## _wp16246

## Source details

**Canonical URL:** [_wp16246](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16246.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16246.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16246.pdf.json)

---

### I. Introduction and research question
- Research question:
  - Whether gross government debt or debt net of government financial assets is the relevant measure for explaining sovereign spreads (or interest rates) on sovereign bonds for emerging markets.
- Theoretical considerations highlighted:
  - If governments own publicly known, liquid, and non–policy-tied assets, markets can value and potentially liquidate those assets, making net debt more relevant.
  - If markets lack comprehensive public data on assets, assets may be ignored and gross debt would be the relevant variable; private financial expertise could nonetheless make assets relevant despite limited public data.

### II. Theoretical background and mechanisms
- Standard arguments linking government debt to sovereign spreads:
  - New government liabilities raise sovereign default risk, increasing spreads required by investors.
  - Rapid expansion of government debt creates excess supply relative to diversified portfolio benchmarks, requiring higher returns (spreads) to attract demand.
- Literature positioning:
  - Much prior empirical work focuses on gross debt due to historical limitations in net debt data.
  - A smaller literature examines net government debt primarily for advanced economies.
  - Empirical gap identified: no known study (as of source) empirically investigates gross vs net debt relevance for sovereign spreads in emerging markets (exception: ongoing work by Henao-Arbelaez and Sobrinho (2016)).

### III. Data and empirical methodology
- Data sources and sample:
  - WEO database provides general government net debt series; WEO series reported within Fiscal Monitor since spring 2010; covers 28 advanced economies and 67 emerging and developing countries.
  - Alternative datasets: OECD main economic indicators; GFS (provides asset breakdown but only 10 emerging markets in authors’ sample and only four with more than 10 observations).
  - Annual series for 31 emerging economies over 1998–2014 (unbalanced).
  - Sovereign bond spreads from JP Morgan’s EMBIG; in some specs 5-year CDS from Bloomberg.
  - VIX from CBOE; 10-year US interest rate from Bloomberg.
  - GDP growth, inflation, primary fiscal balance from WEO; official international reserves from IFS.
- Key definitions (GFSM / WEO):
  - Gross debt: all liabilities that require payment of interest and/or principal (SDRs, currency and deposits, debt securities, loans, insurance, pensions and standardized guarantee schemes, and other accounts payable).
  - Financial assets corresponding to debt instruments: monetary gold and SDRs, currency and deposits, debt securities, loans, insurance, pension, and standardized guarantee schemes, and other accounts receivable.
  - Net debt: difference between gross debt and financial assets.
  - Note: official reserves held by the Central Bank are excluded from government financial assets in most WEO reporting (18 out of 20 IMF desks confirmed reserves excluded; 2 cases included for comparability).
- Empirical specification:
  - Dependent variable y_i,t is EMBIG spread.
  - X_i,t stands for gross debt, net debt, government financial assets, or combinations thereof.
  - Z_i,t are macro controls including VIX and US interest rates.
  - Fixed effects panel regressions preferred via Hausman test; various robustness estimators applied.

### IV. Preview and baseline empirical findings
- Preview without controls (Table 3):
  - Gross debt, net debt, and financial assets individually statistically significant with expected signs.
  - When gross debt and assets included jointly, coefficients similar in magnitude and opposite in sign; F-test implies sum of coefficients not statistically different from zero.
  - Net debt dominates: when net debt included jointly with gross debt or assets, those latter variables lose statistical significance while net debt remains virtually unchanged.
- Baseline with controls (Table 4) — findings and economic magnitudes:
  - Gross debt and financial assets offset each other; net debt dominates.
  - Exact reported magnitudes:
    - An increase in the net debt-to-GDP ratio by 10 percentage points is associated with an increase of 100–120 basis points in the country’s EMBIG spread.
    - A reduction in the level of net debt by half a standard deviation in the sample (about 16.5 percentage points) is associated with lower spreads in the order of about 150–200 basis points.
    - An increase in inflation by one percentage point leads to an increase in sovereign spreads by about 10 basis points.
    - An increase in GDP growth by one percentage point is associated with a reduction in sovereign spreads by about 10 basis points.
    - Sovereign spreads rise by about 13 basis points for each unit increase in VIX.
  - US interest rate typically enters with expected positive sign but is often not statistically significant (counterbalancing effects because dependent variable is spread over US Treasuries).

