## Appendix Tables — _wp10117

## Source details

**Canonical URL:** [Appendix Tables — _wp10117](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10117.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10117.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10117.pdf.json)

---

### Introduction and model
- Focus: determinants of the investment grade/speculative grade threshold for sovereign credit ratings in emerging market economies.
- Sample and estimation:
  - Sample of 48 emerging market economies.
  - Period: 1993−2008.
  - Dependent variable: binary indicator equal to 1 for countries assigned investment grade status by at least two out of the three agencies (Moody’s, S&P, Fitch).
  - Estimation technique: random effects binomial logit model.
- Key model specification (notation preserved):
  - IG_it = Λ(α + βX_it + λZ_i + a_i + μ_it).
- Dependent variable construction:
  - IG_it = 1 if investment grade by at least two agencies at end-December rating.
  - Investment/speculative grade status coincided across the three agencies for 94 percent of all observations.
- Endogeneity control: lagged values of time-varying regressors used.

### Core empirical findings
- Preferred specification: random effects restricted model (column D in Table 6).
  - Likelihood-ratio test rejected no variation in country-specific errors for pooled regression.
  - Hausman tests did not reject equality of random and fixed effects coefficients, supporting random effects.
- Core determinants identified (five variables):
  - External public debt (share of GDP).
  - Domestic public debt (share of GDP).
  - Political risk (ICRG index).
  - Exports (share of GDP).
  - Broad money (share of GDP) — financial depth measure.
- Exact magnitudes and behavioral findings:
  - Rating agencies distinguish between types of debt: high public debt indicators matter; private external debt not significant in the preferred specification.
  - Agencies attach greater risk to external public debt than to domestic public debt, with the coefficients of the former being more than 2½ times larger than the latter.
  - Political risk index significant and positively related to investment grade rating.
  - Exports to GDP significant and positive.
  - Broad money to GDP significant and positive, confirming relevance of financial depth.

### Model fit, classification performance, and dynamics
- Classification thresholds and performance:
  - Overall in-sample probability of being investment grade used as cut-off: 40 percent.
  - Model correctly classifies 86 percent of all observations.
  - Type I error (failing to predict investment grade status): 8.9 percent.
  - Type II error (failing to predict speculative grade status): 5.2 percent.
- Upgrades and downgrades prediction:
  - Downgrades: model correctly predicts 6 out of 9 downgrades (68 percent).
    - Failed to predict downgrades of Korea, Malaysia and Thailand (Asian crisis context; reversed within two years).
  - Upgrades: model correctly predicts 11 of the 17 upgrades (65 percent).
    - Failed to predict upgrades of Mexico and South Africa (predicted with a three year lag) and Colombia and Uruguay (lost investment grade within 3−5 years of upgrade).
- Behavior around switches (exact figures preserved):
  - Median decline in external public debt to GDP at upgrades: 16 percentage points of GDP.
  - In some cases (Russia and Bulgaria) decline was more than 30 percentage points.
  - Downgrades typically followed a large jump in debt, with median increase in total public debt of 23 percentage points of GDP (mostly external public debt).

### Nonlinear marginal effects and distributional ceilings
- Nonlinearities and example marginal effects:
  - Marginal effects larger for values in middle of distribution.
  - Example: a 10 percentage point decrease in external public debt to GDP would increase the probability of investment grade by 16 percentage points on average; effect almost double for countries in the second quartile.
  - Example: a 10 point increase in the political risk index would increase the probability of investment grade by 10 percentage points on average.
- Distributional ceilings observed:
  - No investment grade observations for external public debt beyond 64 percent of GDP.
  - No investment grade observations for domestic public debt beyond 85 percent of GDP.
  - Investment grade observations occur throughout the distributions of exports, broad money and political risk — no clear minimum threshold for those determinants.

