.
  - Aid dynamics: AID2GNI(-1) is strongly persistent (coefficients 0.517 *** and 0.534 *** reported).
  - Contemporaneous capital flight not a significant determinant of aid; capital flight leads to higher aid with a one-year lag.
- FDI:
  - Net FDI inflows reduce capital flight; predetermined FDI similar to contemporaneous effect (Table 10).
  - Evidence of nonlinear FDI effects and FDI × EXCONST interactions: good institutional quality helps absorb FDI without inducing flight.

### Robustness and heterogeneity
- Sample splits by region and income:
  - Regions: Africa, Asia, Western Hemisphere, Transition Economies.
  - Income groups: low income: per capita real GDP ≤ $735; lower middle: $736–$2,935; upper middle: $2,936–$9,075.
- Findings generally robust across regions and income groups.
- Exception: Asia shows some regressions with a significantly positive relationship between growth rate and capital flight, possibly reflecting capital account liberalization and portfolio diversification as economies grow.

### Conclusions and policy-relevant implications
- Main empirical conclusions:
  - Macroeconomic policy variables and conditions significantly influence capital flight even after controlling for country fixed effects and institutional quality.
  - Institutional quality, particularly constraints on executive power (EXCONST), has an independent and significant effect on capital flight.
  - Strong evidence for a “revolving door” between external borrowing and capital flight: both debt-fueled capital flight and financing-need channels operate.
  - Composition of external financing matters:
    - Debt tends to stimulate capital flight.
    - FDI and aid tend to reduce capital flight.
    - Short-term debt accumulation has the most severe impact on capital flight.
  - Capital flight is a mechanism through which weak institutions contribute to macroeconomic volatility and higher debt accumulation; good institutions can partly offset these effects by facilitating absorption of aid/FDI and access to external finance when needed.
- Policy implications suggested:
  - Debt relief and foreign aid initiatives may reduce prospective taxation to finance debt repayments and thereby reduce capital flight; leveraging relief to reduce flight could amplify assistance effects.
  - Foreign aid or debt relief should be complemented by sound macro policies and improved institutional environment to ensure resources are allocated to productive domestic projects and to limit elite extraction and flight.
  - Strengthening constraints on executive power and broader governance reforms can reduce the propensity for capital flight and its adverse macroeconomic consequences.

*Source: _wp05199 - Appendix I provides a summary of the extant literature (source PDF content).*

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

### _wp05199 - References..............................................................................................................

### Context and policy background
- In June 2005, finance ministers of the Group of Eight (G-8) agreed to cancel at least $40 billion in debt owed by the world’s poorest nations.
- Under the G-8 proposal:
  - 18 nations as a group will be spared $1 billion to $2 billion per year in debt service for loans from lenders such as the World Bank, the IMF, and the African Development Bank.
  - 20 other countries could be eligible for debt relief if they meet targets for good governance and tackling corruption.
  - The group pledged to double aid to Africa and envisaged $50 billion in additional aid by 2010, with half of the increase going to Africa.
- Sachs et al. (2004) frame poverty in many poor nations, especially in Africa, as a "poverty trap" driven by high transport costs, low agricultural productivity, high disease burdens, unfavorable geopolitical factors, slow diffusion of technology, low savings rates, and high population growth among the rural poor.
- Boyce and Ndikumana (2001) estimate that Africa is a net creditor to the rest of the world in the sense that private assets held abroad as measured by accumulated capital flight exceed the total stock of external debt.

### Research objectives and contributions
- Primary objective: understand the relationship between capital flight and debt, and between capital flight and foreign aid.
- Contribution: provide the first set of panel data estimates on capital flight for a large set of developing countries that includes both macro policy and institutional variables as regressors.
- Specific research aims:
  - Explore whether distorted macroeconomic policies and/or weak institutions are responsible for capital flight and high external indebtedness.
  - Investigate whether capital flight acts as a mediating channel between weak institutions and poor macroeconomic outcomes.
  - Assess implications for the effectiveness of debt relief and aid in increasing domestic savings and reducing poverty.

### Data and methodology
- Data set includes 134 developing countries over a 32-year time period from 1970 to 2001.
- Methodological notes:
  - Use of panel data estimates on capital flight.
  - Inclusion of macroeconomic variables and indices of institutional quality as regressors.
  - Clarification and construction of the residual measure of capital flight, including an adjustment for changes in debt valuation.
  - Examination of both cross-sectional and panel empirical strategies (see Tables and Appendices listed in the source for detailed econometric specifications and variable lists).