### V. Robustness checks and alternative specifications
- Country-group robustness (Table 5a):
  - Results hold for full EMEs sample, LAC, EMEs outside Europe and LAC, and EMEs excluding MENA.
  - European EMEs qualitatively similar though coefficients for assets (and to a lesser extent gross debt) decline in absolute value.
- Methodology robustness (Table 5b):
  - Lagging RHS variables, IV with two lags (Newey-West HAC SEs), time fixed effects, dynamic specifications with lagged dependent variable, GMM (Arellano-Bond), and mean-group estimator — baseline results remain qualitatively similar.
  - F-tests generally suggest gross debt + assets sum not different from zero (exception: emerging Europe in some cases).
- CDS robustness (Table 6):
  - Replacing EMBIG with CDS does not alter main conclusions (smaller sample but similar results); F-test p-value for gross debt + assets = 0 is 0.91 in CDS panel.
- Including international reserves and primary balance (Table 7):
  - International reserves enter with negative and statistically significant coefficients (coefficient about half the magnitude of financial assets).
  - Primary fiscal balance not statistically significant when included alongside stock variables.
- Distress episodes (Table 8):
  - Two distress types: global volatility (VIX > 20) and domestic distress (EMBI > mean + 2σ), implemented via dummies and interactions.
  - High global volatility: coefficients for gross debt and assets similar to baseline; interaction terms suggest possible amplification but are not statistically significant.
  - Domestic distress (EMBI > mean + 2σ): interaction terms statistically significant and imply domestic distress amplifies the effect of debt and assets on spreads, with assets’ role particularly amplified.
- Nonlinearities and composition of net debt (Table 9):
  - Gross debt-to-financial assets ratio used as leverage proxy.
  - Positive coefficient on gross debt/assets ratio suggests higher leverage may be penalized for the same net debt; effect small and not robust to inclusion of controls.
  - Example magnitude from sample: decline in gross debt-to-assets ratio from country average ~16 to 5 (via across-the-board increase in both gross debt and assets of 5 percent of GDP) associated with ~5 basis points lower spreads in first-column result, but loses significance with controls.

### VI. Additional explorations: advanced economies and local-currency rates
- Advanced economies (Table 10 and Table 11):
  - No visible effect of debt and assets on advanced-economy CDS spreads.
  - For local currency long-term interest rates in advanced economies:
    - Gross debt and net debt individually relevant; jointly gross debt seems to dominate net debt.
    - Estimated coefficients much smaller than for emerging market spreads (about one fifth the effect found for EMEs).
    - Assets typically insignificant or have unexpected sign, possibly reflecting lower liquidity/usability of advanced-economy government assets.
    - VIX shows opposite sign compared to EMEs (higher VIX associated with lower advanced-economy local rates), consistent with flight-to-quality.
- Emerging markets local currency interest rates (Table 12 and Table 13):
  - For EMEs’ 10-year local currency interest rates:
    - Net debt and assets both matter and each dominates gross debt when entered jointly.
    - Financial assets have coefficient comparable to net debt in baseline spread specifications for EMEs; gross debt often not significant.
  - Restricted-sample check: EMBIG regressions restricted to local-rate sample continue to show net debt dominance; gross debt and assets offset each other.
  - Indicates potential market segmentation by currency denomination (FX vs local currency) as an avenue for further investigation.

