## _wp11280 - Section 2 describes the data and the identification

## Source details

**Canonical URL:** [_wp11280 - Section 2 describes the data and the identification](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11280.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11280.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11280.pdf.json)

---

### II. DATA — datasets and descriptive statistics
- Datasets used to identify debt crisis episodes:
  - Laeven and Valencia (2008): identify 63 crisis episodes; distribution by period: seven episodes in 1970–1979, 41 between 1980 and 1989, seven in 1990–1999, and eight after 1999.
  - De Paoli et al. (2006): identify 39 sovereign default episodes over 1970–2000; criterion includes arrears on principal reaching at least 15 percent of total commercial debt outstanding and/or rescheduling with private creditors as listed in World Bank’s Global Development Finance.
  - Reinhart et al. (2003): identify 31 debt crisis episodes over 1970–2001.
  - Detragiache and Spillimbergo (2001): cover 54 episodes; defaults when arrears of principal on external obligations to commercial creditors exceed 5 percent of total commercial debt outstanding (excluding episodes within four years of previous defaults) and/or rescheduling with private creditors listed in World Bank’s Global Development Finance.
  - Levy-Yeyati and Panizza (2011): identify 20 default episodes over 1980–2003 (excluding episodes within three years of previous defaults).
- Descriptive statistics (first-row averages highlighted):
  - At time of debt crises (average, first row): gross public debt-to-GDP ≈ 80 percent; public foreign gross debt-to-GDP ≈ 55 percent; GDP growth ≈ –2 percent.
  - Table 2 dataset-specific highlights (averages and ranges preserved):
    - LV: N. Crises 63; Debt over GDP Average 78.3 Max 119.4 Min 34.4 S.D. 25.5; Foreign debt over GDP Average 55.9 Max 86.3 Min 26.5 S.D. 19.6; GDP Growth Average -2.1 Max 7.5 Min -14.4 S.D. 5.1.
    - DHS: N. Crises 39; Debt over GDP Average 111.9 Max 166.6 Min 81.0 S.D. 37.6; Foreign debt over GDP Average 59.7 Max 95.9 Min 7.6 S.D. 32.7; GDP Growth Average -2.5 Max 10.6 Min -32.1 S.D. 7.7.
    - RRS: N. Crises 31; Debt over GDP Average 68.6 Max 85.2 Min 47.4 S.D. 19.3; Foreign debt over GDP Average 53.0 Max 65.4 Min 39.4 S.D. 13.0; GDP Growth Average -2.2 Max 5.9 Min -14.4 S.D. 5.4.
    - DS: N. Crises 54; Debt over GDP Average 63.8 Max 142.0 Min 10.8 S.D. 39.7; Foreign debt over GDP Average 41.0 Max 70.6 Min 6.0 S.D. 23.3; GDP Growth Average 0.7 Max 15.4 Min -14.4 S.D. 6.4.
    - LY: N. Crises 21; Debt over GDP Average 64.5 Max 96.6 Min 21.0 S.D. 26.1; Foreign debt over GDP Average 46.7 Max 78.4 Min 21.0 S.D. 20.9; GDP Growth Average -2.2 Max 6.5 Min -14.1 S.D. 5.3.
    - Average across datasets: Debt over GDP 77.4; Foreign debt over GDP 51.2; GDP Growth -1.7.
- Additional data sources:
  - Banking and currency crises: Laeven and Valencia (2008) (systemic banking crises 1970–2007; currency crisis defined as nominal depreciation of at least 30 percent and at least a 10 percent increase in rate of depreciation compared to year before).
  - Real GDP: World Bank Economic Indicators.
  - Public (domestic and foreign) debt: Panizza (2008).
- Definition note:
  - Foreign debt is defined as public debt issued in foreign countries and under the jurisdiction of a foreign court.

### III. EMPIRICAL ANALYSIS — methodology overview
- Objectives:
  - Assess short-term and medium-term effects of debt crises on output.
  - Control for reverse causality and identification versus banking and currency crises.
  - Investigate impact of total and foreign public debt-to-GDP ratios and existence of debt thresholds.
  - Extend analysis to response of output up to 8 years after a debt crisis.
- Core short-term empirical specification (Equation 1):
  - Dependent variable: log of real GDP for country i at time t.
  - Key regressor: debt crisis dummy (1 if debt crisis occurred in country i at time t; 0 otherwise).
  - Controls (vector X): trade openness (share of total exports and imports in GDP), population growth, private credit growth, real exchange rate growth, initial (lagged) level of GDP, and two lags of real GDP growth.
  - Estimation: two-step GMM-system estimator (Roodman’s xtabond2) with Windmeijer standard errors; predetermined/endogenous instrumenting as specified.
- Robustness and diagnostic checks:
  - Hansen J-test p-value range across specifications: 0.3 to 1.
  - Arellano-Bond test for second-order autocorrelation p-value range: 0.2 to 1.
  - Addressed instrument proliferation: difference-Hansen-test, collapsing instruments, checking GMM-difference estimator.
  - Additional checks: exclusion of crisis observations following positive-to-negative growth transitions; focusing on crises during contemporaneous and lagged positive output gap (HP filter smoothness parameter = 100); using predicted default probabilities as regressor (logit-based; point estimates sensitive: range from 1 to 25 percentage points).