### Literature framing and hypotheses
- Two principal strands in capital flight literature are distinguished: determinants and associations.
- Competing perspectives on development and aid:
  - Sachs et al. (2004): emphasize savings traps and the need for increased foreign and domestic savings to achieve Millennium Development Goals (reduce global poverty by half by 2015).
  - Acemoglu, Robinson, and Johnson (2001, 2003): emphasize weak institutions as the main drag on growth and a source of volatility and crises.
- Hypothesis explored: if capital flight has an institutional cause, then policy prescriptions focusing solely on correcting macro policy distortions may be ineffective without concurrent institutional reform.

### Paper organization (as given in the source)
- Section II: literature on capital flight and the role of institutional quality.
- Section III: research methodology, variables, testable hypotheses, data sources, construction of variables, and residual measure of capital flight (including debt-valuation adjustment).
- Section IV: anatomy of capital flight and repatriation; relation of macro variables and confidence indicators to peak experiences along a timeline.
- Section V: empirical links between capital flight, macroeconomic variables, indices of institutional quality, and debt accumulation.
- Section VI: conclusions.

*Source: _wp05199 - References.*

### Appendix I provides a summary of the extant literature.

### _wp05199 - Appendix I provides a summary of the extant literature.

### Literature on determinants of capital flight
- Macroeconomic policies and outcomes identified as significant determinants: overvalued exchange rates, high budgetary deficits, high inflation, interest rate differentials, and domestic tax and trade policies (Cuddington, 1987; Lessard and Williamson, 1987; Boyce, 1992; Dooley and Kletzer, 1994; Henry 1996; Bhattacharya 1999).
- Political risk and institutional variables increasingly emphasized:
  - Gibson and Tsakalotos (1993): political risk and expected depreciation significant for five European countries.
  - Fatehi (1994): political instability adversely influences FDI in 17 Latin American countries; argues same factors influence capital flight.
  - Lensink, Hermes, and Murinde (1998): political risk matters for cross-sectional capital flight relationships when macro variables excluded.
- Capital flight associated contemporaneously with other adverse macro outcomes:
  - Low growth (Varman-Schneider, 1991).
  - Increased aid inflows (Collier, Hoeffler, and Pattillo, 2004).
  - High external debt (Boyce, 1992; Chipalkatti and Rishi, 2001; Demir, 2004).
  - Financial and currency crises (World Bank, 1998; Moghaddam et al., 2003).
- Institutions and macroeconomic performance literature:
  - North and Thomas (1973), Bardhan (1984) model institutions–performance link.
  - Acemoglu et al. (2003): focus on institutions’ effects on volatility, crises, growth; identify “constraints on the executive” as key institutional quality measure.
- Paper’s framing: connects capital flight literature to institutions literature, positing capital flight may be a byproduct of redistributive strategies by a weakly constrained executive (“robbing the riches”).

### Methodology and data construction
- Sample and panel design:
  - Panel of developing and emerging market countries pooled (panel spans years 1970–2001 in descriptive work; econometric samples vary).
  - F-tests indicate country and period effects required; Hausman (1978) tests indicate random effects inconsistent → use country fixed effects and period fixed effects.
  - 2SLS estimates included to address endogeneity in the debt–flight relationship.
- Core empirical specifications:
  - (1) Capital flight = f(institutional quality, macro policies and conditions, foreign financing)
  - (2) Debt accumulation or other foreign financing = g(capital flight, institutional quality, other macro policies and conditions)
- Macroeconomic variables considered (listed in source): interest rate differentials, inflation, growth rates, budget balances, investment, growth in domestic credit, level of foreign reserves, misaligned exchange rates, financial crises, trade openness, stock of debt, foreign financing inflows including short-term and long-term debt, aid, and FDI.
- Institutional quality measures:
  - Constraints on executive power (Polity IV) and political/economic confidence (ICRG).
  - Also explore World Bank, Fraser Institute, Milken Institute, Heritage Foundation governance indicators where available.
- Data sources:
  - Macroeconomic data: IMF’s International Financial Statistics; supplemented by Asian Development Bank’s Asia: Key Indicators and World Bank’s African Development Indicators where applicable.