### VII. Illustrative evidence and examples
- Cross-sectional depiction for emerging markets in 2014 (Figure 1):
  - Positive relation between gross sovereign debt and sovereign spreads.
  - Negative relation between assets and sovereign spreads.
  - Both relations “seem to hold with the right sign,” suggesting net debt matters.
- Country case (Figure 2): Brazil and Colombia (2003–2012)
  - Gross debt declined in both countries, somewhat more in Colombia.
  - Brazil accumulated significantly more financial assets, implying sharper drop in net debt.
  - Studies based on gross debt would predict larger spread drop for Colombia; net-debt focus predicts larger drop for Brazil.
  - Empirical evidence: spreads dropped for both countries, but the drop was larger for Brazil.

### VIII. Conclusions, policy implications, and research avenues
- Key empirical conclusions:
  - Both gross debt and net debt have significant impacts on long-term sovereign bond spreads in EMEs.
  - Net debt’s effect dominates gross debt.
  - Gross debt and financial assets offset each other, implying net debt is the relevant measure for spreads.
- Policy-relevant implications:
  - Countries cannot lower spreads by reducing gross debt through asset sales, nor lower spreads by building assets financed with new gross debt — markets focus on net debt.
  - Net debt may be a superior measure for assessing debt sustainability (as measured by spreads), subject to data quality caveats.
- Magnitude summary reiterated:
  - A 10 percentage point increase in net debt-to-GDP typically associated with an increase in sovereign spreads by about 100–120 basis points.
- Robustness summary:
  - Findings stable across specifications, risk measures (EMBIG, CDS), country samples, and estimation methods.
  - Global volatility (high VIX) does not significantly alter gross-debt/assets effects; domestic distress amplifies debt/assets effects on spreads.
  - No clear evidence of large or robust nonlinearities from composition of net debt.
- Suggested further research directions:
  - Improve depth and quality of data on government assets; broaden coverage to more emerging and frontier markets.
  - Investigate composition, liquidity, maturity, and signaling role of government financial assets; disentangle central bank reserves versus government assets considering central bank independence.
  - Explore why gross debt effect on local currency rates in advanced economies is about five times smaller than effect on EM spreads.
  - Study market segmentation between local currency–denominated and foreign currency–denominated debt.
  - Undertake event studies around large asset accumulation/divergence between gross and net debt; assess asymmetry of asset accumulation vs reduction; analyze sensitivity to restructuring, re-profiling, default, or IMF programs.
  - Compare returns on government asset portfolios and yields on sovereign liabilities to weigh insurance benefits versus opportunity costs of investing government assets.

### IX. Annex highlights (figures and panels)
- Annex I: Evolution of Gross Debt and Net Debt — Regional Groupings
  - Time series covered: 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014.
  - Panels present series as % of GDP and in US$ billions for regional groupings: Advanced Economies; Emerging and Developing Economies; Latin America and the Caribbean; Emerging Europe; Middle East, North Africa, Afghanistan, and Pakistan; Sub-Saharan Africa; Commonwealth of Independent States; Emerging and Developing Asia.
  - Series labels used: net debt; gross debt (when net debt available); gross debt (all countries).
  - Axis label ranges visible in panels (examples preserved as presented): Advanced Economies % of GDP axis: 0 20 40 60 80 100 120; US$ billions axis: 0 10000 20000 30000 40000 50000 60000. (Other regional axis labels preserved in annex figures.)
- Annex II: Evolution of Gross Debt and Net Debt, by Country
  - Time series covered in country panels: 2000 2005 2010 2015.
  - Country panels plot gross debt (% of GDP) and net debt (% of GDP) for listed countries (Belize; Bolivia; Brazil; Bulgaria; Chile; Colombia; Costa Rica; Egypt; Ghana; Hungary; Jordan; Kazakhstan; Latvia; Lebanon; Lithuania; Mexico; Morocco; Namibia; Nigeria; Pakistan; Panama; Peru; Poland; Serbia, Republic of; South Africa; Trinidad and Tobago; Turkey; Ukraine; Uruguay; Vietnam).

*Source: _wp16246 - 1. Summary Statistics; Section IV; Annex I and Annex II.*

### 1. Summary Statistics...................................................................................................

### _wp16246 - 1. Summary Statistics...................................................................................................