### Public debt implications for BB/Ba (one notch below investment grade) countries
- 2009 cross-section diagnostic (BB/Ba countries):
  - External public debt in all these countries is above the investment grade median; domestic public debt is not uniformly above the investment grade median.
  - Exports to GDP, broad money to GDP, and ICRG index for these BB/Ba countries generally below investment grade medians.
- Policy-relevant marginal effect assessment:
  - Reducing public debt, particularly external public debt, substantially increases probability of upgrade.
  - Figure-based assessment (described in source): external public debt reduction needed by each BB/Ba country to raise probability of attaining investment grade status to 40 percent (holding other variables at 2009 levels).
  - Required debt reduction depends on current total public debt and other indicator values:
    - Example contrasts preserved from source: a country with total public debt close to 40 percent of GDP may need only a few percentage points reduction; a country with total public debt closer to 30 percent of GDP may need to reduce debt by 10 percentage points.

### Policy conclusions (as stated in the source)
- Main conclusions:
  - Investment grade status by the three major agencies can be explained by a small set of variables: external public debt, domestic public debt, political risk, exports, and financial depth (broad money).
  - The preferred panel random effects model correctly predicts 86 percent of investment grade status observations and about two thirds of upgrades and downgrades to/from investment grade.
  - Policy recommendation: emerging market efforts to increase the likelihood of an upgrade should focus on faster public debt reduction, particularly external public debt, because rating agencies assign larger weight to external debt over domestic debt.
  - Other areas (trade, financial depth, political risk) are important but likely require gradual improvement and are not as directly responsive to short-term macroeconomic policy maneuvers.

### Appendix I — explanatory variables (listing of regressors and interpretation)
- Macroeconomic variables:
  - Per capita income (U.S. dollars).
  - Real GDP growth and potential GDP growth.
  - Inflation rate.
  - Unemployment rate.
- External sector variables:
  - Exports to GDP.
  - External current account to GDP.
  - Private and public external debt to GDP (distinction between private and public external debt).
  - Net international reserves to GDP.
- Government sector variables:
  - Primary balance to GDP (used to avoid endogeneity with credit rating).
  - Public debt to GDP (distinguished between domestic and external public debt).
- Financial depth:
  - Broad money to GDP (proxy for financial intermediation and government financial flexibility).
- Other regressors:
  - Political risk (ICRG political risk index).
  - Default history (binary: 1 if defaulted on external debt at least once in previous ten years).
  - Regional dummies (Latin America and the Caribbean, Europe, East Asia).
  - Time dummies (annual dummies).

### Appendix II — robustness checks and agency-specific findings
- Alternative debt measures:
  - Using foreign and local currency denominated debt yields similar results; foreign currency debt has a much larger coefficient than local currency debt (Table A1, column A).
  - Maturity structure proxies (short term public debt measures) not significant.
- Debt scaled by exports:
  - Debt-to-exports ratio negative and significant but deteriorates model fit (Column B).
- Alternative political risk measure:
  - World Bank Governance Indicators’ political stability index positive and significant but deteriorates model fit (column C).
- Alternative financial depth measures:
  - Net credit to private sector and market capitalization as percent of GDP not significant; broad money to GDP best captures agencies’ concerns.
- Agency comparisons:
  - Dependent variable redefined by each agency yields similar results; agencies agree on investment/speculative classification in 94 percent of cases.
  - All agencies attribute more weight to external public debt over domestic public debt.
  - Agency-specific differences noted:
    - Moody’s places more weight on political risk; political risk not significant for S&P in some specifications.
    - Exports to GDP significant for Moody’s and significant with a lower coefficient for S&P.
    - Broad money to GDP significant for S&P and significant with a lower coefficient for Fitch.
  - Additional agency-specific variables:
    - Moody’s: private external debt had a positive significant effect in some specifications.
    - S&P: inflation negative and significant in some specifications.
    - Fitch: net international reserves positive and significant in some specifications.

*Source: _wp10117 - Appendix I. Explanatory Variables*

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

### References

### Tables
- 1. Sovereign Credit Ratings by Agency .............................................................................4
- 2. Main Studies on the Determinants of Sovereign Ratings ..............................................5
- 3. Explanatory Variables and Expected Sign .....................................................................6
- 4. Descriptive Statistics of Countries in the Sample and Investment Grade Status ...........7
- 5. Test of Equality of Means and Medians of Country Characteristics, by Investment 
  Grade Rating .................................................................................................................8
- 6. Regression Results .........................................................................................................9
- 7. Average Partial Effects on the Probability of Investment Grade Status ......................12