### Capital flight definition and construction
- Use the “residual” definition of capital flight (World Bank / Cuddington 1986; Cumby and Levich 1987; Dornbusch 1990; Hermes, Lensink, and Murinde, 2002).
- Residual extracts measured components of the balance of payments identity; residual interpreted as net outflows of debt and equity portfolio investment and outflows of other debt instruments (including foreign bank accounts).
- Residual constructed as (equation preserved exactly):
  - KFit = ∆Debtit + DFIit + CASit + CHORit
    - (presented in source as: ititititit CHORCASDFIDebtKF+++∆= )
- Exchange-rate adjustment for long-term debt changes follows Boyce and Ndikumana (2001) methodology (equations (4) and (5) reproduced in source), producing Debtadj and adjusted residual:
  - KFit = ∆Debtadjit + DFIit + CASit + CHORit
    - (presented in source as: ititititit CHORCASDFIDebtadjKF+++∆= )
- Note: misinvoicing treated differently and not adjusted for in residual measure (discussion in footnote 5).

### Timeline and descriptive patterns around peak flight/repatriation
- Panel: 134 developing and emerging market countries, 1970–2001; episodes in top quartile of capital flight and top quartile of repatriation aligned at “time zero” (year of maximum flight/repatriation).
- Macroeconomic indicators (median deviations from baseline, top-quartile vs. middle half):
  - Growth rate: declines precipitously in the two years prior to maximum capital flight; growth slightly above normal during repatriation episodes.
  - Inflation: downward trend for both, but higher level during flight episodes versus repatriation.
  - Budget balances: slump prior to and during flight.
  - Government foreign borrowing: surges around flight episodes.
- Confidence indicators (ICRG: political, financial, economic, composite):
  - Confidence declines dramatically before, during, and after capital flight; nadir in year of maximum flight.
  - For repatriation, sentiment lower than normal three years before but rises steadily, peaking two years after maximum repatriation.
- Crisis probabilities:
  - Banking crisis probability rises prior to capital flight (constructed from Caprio and Klingebiel (2003) banking crisis dates).
  - Currency crisis probability (top quartile of exchange market pressure index by construction = 25 percent mean) considerably higher in years before and during capital flight; similar to mean for repatriation episodes.

### Empirical findings — cross-section results
- Politically connected firms (Faccio 2005 measures: Polcon1, Polcon2, Polcon3) are positively associated with average capital flight (1992–2001). Percentage political connections explain close to 50 percent of variation in these regressions (high R-squared in Table 1).
- Other cross-section findings:
  - Fraser Institute’s control of irregular payments indicator (higher = less irregular payments) negative and significant when controlling for political connections and income per capita.
  - Milken Institute’s banking governance: higher-quality banking governance significantly reduces capital flight in countries with politically connected firms.
  - Settlers’ mortality (Acemoglu, Johnson, Robinson 2001) positively associated with average capital flight over 1971–2001 when controlling for per capita GDP (but lower R-squared than political connection regressions).
- Caveats: cross-section samples small (e.g., 21 countries for Faccio overlap), potential omitted country effects bias → motivate panel analysis.

### Empirical findings — panel determinants of capital flight
- Institutional quality (Polity EXCONST: constraints on executive power) is highly significant for capital flight to GDP (KF2GDP) in fixed-effects panel regressions (Table 2).
- Macroeconomic determinants remain significant after controlling for institutional quality:
  - Low growth (GRRT), high fiscal deficits (BUDBALGDP), currency crises (CURRCRISIS) → significantly higher capital flight.
  - Domestic credit growth (DOMCRGR) significantly higher two years prior to capital flight → credit booms provide resources for flight.
- Persistence and dynamics:
  - Capital flight is persistent; lagged KF2GDP significant.
  - One-year lags of growth, fiscal balance, currency crisis, and executive constraints also significant with similar magnitudes → authors predominantly use predetermined (lagged) variables to mitigate endogeneity concerns.
- Variables insignificant in panel once fixed effects and other controls included: INFLATION; TRADE2GDP; FOREX2GDP; INTRT-USINTRT.
- Estimation robustness:
  - Fixed effects used due to Hausman tests rejecting random effects.
  - Arellano-Bond and Arellano-Bover GMM implemented as robustness; results qualitatively similar though coefficients differ in magnitude; Arellano-Bover estimates similar to fixed effects with higher significance.
- Interaction of domestic credit growth and institutional quality (Table 3):
  - Lagged DOMCRGR × country-average governance indicators (control of corruption, political stability, government effectiveness, rule of law, Fraser indices) → interaction terms significantly negative.
  - Interpretation: better institutions attenuate capital flight induced by domestic credit booms.