### I. Introduction and research question
- The paper investigates whether gross government debt or debt net of government financial assets is the relevant measure for explaining sovereign spreads (or interest rates) on sovereign bonds for emerging markets.
- The empirical question arises because recent data availability on government financial assets allows testing whether net debt matters, while much of the prior empirical literature focused on gross debt.
- Two opposing theoretical considerations are highlighted:
  - If governments own publicly known, liquid, and non–policy-tied assets, markets can value and potentially liquidate those assets, making net debt more relevant.
  - If markets lacked comprehensive public data on assets, assets may have been ignored and gross debt would have been the relevant variable; financial experts’ private knowledge could, however, make assets relevant despite limited public data.

### II. Theoretical background and mechanisms
- Standard economic arguments motivating a positive relationship between government debt and sovereign spreads:
  - An increase in government debt through issuance of new liabilities raises sovereign default risk, which should increase spreads to compensate investors.
  - Rapid expansion of government debt creates an excess supply of such assets relative to diversified portfolio benchmarks investors aim to hold, requiring higher returns (spreads) to attract demand.
- The literature has been agnostic on whether gross or net debt is more relevant for spreads, primarily due to historical limitations in net debt data availability.
- Reference: Gruber and Kamin (2012) is cited as an example of theoretical arguments about the relationship between government debt levels and spreads/interest rates on government debt.

### III. Illustrative evidence from figures and examples
- Cross-sectional depiction for emerging markets in 2014 (Figure 1) shows:
  - A positive relation between gross sovereign debt and sovereign spreads.
  - A negative relation between assets and sovereign spreads.
  - Both relations "seem to hold with the right sign," which tends to suggest net debt matters.
- Country case example (Figure 2): Brazil and Colombia (2003–2012)
  - Gross debt declined in both countries, somewhat more in Colombia than in Brazil.
  - Brazil accumulated significantly more financial assets over the same period, implying a sharper drop in net debt.
  - Studies based exclusively on gross debt would predict a larger drop in spreads for Colombia; focusing on net debt predicts a larger drop for Brazil.
  - Empirical evidence supports the net-debt–based conclusion: spreads dropped for both countries, but the drop was larger for Brazil.

### IV. Policy relevance
- The question of gross versus net debt matters for public asset–liability management decisions, notably for countries with significant government asset positions.
- Implications include:
  - The perceived cost of building emergency buffers via government borrowing depends on whether gross or net debt drives spreads.
  - Findings are especially relevant for countries that have been making, or plan to make, asset–liability decisions where the cost of borrowing is a key factor.

### V. Data and organization (as presented)
- The paper uses recently available data on government financial assets for emerging markets to test relative importance of gross debt and assets for sovereign spreads.
- Figures referenced:
  - Figure 1: Government Debt and Sovereign Spreads (2014): Gross or Net Debt?
  - Figure 2: Example: Debt, Assets, and Spreads in Brazil and Colombia (2003–2012)
- Paper organization (overview):
  - Section II: short literature review.
  - Section III: methodology, dataset, and empirical specification.
  - Subsequent sections: empirical results (baseline and alternative specifications) and robustness checks.

*Source: _wp16246 - 1. Summary Statistics...................................................................................................*

### Section IV, which is followed by Section V that gives a glance at results from some

### _wp16246 - Section IV, which is followed by Section V that gives a glance at results from some

### Literature review: positioning and gap
- Seminal and subsequent work document determinants of sovereign bond spreads in emerging markets (Edwards 1984; Ciarlone, Piselli, and Trebeschi 2009; Bellas, Papaioannou, and Petrova 2010; Baldacci, Gupta, and Mati 2011; Csonto and Ivaschenko 2013).
- Many studies analyze gross government debt (e.g., Dell’Erba, Hausmann, and Panizza 2013; Akitoby and Stratmann 2008; Baldacci and Kumar 2010; Dell’Erba and Sola 2013; Ardagna, Caselli, and Lane 2007; Jaramillo and Weber 2013) but typically ignore government assets.
- A smaller literature examines net government debt primarily for advanced economies (Gruber and Kamin 2012; Chinn and Frankel 2005; Haugh, Ollivaud, and Turner 2009). Theoretical work on gross debt and assets includes Bianchi, Hatchondo, and Martinez (2012).
- Empirical gap: no known study (as of source) empirically investigates whether gross or net debt is more relevant for sovereign spreads in emerging markets, with the exception of ongoing work by Henao-Arbelaez and Sobrinho (2016).