### III.A Short Term — main findings and robustness
- Main short-term effect estimates:
  - Baseline two-step GMM (Table 3, column I): debt crises reduce contemporaneous output growth by about 6 percentage points (reported coefficient -5.566).
  - Range across specifications (Table 3, columns II–VII): estimated impact ranges from about 5 to 6 percentage points (examples: -5.384, -5.529, -6.065, -6.412).
  - Control variables typically positive and often statistically significant: trade openness, population growth, credit growth, and first lag of real GDP growth.
- Reverse causality robustness:
  - Excluding observations with growth_t < 0 and growth_t-1 > 0: debt crises remain statistically significant and negative.
  - Focusing on debt crises in periods of contemporaneous and lagged positive output gap: estimated effect −7.5 percent (Table 5, Column IV).
  - Focusing on crises when contemporaneous growth did not slow down: estimated effect −9.3 percent (Table 5, Column V).
  - Caveat: results for crises in “good times” may reflect harsher market punishment (Grossman and Huyck 1988).
- Alternative estimation and datasets:
  - Using alternative debt-crisis datasets: point estimates range from 5 to 10 percentage points (Table 6: LV -5.566; DHS -5.096; LP -9.984; DS -7.143; RRS -9.319).
  - Adding terms of trade and investment-to-GDP controls: these variables statistically insignificant; effect of debt crises changes only slightly and not significantly (Table 7).
  - OLS re-estimation: OLS estimate -4.316 (Table 7, column III).
- Comparison to previous literature:
  - Baseline estimate (≈6 percentage points) higher than some previous studies but not statistically different; cannot reject hypothesis that baseline differs from lowest prior estimate (0.6) found in Sturzenegger (2004).

### Debt crises versus currency and banking crises — identification of marginal effects
- Specification estimating marginal contributions of debt, currency, banking, twin, and triple crises (Equation 2 / Equation 3):
  - Twin crises defined: onset of a given crisis occurs two years before, during, or two years after onset of another type of crisis.
  - Triple crises defined similarly across three crisis types within two years window.
- Findings:
  - Debt crises significantly reduce output growth with estimated impacts ranging from 5 to 8 percentage points across specifications (Table 8: examples -4.963, -5.435, -8.913, -8.064, -8.740).
  - Full specification suggests effect of debt crises is more detrimental than currency or banking crises.
  - Among twin and triple dummies, only the twin banking-currency crisis dummy is negative and statistically significant.
  - Results robust to different year windows (one year and three years).
- Isolation approach:
  - Restricting to debt crises not preceded/followed by banking or currency crises within two years reduces sample to 20 episodes; estimated contemporaneous output reduction ≈ 8 percentage points (Table 8, Column V).

### Debt thresholds — nonlinear effects of high debt on growth
- Motivating descriptive evidence referenced:
  - Reinhart and Rogoff (2010a): gross public debt-to-GDP > 90 percent associated with median growth falling by 1 percentage point; external debt reaching 60 percent of GDP associated with annual growth decline ≈ 2 percentage points.
- Threshold tests (Table 9) — total gross debt-to-GDP:
  - No statistical evidence of a linear relationship between growth and debt (Table 9, panel I linear Debt_GDP t -0.011 (-1.52)).
  - Output reduced by about 1.8 percentage points when debt-to-GDP ratio exceeds 70 percent (Table 9, II >70% -1.776 (-2.62)***).
  - 80 and 90 percent thresholds associated with decline in output growth greater than 2 percentage points (Table 9, III >80% -2.546 (-3.14)***; IV >90% -2.239 (-2.79)***).
  - Higher thresholds not contributing significantly to additional negative effects beyond these.
- Foreign debt-to-GDP thresholds (Table 9):
  - Weak statistical evidence of a linear relationship (V linear Foreign debt GDP t -0.011 (-1.72)*).
  - Output growth reduced by about 2.4 percentage points when foreign debt-to-GDP exceeds 80 percent (VIII >80% -2.418 (-2.09)**).
  - Lower thresholds (60 and 70 percent) show weaker significance (VI >60% -1.176 (-1.79)*; VII >70% -1.538 (-1.67)*).
  - Higher thresholds beyond 80 percent not contributing significantly to additional negative effects.
- Joint inclusion of threshold and crisis dummies:
  - Both dummies statistically significant.
  - Debt crises reduce output by about 4–5 percentage points when both included.
  - Higher debt levels (total and foreign) reduce output by about 1.5–2 percentage points.
- Interpretation:
  - High debt has a negative effect but considerably lower than effect of debt crises, implying that mechanisms beyond large debt burdens (e.g., crisis-specific channels) drive much of the negative output impact.