### Figures
- 1. Emerging Markets: Value of Explanatory Variables at Time of Upgrade to 
  Investment Grade and Five Years Earlier ..................................................................11
- 2. Histogram of Explanatory Variables by Investment Grade/Speculative Grade  
  Status ..........................................................................................................................13
- 3. Explanatory Variables for Emerging Markets with BB/Ba Sovereign Credit       
  Rating, 2009 ...............................................................................................................14
- 4. Proposed Adjustment for Emerging Markets with BB/Ba Sovereign Credit  
  Rating .........................................................................................................................15

### Appendices
- I. Explanatory Variables ..................................................................................................16
- II. Robustness Analysis ....................................................................................................18

*Source: _wp10117 - References*

### Appendix Tables

### Appendix Tables

### Introduction
- Focus: determinants of the investment grade/speculative grade threshold for sovereign credit ratings in emerging market economies.
- Sample and model:
  - Sample of 48 emerging market economies.
  - Period: 1993−2008.
  - Dependent variable: binary indicator equal to 1 for countries assigned investment grade status by at least two out of the three agencies (Moody’s, S&P, Fitch).
  - Estimation technique: random effects binomial logit model.
- Main finding: a small number of variables largely explain investment grade status; policy relevance centers on public debt management, especially external public debt.

### Background and literature review
- Investment grade definition:
  - S&P and Fitch: BBB or above.
  - Moody’s: Baa or above.
- Importance of sovereign ratings:
  - Determine borrowing costs in international capital markets.
  - Set ceilings for domestic banks and companies’ ratings.
  - Affect portfolio choices of institutional investors with risk bounds.
- Prior literature highlights (selected):
  - Cantor and Packer (1996): per capita income, GDP growth, inflation, external debt, level of economic development, default history.
  - Afonso (2003): GDP per capita key for developed countries; external debt key for developing countries.
  - Mixed results for political variables across studies.

### Empirical model specification
- Dependent variable:
  - IG_it = 1 if investment grade by at least two agencies at end-December rating.
  - Investment/speculative grade status coincided across the three agencies for 94 percent of all observations.
- Regression framework:
  - Panel random effects binomial logit preferred to pooled and fixed effects.
  - Model: IG_it = Λ(α + βX_it + λZ_i + a_i + μ_it) (notation preserved from source).
- Explanatory variables:
  - Macro, external, government sector, financial depth and other indicators (see Table 3 in source).
  - Political risk measured by ICRG index (higher value = lower risk).
- Endogeneity control: lagged values of time-varying regressors used.

### Data and estimation results
- Sample composition: 48 emerging market countries (list in source).
- Descriptive and statistical tests:
  - Welch tests for equality of means and Wilcoxon rank-sum tests for distributional differences between investment and speculative grade groups.
  - Investment grade countries tend to outperform speculative grade countries on most regressors.
- Preferred specification: random effects restricted model (column D in Table 6).
  - Likelihood-ratio test rejected no variation in country-specific errors for pooled regression.
  - Hausman tests did not reject equality of random and fixed effects coefficients, supporting random effects.
- Core determinants identified (five variables):
  - External public debt (share of GDP).
  - Domestic public debt (share of GDP).
  - Political risk (ICRG index).
  - Exports (share of GDP).
  - Broad money (share of GDP) — financial depth measure.
- Key empirical findings (exact language and magnitudes preserved):
  - Rating agencies distinguish between types of debt: high public debt indicators matter, private external debt not significant.
  - Agencies attach greater risk to external public debt than to domestic public debt, with the coefficients of the former being more than 2½ times larger than the latter.
  - Political risk index significant and positively related to investment grade rating.
  - Exports to GDP significant and positive.
  - Broad money to GDP significant and positive, confirming relevance of financial depth.
- Model fit and classification performance:
  - Overall in-sample probability of being investment grade used as cut-off: 40 percent.
  - Model correctly classifies 86 percent of all observations.
  - Type I error (failing to predict investment grade status): 8.9 percent.
  - Type II error (failing to predict speculative grade status): 5.2 percent.
  - Downgrades: model correctly predicts 6 out of 9 downgrades (68 percent).
    - Failed to predict downgrades of Korea, Malaysia and Thailand (Asian crisis context; reversed within two years).
  - Upgrades: model correctly predicts 11 of the 17 upgrades (65 percent).
    - Failed to predict upgrades of Mexico and South Africa (predicted with a three year lag) and Colombia and Uruguay (lost investment grade within 3−5 years of upgrade).
- Behavior around switches:
  - Median decline in external public debt to GDP at upgrades shown in Figure 1: 16 percentage points of GDP.
  - In some cases (Russia and Bulgaria) decline was more than 30 percentage points.
  - Downgrades typically followed a large jump in debt, with median increase in total public debt of 23 percentage points of GDP (mostly external public debt).
- Nonlinear marginal effects (Table 7 and discussion):
  - Marginal effects larger for values in middle of distribution.
  - Example: a 10 percentage point decrease in external public debt to GDP would increase the probability of investment grade by 16 percentage points on average; effect almost double for countries in the second quartile.
  - Example: a 10 point increase in the political risk index would increase the probability of investment grade by 10 percentage points on average.
  - Distributional ceilings observed:
    - No investment grade observations for external public debt beyond 64 percent of GDP.
    - No investment grade observations for domestic public debt beyond 85 percent of GDP.
  - Investment grade observations occur throughout the distributions of exports, broad money and political risk — no clear minimum threshold for those determinants.