### Methodology: data sources, sample, and empirical specification
- Data on net debt: WEO database provides the widest country coverage on general government net debt series; WEO series reported within Fiscal Monitor since spring 2010 and currently encompasses 28 advanced economies and 67 emerging and developing countries.
- Alternative datasets: OECD main economic indicators (more restrictive sample), GFS (provides asset breakdown but would restrict sample severely: only 10 emerging markets with data in the authors’ sample and only four with more than 10 observations).
- Key GFSM definitions (as used by WEO):
  - Gross debt: all liabilities that require payment of interest and/or principal (SDRs, currency and deposits, debt securities, loans, insurance, pensions and standardized guarantee schemes, and other accounts payable).
  - Financial assets corresponding to debt instruments: monetary gold and SDRs, currency and deposits, debt securities, loans, insurance, pension, and standardized guarantee schemes, and other accounts receivable.
  - Net debt: difference between gross debt and financial assets.
- Note: official reserves held by the Central Bank are excluded from the definition of government financial assets in most WEO reporting (survey: 18 out of 20 IMF desks confirmed reserves excluded; 2 cases included for comparability).
- Overall dataset used in empirical analysis:
  - Annual series for 31 emerging economies over the period 1998–2014 (unbalanced).
  - General government gross and net debt from WEO; sovereign bond spreads from JP Morgan’s EMBIG; in some specifications 5-year CDS from Bloomberg.
  - VIX from CBOE; 10-year US interest rate from Bloomberg.
  - GDP growth, inflation, primary fiscal balance from WEO; official international reserves from IFS.
- Empirical specification (panel regressions, fixed effects preferred via Hausman test):
  - Dependent variable y_i,t is EMBIG spread.
  - X_i,t stands for gross debt, net debt, government financial assets, or combinations.
  - Z_i,t are macro controls including VIX and US interest rates.
  - Fixed effects panel regressions reported unless otherwise stated.

### Results — preview and baseline findings
- Preview without controls (Table 3):
  - Gross debt, net debt, and financial assets are individually statistically significant with expected signs.
  - When gross debt and assets are included jointly, coefficients are similar in magnitude and opposite in sign; F-test implies sum of coefficients not statistically different from zero.
  - Net debt dominates: when net debt included jointly with gross debt or assets, the latter lose statistical significance while net debt remains virtually unchanged.
- Baseline with controls (Table 4):
  - Findings remain largely unchanged after adding controls (VIX, US interest rate, GDP growth, inflation, etc.).
  - Gross debt and financial assets offset each other; net debt dominates.
  - Economic magnitudes:
    - An increase in the net debt-to-GDP ratio by 10 percentage points is associated with an increase of 100–120 basis points in the country’s EMBIG spread.
    - A reduction in the level of net debt by half a standard deviation in the sample (about 16.5 percentage points) is associated with lower spreads in the order of about 150–200 basis points.
    - An increase in inflation by one percentage point leads to an increase in sovereign spreads by about 10 basis points.
    - An increase in GDP growth by one percentage point is associated with a reduction in sovereign spreads by about 10 basis points.
    - Sovereign spreads rise by about 13 basis points for each unit increase in VIX.
  - US interest rate typically enters with expected positive sign but is often not statistically significant (counterbalancing effects noted because dependent variable is spread over US Treasuries).