### III.B Medium Term — local projections, ARDL, and persistence of losses
- Methodology:
  - Local projections (Jorda 2005; Teuling and Zubanov 2010): estimate impulse response functions (IRFs) for k = 0,...,8 (Equation 3); lags chosen = 2; White robust standard errors; country fixed effects and time trends included.
  - ARDL (4,4) specification (Equation 4) used as additional robustness; IRFs obtained by simulating a one-year crisis and Monte Carlo simulations (1,000 trials) for 95 percent confidence bands.
  - Tests for bias from fixed effects interaction with crisis arrival rates addressed by re-estimating without country fixed effects.
- Medium-term findings:
  - Local projections (Figure 1): debt crises have long-lasting effects; eight years after occurrence output lower by about 10 percent.
  - Robustness to controls: including common time trend and time fixed effects leaves results statistically significant and broadly unchanged (Figure 2 panels A and B).
  - Excluding country fixed effects: negligible bias; eight-year impact ≈ 10 percent lower output (Figure 3).
  - ARDL (4,4) OLS and GMM estimates (Figure 4): eight years after occurrence output lower by about 9–12 percent.
  - Additional robustness: restricting sample to (i) debt crises with contemporaneous non-negative growth and (ii) crises not associated with banking or currency crises in the eight years before, during, or after onset — both corroborate negative medium-term impacts (Figure 5 Panels A and B).
- Note on reverse causality:
  - Reverse causality is less of a concern for medium-term (k up to 8) estimates because changes in output are measured for subsequent periods t+1 to t+8.

### IV. CONCLUSIONS AND ISSUES FOR FUTURE RESEARCH — synthesis and research directions
- Key empirical conclusions:
  - Short term:
    - Debt crises reduce contemporaneous output growth by about 6 percentage points in baseline two-step GMM estimates (Table 3 baseline -5.566).
    - Estimates robust across specifications and checks; range across datasets and specifications approximately 5 to 10 percentage points.
    - For debt crises not accompanied by banking/currency crises within two years, contemporaneous output reduction ≈ 8 percentage points.
  - Medium term:
    - Debt crises associated with persistent output losses; eight years after occurrence output about 10 percent lower (local projections) with ARDL range 9–12 percent.
  - Debt levels vs crises:
    - Exceeding thresholds: gross debt-to-GDP > 70 percent associated with ≈ 1.8 percentage point lower growth; foreign debt-to-GDP > 80 percent associated with ≈ 2.4 percentage point lower growth.
    - Effect of high debt levels materially smaller than effect of debt crises.
- Policy implications (implied by magnitudes):
  - Large and persistent output losses from debt crises underscore significant macroeconomic risks of default and debt restructurings; policymakers should be alerted by magnitudes of estimated losses.
- Suggested extensions and research avenues:
  - Empirically examine determinants of heterogeneous output responses across countries and crisis episodes (differentiate debt restructuring types: debt versus flow restructuring; preemptive/voluntary debt exchanges vs arrears or outright default).
  - Expand analysis of debt thresholds using non-parametric or semi-parametric approaches.
  - Investigate interactions between public (total and foreign) debt shares and other factors: trade openness, domestic saving, financial integration, financial development, measures of perceived country risks.

*Source: _wp11280 - Section 2 describes the data and the identification (PDF chapter/section).*

### Section 2 describes the data and the identification

### _wp11280 - Section 2 describes the data and the identification

### II. DATA — datasets and descriptive statistics
- Datasets used to identify debt crisis episodes:
  - Laeven and Valencia (2008): identify 63 crisis episodes; distribution by period: seven episodes in 1970–1979, 41 between 1980 and 1989, seven in 1990–1999, and eight after 1999. Sources: Beim and Calomiris (2001), World Bank (2002), Sturzenegger and Zettelmeyer (2006), and IMF Staff Reports.
  - De Paoli et al. (2006): identify 39 sovereign default episodes over 1970–2000; criterion includes arrears on principal reaching at least 15 percent of total commercial debt outstanding and/or rescheduling with private creditors as listed in World Bank’s Global Development Finance.
  - Reinhart et al. (2003): identify 31 debt crisis episodes over 1970–2001 using Beim and Calomiris (2001) and Standard and Poor’s Credit Week.
  - Detragiache and Spillimbergo (2001): cover 54 episodes; defaults when arrears of principal on external obligations to commercial creditors exceed 5 percent of total commercial debt outstanding (excluding episodes within four years of previous defaults) and/or rescheduling with private creditors listed in World Bank’s Global Development Finance.
  - Levy-Yeyati and Panizza (2011): identify 20 default episodes over 1980–2003 (excluding episodes within three years of previous defaults); classify beginning years of foreign currency bank or bond debt default using Standard and Poor’s Credit Week, World Bank’s Global Development Finance, and the financial press.
- Descriptive statistics (Table 2, first row highlighted):
  - At time of debt crises (average, first row): gross public debt-to-GDP ≈ 80 percent; public foreign gross debt-to-GDP ≈ 55 percent; GDP growth ≈ –2 percent.
  - Noted: considerable dispersion around these averages.
- Additional data sources:
  - Banking and currency crises episodes: Laeven and Valencia (2008) (systemic banking crises 1970–2007; currency crisis defined as nominal depreciation of at least 30 percent and at least a 10 percent increase in rate of depreciation compared to year before).
  - Real GDP: World Bank Economic Indicators.
  - Public (domestic and foreign) debt: Panizza (2008).
- Definition note:
  - Foreign debt is defined as public debt issued in foreign countries and under the jurisdiction of a foreign court.