### Empirical findings — time-invariant regressors
- Hausman-Taylor style regressions (Table 4) including time-invariant governance measures:
  - World Bank governance indices (control of corruption, political stability, government effectiveness, rule of law) negatively related to capital flight.
  - Fraser Institute variables insignificant in these panel-with-time-invariant regressions.
  - Percent politically connected firms positively related to capital flight (consistent with cross-section results).

### Empirical findings — debt and the “revolving door”
- Strong contemporaneous positive relationship between increases in external debt and capital flight (Table 5).
  - In OLS/2SLS estimates: evidence that 21 cents of each dollar of additional total debt flows back out as capital flight (units in percent of GDP; Table 5 statement).
  - Conversely, one dollar of capital flight draws in 69 cents of new borrowing (Table 5 statement); in 2SLS this increases to 81 cents.
- Short-term debt matters more (Table 6):
  - One dollar increase in short-term debt associated with 84 cents of capital flight (OLS, Table 6).
  - In 2SLS, one dollar increase in short-term debt generates 92 cents of capital flight, versus 13 cents for total debt in 2SLS (Table 6 and Table 5 comparisons).
- Endogeneity and 2SLS:
  - Hausman tests reject exogeneity of debt in the capital flight equation and of flight in the debt equation → employ 2SLS with instruments:
    - Instruments for flight: lagged ratio of capital flight to GDP, budget balance to GDP, growth rates.
    - Instruments for change in debt: lagged trade openness, lagged debt ratio (TOTDEBT2GDP), lagged government expenditure ratio (GOVEXP2GDP).
    - F-statistics: 58 for debt instruments; 25 for capital flight instruments → authors report strong instruments; J-test fails to reject overidentifying restrictions.
- Channels and interpretation:
  - Debt-fueled capital flight: external borrowing may provide liquidity that elites siphon abroad or raise expectations of devaluation/taxation.
  - Financing-need channel: capital flight reduces domestic savings and prompts foreign borrowing to fill gap.
  - Interaction with institutions:
    - Weak institutions increase propensity to accumulate debt via capital flight; constraint on executive power reduces debt accumulation only when capital flight excluded from debt equations — when capital flight included, institution coefficient disappears → suggests capital flight is conduit through which weak institutions raise debt.
    - Contemporaneous results show debt-fueled flight more pronounced in countries with good institutions and low inequality (plausible endogeneity or timing lags); using predetermined (lagged) debt, debt fuels subsequent capital flight more in countries with weak institutions and high income inequality.

### Empirical findings — aid and FDI
- Aid:
  - Aid inflows reduce contemporaneous capital flight (Table 8).
  - Aid does not appear to influence capital flight in subsequent year (AID2GNI(-1) insignificant in Table 8).
  - Aid appears exogenous to capital flight after controlling for fixed effects (Hausman test fails to reject exogeneity).
  - Nonlinear relationship: high levels of aid tend to stem capital flight; interaction AID2GNI × EXCONST(-1) strongly negative → good institutions help absorb aid without flight.
- Aid determinants (Table 9):
  - AID2GNI(-1) is strongly persistent (coefficient 0.517 *** and 0.534 *** reported).
  - Contemporaneous capital flight not a significant determinant of aid; capital flight only leads to higher aid with a one-year lag.
- FDI:
  - Net FDI inflows reduce capital flight; predetermined FDI similar to contemporaneous effect (Table 10).
  - Nonlinearity and interaction: some evidence of nonlinear FDI effect and FDI × EXCONST interactions; good institutional quality helps absorb FDI without inducing flight.

### Robustness across regions and income groups
- Sample splits: four regions (Africa, Asia, Western Hemisphere, Transition Economies) and three income groups (low income: per capita real GDP ≤ $735; lower middle: $736–$2,935; upper middle: $2,936–$9,075).
- Generally results robust across regions and income groups.
- Notable exception: in Asia, some regressions show a significantly positive relationship between growth rate and capital flight, possibly reflecting capital account liberalization and portfolio diversification as economies grow.