### Public debt implications for emerging markets
- Focus for BB/Ba (one rating category below investment grade) countries:
  - 2009 values for the five determinants show external public debt in all these countries is above the investment grade median; domestic public debt is not uniformly above the investment grade median.
  - Exports to GDP, broad money to GDP, and ICRG index for these BB/Ba countries generally below investment grade medians.
  - Policy implication: progress on trade, financial depth and political risk is gradual and not directly linked to macroeconomic policy; fiscal consolidation can more directly reduce debt levels.
- Marginal effects assessment for BB/Ba countries:
  - Reducing public debt, particularly external public debt, substantially increases probability of upgrade.
  - Figure 4 (described): external public debt reduction needed by each country to raise probability of attaining investment grade status to 40 percent (holding other variables at 2009 levels).
  - Required debt reduction depends on current total public debt and other indicator values:
    - Example contrasts from source text:
      - A country with total public debt close to 40 percent of GDP may need only a few percentage points reduction.
      - A country with total public debt closer to 30 percent of GDP may need to reduce debt by 10 percentage points.

### Summary and conclusions
- Main conclusions:
  - Investment grade status by the three major agencies can be explained by a small set of variables: external public debt, domestic public debt, political risk, exports, and financial depth (broad money).
  - The preferred panel random effects model correctly predicts 86 percent of investment grade status observations and about two thirds of upgrades and downgrades to/from investment grade.
  - Policy recommendation: emerging market efforts to increase the likelihood of an upgrade should focus on faster public debt reduction, particularly external public debt, because rating agencies assign larger weight to external debt over domestic debt.
  - Other areas (trade, financial depth, political risk) are important but likely require gradual improvement and are not as directly responsive to short-term macroeconomic policy maneuvers.

*Appendix Tables — content unit from the source PDF*

### Appendix I. Explanatory Variables

### Appendix I. Explanatory Variables

### Macroeconomic variables
- Per capita income in U.S. dollars.
  - Higher per capita income tends to suggest a larger potential tax base and a greater ability to repay debt.
  - Serves as a proxy for the level of economic development, which might influence default risk.
- Real GDP growth and potential GDP growth.
  - Higher economic growth tends to decrease the relative debt burden and may help in avoiding insolvency.
- Inflation rate.
  - A low inflation rate reveals sustainable monetary and exchange rate policies and serves as a proxy of the quality of economic management.
- Unemployment rate.
  - A country with low unemployment tends to have more flexible labor markets making it less vulnerable to changes in the global environment.

### External sector variables
- Exports to GDP.
  - A higher ratio suggests a greater capacity to obtain hard currency to repay foreign currency denominated debt.
  - Unlike most previous studies, exports are introduced here as an independent regressor.
- External current account to GDP.
  - A large current account deficit suggests a high dependence on foreign capital, which can be a source of risk to macroeconomic stability.
- Private and public external debt to GDP.
  - Higher external indebtedness implies higher risk of fiscal or balance of payments stress.
  - This paper distinguishes between private and public externally issued debt to allow for differences in the weight assigned by rating agencies to each one.
- Net international reserves to GDP.
  - The higher the ratio, the more resources are available to service foreign debt and the lower the vulnerability to liquidity shocks.