### III. EMPIRICAL ANALYSIS — methodology overview
- Objectives:
  - Assess short-term and medium-term effects of debt crises on output.
  - Control for reverse causality and identification versus banking and currency crises.
  - Investigate impact of total and foreign public debt-to-GDP ratios and existence of debt thresholds.
  - Extend analysis to response of output up to 8 years after a debt crisis.
- Core short-term empirical specification (Equation 1):
  - Dependent variable: log of real GDP for country i at time t.
  - Key regressor: debt crisis dummy (1 if debt crisis occurred in country i at time t; 0 otherwise).
  - Controls (vector X): trade openness (share of total exports and imports in GDP), population growth, private credit growth, real exchange rate growth, initial (lagged) level of GDP, and two lags of real GDP growth.
  - Estimation: two-step GMM-system estimator (Roodman’s xtabond2) with Windmeijer standard errors; predetermined/endogenous instrumenting as specified.
- Robustness and diagnostic checks:
  - Hansen J-test p-value range across specifications: 0.3 to 1 (cannot reject validity of instruments).
  - Arellano-Bond test for second-order autocorrelation p-value range: 0.2 to 1 (cannot reject no second-order serial correlation).
  - Addressed instrument proliferation by: including difference-Hansen-test, collapsing instruments, and checking GMM-difference estimator.
  - Additional checks: exclusion of crisis observations following positive-to-negative growth transitions; focusing on crises during contemporaneous and lagged positive output gap (HP filter smoothness parameter = 100); using predicted default probabilities as regressor (logit-based; point estimates sensitive: range from 1 to 25 percentage points).

### III.A Short Term — main findings and robustness
- Main short-term effect estimates:
  - Baseline two-step GMM (column I, Table 3): debt crises reduce contemporaneous output growth by about 6 percentage points.
  - Range across specifications (columns II–VII, Table 3): estimated impact ranges from about 5 to 6 percentage points.
  - Control variables typically positive and (most of the time) statistically significant: trade openness, population growth, credit growth, and first lag of real GDP growth.
- Reverse causality robustness:
  - Excluding observations with growth_t < 0 and growth_t-1 > 0: debt crises remain statistically significant and negative (Columns II and III, Table 5).
  - Focusing on debt crises occurring in periods of contemporaneous and lagged positive output gap: estimated effect −7.5 percent (Column IV, Table 5).
  - Focusing on debt crises occurring when contemporaneous growth did not slow down: estimated effect −9.3 percent (Column V, Table 5).
  - Caveat: results for crises in “good times” may reflect harsher market punishment (Grossman and Huyck 1988).
- Alternative estimation and datasets:
  - Using alternative debt-crisis datasets (Section 2 alternatives): point estimates range from 5 to 10 percentage points (Table 6); differences likely reflect heterogeneous country responses and crisis severities, not statistically significant.
  - Adding terms of trade and investment-to-GDP controls: these variables statistically insignificant; effect of debt crises changes only slightly and not significantly (Table 7).
  - Re-estimating Equation 1 using OLS (comparison with previous studies): lower impact but difference not statistically significant (Table 7).
- Comparison to previous literature:
  - Baseline estimate (≈6 percentage points) higher than some previous studies but not statistically different; cannot reject hypothesis that baseline differs from lowest prior estimate (0.6) found in Sturzenegger (2004).

### Debt crises versus currency and banking crises — identification of marginal effects
- Specification estimating marginal contributions of debt, currency, banking, twin, and triple crises (Equation 2 / Equation 3):
  - Twin crises defined: onset of a given crisis occurs two years before, during, or two years after onset of another type of crisis.
  - Triple crises defined similarly across three crisis types within two years window.
- Findings:
  - Debt crises significantly reduce output growth with estimated impacts ranging from 5 to 8 percentage points across different specifications (Table 8, Columns I–IV).
  - Full specification suggests effect of debt crises is more detrimental than currency or banking crises.
  - Among twin and triple dummies, only the twin banking-currency crisis dummy is negative and statistically significant.
  - Results robust to different year windows (one year and three years).
- Isolation approach:
  - Restricting to debt crises not preceded/followed by banking or currency crises within two years reduces sample to 20 episodes; estimated contemporaneous output reduction ≈ 8 percentage points (Column V, Table 8).

### Debt thresholds — nonlinear effects of high debt on growth
- Motivating evidence: Reinhart and Rogoff (2010a) descriptive findings referenced: gross public debt-to-GDP > 90 percent associated with median growth falling by 1 percentage point; external debt reaching 60 percent of GDP associated with annual growth decline ≈ 2 percentage points.
- Threshold tests (Table 9):
  - Total gross debt-to-GDP:
    - No statistical evidence of linear relationship between growth and debt.
    - Output reduced by about 1.8 percentage points when debt-to-GDP ratio exceeds 70 percent.
    - 80 and 90 percent thresholds associated with decline in output growth greater than 2 percentage points.
    - Higher thresholds not contributing significantly to additional negative effects.
  - Foreign debt-to-GDP:
    - Weak statistical evidence of linear relationship.
    - Output growth reduced by about 2.4 percentage points when foreign debt-to-GDP exceeds 80 percent.
    - Lower thresholds (60 and 70 percent) not statistically significant at 5 percent.
    - Higher thresholds not contributing significantly to additional negative effects.
- Joint inclusion of threshold and crisis dummies:
  - Both dummies statistically significant.
  - Debt crises reduce output by about 4–5 percentage points when both included.
  - Higher debt levels (total and foreign) reduce output by about 1.5–2 percentage points.
- Interpretation:
  - High debt has a negative effect but considerably lower than effect of debt crises, implying that mechanisms beyond large debt burdens (e.g., crisis-specific channels) drive much of the negative output impact.