### Robustness checks and alternative specifications
- Robustness across country groups (Table 5a): results hold for full EMEs sample, LAC, EMEs outside Europe and LAC, and EMEs excluding MENA. European EMEs qualitatively similar though coefficients for assets (and to a lesser extent gross debt) decline in absolute value.
- Robustness across methodologies (Table 5b): lagging RHS variables, IV with two lags (Newey-West HAC SEs), time fixed effects, dynamic specifications with lagged dependent variable, GMM (Arellano-Bond), and mean-group estimator — baseline results remain qualitatively similar. F-tests generally suggest gross debt + assets sum not different from zero (with exception: emerging Europe in some cases).
- CDS robustness (Table 6): replacing EMBIG with CDS does not alter main conclusions (sample smaller but results similar); F-test p-value for gross debt + assets = 0 is 0.91 in CDS panel.
- Including international reserves and primary balance (Table 7):
  - International reserves enter with negative and statistically significant coefficients (coefficient about half the magnitude of financial assets).
  - Primary fiscal balance is not statistically significant when included alongside stock variables, suggesting fiscal developments may be captured by gross debt and assets.
- Distress episodes (Table 8):
  - Two types: global volatility episodes (VIX > 20) and domestic distress episodes (EMBI > mean + 2σ), implemented via dummies and interactions with gross debt and assets.
  - During high global volatility, coefficients for gross debt and assets similar to baseline; interaction terms suggest possible amplification but are not statistically significant.
  - During domestic distress (EMBI > mean + 2σ), interaction terms are statistically significant and imply domestic distress amplifies the effect of debt and assets on spreads (assets’ role particularly amplified).
- Nonlinearities and composition of net debt (Table 9):
  - Introduce gross debt-to-financial assets ratio as proxy for “leverage.”
  - Positive coefficient on gross debt/assets ratio suggests higher leverage (more gross debt per unit of assets) may be penalized for the same net debt; effect small and not robust to inclusion of controls.
  - Example magnitude from sample: decline in gross debt-to-assets ratio from country average ~16 to 5 (via across-the-board increase in both gross debt and assets of 5 percent of GDP) associated with ~5 basis points lower spreads (first-column result), but loses significance with controls.

### Additional explorations: advanced economies and local currency interest rates
- Advanced economies (Table 10 and Table 11):
  - Credit default spreads for advanced economies: no visible effect of debt and assets on CDS spreads.
  - Local currency long-term interest rates in advanced economies:
    - Gross debt and net debt individually relevant; when jointly included gross debt seems to dominate net debt.
    - Estimated coefficients much smaller than for emerging market spreads (about one fifth the effect found for EMEs).
    - Assets typically insignificant or have unexpected sign, possibly reflecting lower liquidity/usability of advanced-economy government assets.
    - VIX shows opposite sign compared to EMEs (higher VIX associated with lower advanced-economy local rates), consistent with flight-to-quality.
- Emerging markets local currency interest rates (Table 12 and Table 13):
  - For emerging markets’ 10-year local currency interest rates:
    - Net debt and assets both matter and each dominates gross debt when entered jointly.
    - Financial assets have a coefficient comparable to net debt in baseline spread specifications for EMEs; gross debt often not significant.
  - Restricted-sample check (Table 13): when EMBIG regressions are restricted to the local-rate sample, baseline EMBIG results continue to hold — net debt dominates and gross debt and assets offset each other.
  - Suggests potential market segmentation by currency denomination of debt (FX vs local currency) as an avenue for further investigation.