### Government sector variables
- Primary balance to GDP.
  - A low primary balance indicates the government lacks the ability or the will to increase taxes to cover current expenses.
  - A weak fiscal position implies a higher likelihood that external shocks result in a default.
  - Regressions use the primary balance instead of the overall balance to avoid possible endogeneity with the credit rating.
- Public debt to GDP.
  - The higher the debt burden, the larger the transfer effort required to service obligations and therefore a higher risk of default.
  - This paper distinguishes between domestic and external public debt to allow for differences in weights.

### Financial depth
- Broad money to GDP.
  - Countries with deeper and more diversified finance are better positioned when private savings are low or financial systems are repressed.
  - Financial depth is a useful indicator of government financial flexibility; high levels of financial intermediation (proxied by broad money to GDP) can be associated with a greater capacity to sustain a given domestic debt burden.
  - Existing literature has not included this variable.

### Other regressors
- Political risk.
  - Measured by the International Country Risk Guide (ICRG) political risk index as a proxy for willingness to repay; rule of law and respect for property rights provide confidence that institutions will honor obligations.
- Default history.
  - Binary variable equal to 1 when the sovereign has defaulted on its external debt at least once in the previous ten years.
- Regional dummies.
  - Dummies included for the Latin America and the Caribbean, Europe, and East Asia.
- Time dummies.
  - Annual dummies included to capture possible bunching in rating movements due to global conditions.

### Appendix II. Robustness Analysis — alternative measures and results
- Using foreign and local currency denominated debt in place of external and domestic debt variables:
  - Renders similar results to the preferred specification (Table A1, column A).
  - Both foreign and local currency denominated debt have a significant and negative impact on the probability of investment grade status, with foreign currency debt having a much larger coefficient than local currency debt.
  - Measures capturing the maturity structure of public debt (proxied by short term public debt at original maturity as a percent of GDP and as a percent of total public debt) did not prove to be significant.
- Using debt scaled by exports as a regressor (instead of having debt and exports to GDP enter separately):
  - Coefficient for the debt to exports ratio is negative and significant (consistent with Mulder and Perrelli (2001) and Rowland and Torres (2004)), without altering the sign or significance of other coefficients.
  - This specification deteriorates the fit of the model (Column B).
- Alternative political risk measure:
  - The World Bank Governance Indicators’ political stability index used in place of the ICRG indicator produced a positive and significant coefficient, did not affect sign or significance of other coefficients, but deteriorated model fit (column C).
- Alternative measures of financial depth:
  - Net credit to the private sector and market capitalization of listed companies as a percent of GDP did not prove significant, implying broad money to GDP best captures rating agencies’ concerns regarding financial depth.

### Comparisons across credit agencies and additional findings
- Dependent variable redefined by each agency’s investment grade yields similar results; S&P, Fitch and Moody’s broadly share the same criteria for determining investment grade.
  - Agencies agree on classification across investment/speculative grades in 94 percent of the cases.
- All agencies attribute more weight to external public debt over domestic public debt.
- Agency-specific differences:
  - Moody’s appears to put more weight on the political risk variable than the average; political risk was not significant for S&P.
  - Exports to GDP was significant for Moody’s and significant with a lower coefficient for S&P.
  - Broad money to GDP was significant for S&P and significant with a lower coefficient for Fitch.
- Additional variables used by individual agencies:
  - Moody’s: private external debt had a positive significant effect on investment grade ratings, reflecting that steady private access to international markets indicates market confidence in the corporate sector (column B).
  - S&P: inflation was found to have a negative significant effect, serving as a proxy for the overall quality of macroeconomic management (column D).
  - Fitch: net international reserves were found to have a significant and positive effect, reflecting the importance attributed to strong liquidity buffers (column F).

*Source: Appendix I and Appendix II, _wp10117 - Appendix I. Explanatory Variables*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10117.pdf_