### III.B Medium Term — local projections, ARDL, and persistence of losses
- Methodology:
  - Local projections (Jorda 2005; Teuling and Zubanov 2010): estimate impulse response functions (IRFs) from regressions for k = 0,...,8 (Equation 3); lags chosen = 2; White robust standard errors; country fixed effects and time trends included.
  - ARDL (4,4) specification (Equation 4) used as additional robustness; IRFs obtained by simulating a one-year crisis and Monte Carlo simulations (1,000 trials) for 95 percent confidence bands.
  - Tests for bias from fixed effects interaction with crisis arrival rates addressed by re-estimating without country fixed effects.
- Medium-term findings:
  - Local projections (Figure 1): debt crises have long-lasting effects; eight years after occurrence output lower by about 10 percent.
  - Robustness to controls: including common time trend and time fixed effects (Figure 2 panels A and B) leaves results statistically significant and broadly unchanged.
  - Excluding country fixed effects (Figure 3): negligible bias; eight-year impact ≈ 10 percent lower output.
  - ARDL (4,4) OLS and GMM estimates (Figure 4): eight years after occurrence output lower by about 9–12 percent.
  - Additional robustness: restricting sample to (i) debt crises with contemporaneous non-negative growth and (ii) crises not associated with banking or currency crises in the eight years before, during, or after onset — both corroborate negative medium-term impacts (Figure 5 Panels A and B).
- Note on reverse causality:
  - Reverse causality is less of a concern for medium-term (k up to 8) estimates because changes in output are measured for subsequent periods t+1 to t+8.

### IV. CONCLUSIONS AND ISSUES FOR FUTURE RESEARCH — synthesis and research directions
- Key empirical conclusions:
  - Short term:
    - Debt crises reduce contemporaneous output growth by about 6 percentage points in baseline two-step GMM estimates.
    - Estimates robust across specifications and checks; range across datasets and specifications approximately 5 to 10 percentage points.
    - For debt crises not accompanied by banking/currency crises within two years, contemporaneous output reduction ≈ 8 percentage points.
  - Medium term:
    - Debt crises associated with persistent output losses; eight years after occurrence output about 10 percent lower (robust across local projections and ARDL, with ARDL range 9–12 percent).
  - Debt levels vs crises:
    - Exceeding thresholds: gross debt-to-GDP > 70 percent associated with ≈ 1.8 percentage point lower growth; foreign debt-to-GDP > 80 percent associated with ≈ 2.4 percentage point lower growth.
    - Effect of high debt levels materially smaller than effect of debt crises.
- Policy implications (implied by magnitudes):
  - Large and persistent output losses from debt crises underscore significant macroeconomic risks of default and debt restructurings; policymakers should be alerted by magnitudes of estimated losses.
- Suggested extensions and research avenues:
  - Empirically examine determinants of heterogeneous output responses across countries and crisis episodes (differentiate debt restructuring types: debt versus flow restructuring; preemptive/voluntary debt exchanges vs arrears or outright default).
  - Expand analysis of debt thresholds using non-parametric or semi-parametric approaches.
  - Investigate interactions between public (total and foreign) debt shares and other factors: trade openness, domestic saving, financial integration, financial development, measures of perceived country risks.

*Source: _wp11280 - Section 2 describes the data and the identification (PDF chapter/section).*