### Conclusions and avenues for future research
- Three key empirical conclusions:
  - Both gross debt and net debt have significant impacts on long-term sovereign bond spreads in EMEs.
  - Net debt’s effect dominates gross debt.
  - Gross debt and financial assets offset each other, implying net debt is the relevant measure for spreads.
- Policy-relevant implications:
  - Countries cannot lower spreads by reducing gross debt through asset sales, nor lower spreads by building assets financed with new gross debt — markets focus on net debt.
  - Net debt may be a superior measure for assessing debt sustainability (as measured by spreads), subject to data quality caveats.
- Magnitude summary reiterated:
  - A 10 percentage point increase in net debt-to-GDP typically associated with an increase in sovereign spreads by about 100–120 basis points.
- Robustness: findings stable across specifications, risk measures (EMBIG, CDS), country samples, and estimation methods.
- Distress and nonlinearities:
  - Global volatility (high VIX) does not significantly alter gross-debt/assets effects; domestic distress amplifies debt/assets effects on spreads.
  - No clear evidence of large or robust nonlinearities from composition of net debt.
- Suggested further research directions:
  - Improve depth and quality of data on government assets; broaden coverage to more emerging and frontier markets.
  - Investigate composition, liquidity, maturity, and signaling role of government financial assets; disentangle central bank reserves versus government assets, considering central bank independence.
  - Explore why gross debt effect on local currency rates in advanced economies is about five times smaller than effect on EM spreads.
  - Study market segmentation between local currency-denominated and foreign currency-denominated debt.
  - Undertake event studies around large asset accumulation/divergence between gross and net debt; assess asymmetry of asset accumulation vs reduction; analyze sensitivity to restructuring, re-profiling, default, or IMF programs.
  - Compare returns on government asset portfolios and yields on sovereign liabilities to weigh insurance benefits versus opportunity costs of investing government assets.

*Source: _wp16246 - Section IV (extracted content provided).*

### ANNEX I.

### ANNEX I.

### Evolution of Gross Debt and Net Debt — Regional Groupings
- Time series covered: 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014.
- For each regional grouping, two panel types are shown:
  - Series expressed "as % of GDP".
  - Series expressed "in US$ billions".
- Series labels used consistently across panels:
  - net debt
  - gross debt (when net debt available)
  - gross debt (all countries)

- Advanced Economies
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 20 40 60 80 100 120 (as % of GDP); 0 10000 20000 30000 40000 50000 60000 (in US$ billions).
  - Source: WEO database.

- Emerging and Developing Economies
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 10 20 30 40 50 60 (as % of GDP); 0 2000 4000 6000 8000 10000 12000 14000 (in US$ billions).

- Latin America and the Caribbean
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 10 20 30 40 50 60 (as % of GDP); 0 500 1000 1500 2000 2500 3000 3500 (in US$ billions).

- Emerging Europe
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: -10 0 10 20 30 40 50 60 (as % of GDP); -100 0 100 200 300 400 500 600 700 800 900 1000 (in US$ billions).

- Middle East, North Africa, Afghanistan, and Pakistan
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: -40 -20 0 20 40 60 80 (as % of GDP); -1500 -1000 -500 0 500 1000 1500 (in US$ billions).

- Sub-Saharan Africa
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 10 20 30 40 50 60 70 (as % of GDP); 0 100 200 300 400 500 600 (in US$ billions).

- Commonwealth of Independent States
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 10 20 30 40 50 60 70 80 90 (as % of GDP); 0 100 200 300 400 500 600 (in US$ billions).

- Emerging and Developing Asia
  - Panels: (as % of GDP) and (in US$ billions).
  - Axis labels visible: 0 10 20 30 40 50 60 (as % of GDP); 0 1000 2000 3000 4000 5000 6000 7000 (in US$ billions).

### Annex II — Evolution of Gross Debt and Net Debt, by Country
- Time series covered in country panels: 2000 2005 2010 2015.
- Each country panel plots:
  - gross debt (% of GDP)
  - net debt (% of GDP)

- Countries shown (grouped as presented):
  - Belize, Bolivia, Brazil
  - Bulgaria, Chile, Colombia
  - Costa Rica, Egypt, Ghana
  - Hungary, Jordan, Kazakhstan
  - Latvia, Lebanon, Lithuania
  - Mexico, Morocco, Namibia
  - Nigeria, Pakistan, Panama
  - Peru, Poland, Serbia, Republic of
  - South Africa, Trinidad and Tobago, Turkey
  - Ukraine, Uruguay, Vietnam

*Source: WEO database (figures showing net debt, gross debt (when net debt available), and gross debt (all countries) for regional groupings; country panels showing gross debt and net debt as % of GDP).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16246.pdf_