### Conclusions and policy-relevant implications
- Main empirical conclusions:
  - Macroeconomic policy variables and conditions significantly influence capital flight even after controlling for country fixed effects and institutional quality.
  - Institutional quality, particularly constraints on executive power (EXCONST), has an independent and significant effect on capital flight.
  - Strong evidence for a “revolving door” between external borrowing and capital flight: both debt-fueled capital flight and financing-need channels operate.
  - Composition of external financing matters:
    - Debt tends to stimulate capital flight.
    - FDI and aid tend to reduce capital flight.
    - Short-term debt accumulation has the most severe impact on capital flight.
  - Capital flight is a mechanism through which weak institutions contribute to macroeconomic volatility and higher debt accumulation; good institutions can partly offset these effects by facilitating absorption of aid/FDI and access to external finance when needed.
- Policy implications suggested by findings:
  - Debt relief and foreign aid initiatives may reduce prospective taxation to finance debt repayments and thereby reduce capital flight; leveraging relief to reduce flight could amplify assistance effects.
  - Foreign aid or debt relief should be complemented by sound macro policies and improved institutional environment to ensure resources are allocated to productive domestic projects and to limit elite extraction and flight.
  - Strengthening constraints on executive power and broader governance reforms can reduce the propensity for capital flight and its adverse macroeconomic consequences.

*Italic source attribution: _wp05199 - Appendix I provides a summary of the extant literature (source PDF content).*

### APPENDIX I

### APPENDIX I

### A. Determinants literature

- A1. Macroeconomic factors
  - Cuddington (1987): 7 Latin American countries; Time series analysis over 1974–84
    - Finding: External debt flows, lagged capital flight, inflation, exchange rate, and interest rate differentials are significant contributors to capital flight.
  - Dooley (1988): 5 Latin American countries, Philippines; Time series analysis over 1973–86
    - Finding: Inflation was significantly and positively related to capital flight.
  - Boyce (1992): Philippines; Time series analysis over 1962–86
    - Finding: External debt, budget deficits, and interest rate differentials emerge as significant determinants of Philippine capital flight.
  - Hermes and Lensink (1992): 6 sub-Saharan countries; Pooled data analysis over 1976–87
    - Finding: Capital flight is significantly and positively determined by external debt flows and overvalued exchange rates.
  - Henry (1996): 3 Caribbean countries; Time series analysis over 1971–87
    - Finding: External debt, real interest rate differentials, and the unemployment rate are significant causes of capital flight.

- A2. Nonmacroeconomic factors
  - Fatehi (1994): 17 Latin American countries; Stepwise multiple regression analysis over 1950–82
    - Finding: Political disturbances in some Latin American countries are associated with changes in capital flight from these countries.
  - Nyoni (2000): Tanzania; Time series analysis over 1973–92
    - Finding: Lagged capital flight, real growth rates, interest and exchange rate differentials were significant determinants of capital flight. A political shock dummy in the flight equation had no statistically significant effect.
  - Lensik, Hermes, and Murinde (2000): 84 LDCs; Extreme bounds analysis and cumulative distribution functions estimated for a cross-sectional sample over 1971–90
    - Finding: Political instability, wars, and democracy and political freedom significantly determine capital flight.
  - Hermes and Lensink (2001): 84 LDCs; Cumulative distribution functions estimated for a cross-sectional sample over 1971–91
    - Finding: In addition to macro variables, political instability, civil liberties, and macro policy uncertainty have a significant and positive impact on flight.

### B. Associations literature

- B1. Debt-flight revolving door
  - Boyce (1992): Philippines; Time series analysis over 1962–86
    - Finding: Direct causal linkage between external debt and flight.
  - Chipalkatti and Rishi (2001): India; Time series analysis over 1971–97
    - Finding: Contemporaneous bi-directional causality between external debt and flight.
  - Boyce and Ndikumana (2001): 25 low-income sub-Saharan African countries; Time series and cross-sectional analysis over 1970–96
    - Finding: Significant linkages exist between external borrowing and capital flight.
  - Demir (2004): Turkey; Time series analysis over 1974–2000
    - Finding: There is a contemporaneous bi-directional causality between debt and flight.

- B2. Aid-flight association
  - Collier, Hoeffler, Pattillo (2004): 48 non-OECD countries; Nonlinear estimation (control function approach)
    - Finding: Aid substantially reduces capital flight.

*Source: APPENDIX I, _wp05199 - APPENDIX I*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05199.pdf_