### REFERENCES

### _wp11280 - REFERENCES

### References cited
- Beim, D. and C. Calomiris, 2001, Emerging Financial Markets, Appendix to Chapter 1 (New York: McGraw-Hill/Irwin Publishers).
- Borensztein, E., and U. Panizza, 2009, “The Costs of Sovereign Default,” Staff Papers, International Monetary Fund, Vol. 56(4), pp. 683-741.
- Cai, X., and W. J. Den Haan, 2009, “Predicting Recoveries and the Importance of Using Enough Information,” CEPR Discussion Paper No. 7508 (London: Centre for Economic Policy Research).
- Caprio, G., and others, 2005, “Appendix: Banking Crisis Database,” in Systemic Financial Crises: Containment and Resolution, ed. by Patrick Honohan and Luc Laeven (Cambridge, U.K.: Cambridge University Press).
- Carner, M, T. Grennes, and F. Koeheler-Geib, 2010, “Finding the Tipping Point When Sovereign Debt Turns Bad,” Policy Research Working Paper No. 5391 (Washington: World Bank).
- Cerra, V., and S. Saxena, 2008, “Growth Dynamics: The Myth of Economic Recovery,” American Economic Review, Vol. 98, pp. 439-57.
- Checherita, C., and P. Rother, 2010, “The Impact of High and Growing Government Debt on Economic Growth: An Empirical Investigation for the Euro Area” ECB Working Papers forthcoming.
- Chiang, G., and J. Coronado, 2005, “A Two Step Approach to Assess the Cost of Default for Latin America,” www.urrutiaelejalde.org/Summer-School/2005/papers/coronadochiang.pdf.
- De Paoli, B., G. Hoggarth, and V. Saporta, 2006, “Costs of Sovereign Defaults,” Bank of England Financial Stability Paper No. 1, July.
- De Paoli, B., G. Hoggarth, and V. Saporta, 2009, “Output Costs of Sovereign Crises: Some Empirical Estimates,” Bank of England Working Paper No. 362.
- Detragiache E., and A. Spilimbergo, 2001, “Crises and Liquidity– Evidence and Interpretation,” IMF Working Papers 01/2 (Washington: International Monetary Fund).
- Eichengreen, B., R Hausmann, and U. Panizza, 2003, “Currency Mismatches, Debt Intolerance, and Original Sin: Why They Are Not the Same and Why It Matters,” NBER Working Paper No. 10036 (Cambridge, Massachusetts: National Bureau of Economic Research).
- Furceri, D., and A. Mourougane 2009, “The Effect of Financial Crises on Potential Output: New Empirical Evidence form OECD Countries,” OECD Working Paper No. 699 (Paris: Organisation for Economic Co-operation and Development).
- Furceri, D., and A. Zdzienicka, 2011, “The Real Effects of Financial Crises in the European Transition Economies,” Economics of Transition,17, 1 -25.
- G. Gelos, R. Sahay, and G. Sandleris, 2011, “Sovereign Borrowing by Developing Countries: What Determines Market Access?” Journal of International Economics, Vol. 83(2), pp 243-54.
- Grossman, H. I. and J. B. Huyck, 1988, “Sovereign Debt as a Contingent Claim: Excusable Default, Repudiation, and Reputation,” The American Economic Review, Vol. 78 (5), pp 1088-97.
- Hutchison, M., and N. Ilan, 2005, “How Bad Are Twins? Output Costs of Currency and Banking Crises,” Journal of Money, Credit and Banking, Vol. 37(4), pp. 725-52.
- Jorda, O., 2005, “Estimation and Inference of Impulse Responses by Local Projections,” American Economic Review, Vol. 95, No. 1, pp. 161–82.
- Krugman, P., 1999, “Balance Sheets, the Transfer Problem, and Financial Crises,” International Tax and Public Finance, Springer, Vol. 6(4), pp. 459-72.
- Kumar, M.S., and J.Woo, 2010, “Public Debt and Growth,” IMF Working Paper 10/174 (Washington: International Monetary Fund).
- Leaven, L., and F. Valencia, 2008, “Systemic Banking Crises: a New Database,” IMF Working Paper 08/224 (Washington: International Monetary Fund).
- Levine, R., and D. Renelt, 1992, “A Sensitivity Analysis of Cross-country Growth Regressions,” American Economic Review, Vol. 82, pp. 942–4.
- Levy-Yeyati, E., and U. Panizza, 2011, “The Elusive Costs of Sovereign Defaults,” Journal of Development Economics, Vol. 94 (1), pp. 95-105.
- Panizza, U., 2008, “The External Debt Contentious Six Years after the Monterrey Consensus.” G-24 Discussion Paper No. 51.
- Panizza, U., F. Sturzenegger, and J. Zettelmeyer, 2009, “The Economics and Law of Sovereign Debt and Default,” Journal of Economic Literature, Vol. 47(3), pp. 651-98.
- Reinhart C., K. Rogoff, and M. Savastano, 2003, “Debt Intolerance,” NBER Working Paper No. 9908 (Cambridge, Massachusetts: National Bureau of Economic Research).
- Reinhart C., and K. Rogoff, 2010a, “Growth in a Time of Debt,” American Economic Review, Vol. 100(2), pp. 573-78.
- Reinhart C., and K. Rogoff, 2010b “From Financial Crash to Debt Crisis,” NBER Working Paper No. 15795 (Cambridge, Massachusetts: National Bureau of Economic Research).
- Richmond, C., and D. Dias, 2008, “Duration of Capital Market Exclusion: Stylized Facts and Determining Factors” http://personal.anderson.ucla.edu/christine.richmond/Marketaccess_0808.pdf.
- Romer, C., and D. Romer, 1989, “Does Monetary Policy Matter? A New Test in the Spirit of Friedman and Schwartz,” NBER Macroeconomics Annual, (4) pp. 121-70.
- Roodman, D., 2009a, “How to do xtabond2: An Introduction to Difference and System GMM in Stata,” Stata Journal, Vol. 9(1), pp. 86-136.
- Roodman, D., 2009b, “A Note on the Theme of Too Many Instruments,” Oxford Bulletin of Economics and Statistics, Vol. 71 (1), pp. 135-58.
- Rose, A., 2005, “One Reason Countries Pay Their Debts: Renegotiation and International Trade,” Journal of Development Economics, Vol. 77(1), pp. 189–206.
- Sala-i-Martin, X., 1997, “I Just Run Two Million Regressions,” American Economic Review, Vol 87, pp. 178-83.
- Sala-i-Martin, X., G. Doppelhofer, and R. Miller, 2004, “Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach,” American Economic Review, Vol. 94(4), 813-35.
- Sturzenegger, F., 2004, “Toolkit for the Analysis of Debt Problems,” Journal of Restructuring Finance, Vol. 1(1), pp. 201–03.
- Sturzenegger, F., and J. Zettelmeyer, 2006, Debt Defaults and Lessons from a Decade of Crises, Table 1 in Chapter 1 (Cambridge, Massachusetts: MIT Press).
- Teulings, C.N., and N. Zubanov, 2010, “Economic Recovery a Myth? Robust Estimation of Impulse Responses,” CEPR Discussion Papers No. 7800 (London: Center for Economic Policy Research).
- World Bank, 2002, Global Development Finance, Appendix on Commercial Debt Restructuring, (Washington, D.C.: World Bank).

### Figures (effect of debt crises on output)
- Figure 1. The Effect of Debt Crises on Output—Baseline.
- Figure 2. The Effect of Debt Crises on Output—Robustness Checks
  - A. Common Time Trend
  - B. Time Fixed Effects
  - Note: Dotted lines represent 95 percent confidence bands.
- Figure 3. The Effect of Debt Crises on Output—Robustness for Possible Bias.
- Figure 4. The Effect of Debt Crises on Output—ARDL
  - A. OLS
  - B. GMM
  - Note: Dotted lines represent 95 percent confidence bands.
- Figure 5. The Effect of Debt Crises on Output—Exogeneity and Identification
  - A. Debt Crises with Non-contemporaneous Fall in Output
  - B. Debt Crises Non-overlapping with Other Crises
  - Note: Dotted lines represent 95 percent confidence bands.

### Table highlights and key statistics
- Table 1. Debt Crises Episodes: lists country-year episodes compiled from various datasets. LV= Laeven and Valencia (2008); DHS=De Paoli et al. (2006); RRS= Reinhart et al. (2003); DS=Detragiache and Spillimbergo (2001); LP=Levy-Yeyati and Panizza (2011).
- Table 2. Descriptive Statistics (by dataset)
  - LV: N. Crises 63; Debt over GDP (percent) Average 78.3 Max 119.4 Min 34.4 S.D. 25.5; Foreign debt over GDP (percent) Average 55.9 Max 86.3 Min 26.5 S.D. 19.6; GDP Growth (percent) Average -2.1 Max 7.5 Min -14.4 S.D. 5.1.
  - DHS: N. Crises 39; Debt over GDP (percent) Average 111.9 Max 166.6 Min 81.0 S.D. 37.6; Foreign debt over GDP (percent) Average 59.7 Max 95.9 Min 7.6 S.D. 32.7; GDP Growth (percent) Average -2.5 Max 10.6 Min -32.1 S.D. 7.7.
  - RRS: N. Crises 31; Debt over GDP (percent) Average 68.6 Max 85.2 Min 47.4 S.D. 19.3; Foreign debt over GDP (percent) Average 53.0 Max 65.4 Min 39.4 S.D. 13.0; GDP Growth (percent) Average -2.2 Max 5.9 Min -14.4 S.D. 5.4.
  - DS: N. Crises 54; Debt over GDP (percent) Average 63.8 Max 142.0 Min 10.8 S.D. 39.7; Foreign debt over GDP (percent) Average 41.0 Max 70.6 Min 6.0 S.D. 23.3; GDP Growth (percent) Average 0.7 Max 15.4 Min -14.4 S.D. 6.4.
  - LY: N. Crises 21; Debt over GDP (percent) Average 64.5 Max 96.6 Min 21.0 S.D. 26.1; Foreign debt over GDP (percent) Average 46.7 Max 78.4 Min 21.0 S.D. 20.9; GDP Growth (percent) Average -2.2 Max 6.5 Min -14.1 S.D. 5.3.
  - Average across datasets: Debt over GDP (percent) 77.4; Foreign debt over GDP (percent) 51.2; GDP Growth (percent) -1.7.
- Table 3. Output Growth and Debt Crises (System-GMM baseline and variants)
  - Debt crises t coefficient ranges reported across specifications:
    - (I) -5.566 (z = -2.05)**, N 2403, Hansen test-pvalue 0.323, Arellano-bond AR(2) test-pvalue 0.567.
    - (II) -5.384 (z = -2.04)**, N 2409, Hansen test-pvalue 0.460, AR(2) pvalue 0.995.
    - (III) -5.529 (z = -1.97)**, N 2403, Hansen test-pvalue 0.327, AR(2) pvalue 0.546.
    - (IV) -5.414 (z = -2.01)**, N 2404, Hansen test-pvalue 0.348, AR(2) pvalue 0.622.
    - (V) -6.065 (z = -2.36)**, N 3208, Hansen test-pvalue 0.166, AR(2) pvalue 0.151.
    - (VI) -5.321 (z = -1.98)**, N 2403, Hansen test-pvalue 0.312, AR(2) pvalue 0.590.
    - (VII) -6.412 (z = -2.66)***, N 3398, Hansen test-pvalue 1.00, AR(2) pvalue 0.969.
  - Real GDP growth t-1 consistently positive and significant (e.g., 0.387 (6.34)*** in (I)).
  - Openness t positive and often significant (e.g., 0.735 (2.31)** in (I)).
- Table 4. Checks for instruments validity
  - Debt Crises t: (I) System-GMM -5.566 (-2.05)**; (II) System-GMM-Collapsed Instruments -3.797 (-3.05)**; (III) Difference-GMM -4.000 (-3.67)***.
  - Number of instruments: 118 (I), 15 (II), 99 (III).
  - Hansen test-pvalue: 0.323 (I), 0.17 (II), 0.307 (III).
  - Arellano-Bond AR(2) test-pvalue: 0.567 (I), 0.944 (II), 0.546 (III).
- Table 5. Robustness Check for Exogenous Crises (variants a–d)
  - Debt crises t coefficients:
    - (I) -5.566 (-2.05)**, N 2403.
    - (II) -5.727 (-1.65)*, N 2403.
    - (III) -10.043 (-2.65)***, N 2369.
    - (IV) -7.546 (-3.88)***, N 828.
    - (V) -9.260 (-3.52)***, N 1263.
  - Hansen test-pvalue range: 0.321–0.372.
  - Arellano-bond AR(2) test-pvalue range: 0.055–0.746.
  - Notes define samples: (a) episodes with contemporaneous non-negative output growth; (b) drop observations with contemporaneous negative growth and occurrence of debt crisis; (c) use output-gap non-negative selection; (d) episodes with contemporaneous slowing growth dropped.
- Table 6. Different Datasets (Debt crises t)
  - (I) LV: -5.566 (-2.05)**.
  - (II) DHS: -5.096 (-1.72)*.
  - (III) LP: -9.984 (-2.27)**.
  - (IV) DS: -7.143 (-1.94)**.
  - (V) RRS: -9.319 (-2.63)***.
  - N = 2403 in each column; Hansen test-pvalues reported (e.g., 0.323 (I), 0.290 (II), 0.325 (III), 0.923 (IV), 0.309 (V)).
- Table 7. Different Controls and Estimation Methods
  - Debt crises t:
    - (I) Baseline: -5.566 (-2.05)**, N 2403.
    - (II) Additional controls: -7.025 (-3.51)***, N 2179.
    - (III) OLS: -4.316 (-4.91)***, N 2526.
  - Real GDP growth t-1: 0.387 (6.34)*** in baseline; 0.213 (3.09)*** with additional controls; 0.219 (5.77)*** in OLS.
  - Additional controls include Terms-of-trade t, Investment-GDP ratio t, Saving-GDP ratio t, Private credit-GDP ratio t, Political conflict dummy t (variables included but some coefficients not reported).
- Table 8. Output Growth and Financial Crises: Debt vs. Banking and Currency
  - Debt t coefficients across specifications:
    - (I) Debt t -4.963 (-1.89)*.
    - (II) -5.435 (-1.63)*.
    - (III) -8.913 (-1.97)**.
    - (IV) -8.064 (-1.90)**.
    - (V) -8.740 (-1.93)**.
  - Banking t and Currency t effects reported, with Currency t showing strong negative effects in several columns (e.g., -6.02 (-3.89)*** in (I)).
  - Interaction terms reported (e.g., Currency t * banking t -8.616 (-2.44)** in one specification).
  - N 4863 in all columns; Hansen test-pvalues range 0.865–1.00; AR(2) test-pvalues ~0.308–0.325.
  - Note: Non-contemporaneous episodes of debt, banking and currency crises.
- Table 9. Output Growth, Debt-GDP and Foreign Debt-GDP Ratios
  - Debt-to-GDP ratio results:
    - (I) linear: Debt_GDP t -0.011 (-1.52), N 1505.
    - (II) >70%: -1.776 (-2.62)***.
    - (III) >80%: -2.546 (-3.14)***.
    - (IV) >90%: -2.239 (-2.79)***.
  - Foreign Debt-to-GDP ratio results:
    - (V) linear: Foreign debt GDP t -0.011 (-1.72)*, N 1403.
    - (VI) >60%: -1.176 (-1.79)*.
    - (VII) >70%: -1.538 (-1.67)*.
    - (VIII) >80%: -2.418 (-2.09)**.
  - Control variables: Real GDP growth t-1 positive and significant across panels (e.g., 0.307 (3.73)*** in (I)); Openness t positive and often significant (e.g., 0.922 (2.16)** in (I)).
  - Hansen test-pvalues range 0.537–0.706; Arellano-bond AR(2) test-pvalues range 0.126–0.307.
  - Note: GMM-System Estimator two-step using Windmeijer standard errors; debt and foreign debt crises considered endogenous (instrumented using up to 3 lags).

*Italic: Source: _wp11280 - REFERENCES (content unit supplied).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11280.pdf_