## _wp11100

## Source details

**Canonical URL:** [_wp11100](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11100.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11100.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11100.pdf.json)

---

### Introduction
- Objective: build a new index of fiscal stress that provides early warning signals of fiscal sustainability problems for advanced and emerging economies.
- Innovations:
  - Fiscal crisis definition broadened beyond sovereign debt default/near-default to include: public debt default/restructuring, need to access large-scale official/IMF support, hyperinflation, and extreme financing problems (spikes in sovereign bond spreads).
  - Fiscal stress index based on indicators measuring current fiscal variables and their baseline projections using the conceptual framework in Baldacci, McHugh and Petrova (2011).
  - Thresholds for each indicator estimated via a univariate procedure that maximizes the likelihood of predicting a fiscal crisis.
  - Index measures the number of indicators exceeding thresholds, weighted by relative signaling power.
- Key empirical findings overview:
  - Fiscal stress risks remain elevated in advanced economies and well above pre-crisis years.
  - Fiscal stress is lower for emerging economies due to rebuilding of fiscal buffers and more positive growth prospects, but remains higher than pre-crisis years and signals continued vulnerabilities to shocks.

### Crisis definition and event identification (Methodology)
- Fiscal crisis episode definition: period of extreme government funding difficulties (Cottarelli, 2011).
- Four types of criteria to capture fiscal distress:
  - (i) debt default or restructuring;
  - (ii) implicit default (high inflation);
  - (iii) recourse to exceptional official financing;
  - (iv) sharp deterioration in market access (sovereign yield pressure).
- Operational criteria:
  - Public debt default or restructuring: S&P definition (failure to service debt as payments come due; restructuring/rescheduling alters original debt terms).
  - Large financing: large IMF-supported program defined as access to 100 percent of quota or more.
  - Implicit/internal public debt default via high inflation:
    - Advanced economies: inflation greater than 35 percent per annum.
    - Emerging economies: inflation greater than 500 percent per annum.
  - Extreme sovereign financing constraint via sovereign yield pressure:
    - Sovereign spreads greater than 1,000 basis points or 2 s.d. from the country average (applies to both advanced and emerging economies).
    - For emerging economies, include periods when bond yield spreads exceed 1,000 basis points even if not exceeding 2 s.d.

### Data, sample, and event counting
- Annual data for 29 advanced economies and 52 emerging economies covering 1970-2010 used to identify fiscal stress events.
- Data sources: Standard and Poor (debt default/restructuring), IMF Finance Department database (exceptional IMF programs), IMF IFS / Bloomberg / Datastream (sovereign yields and CDS where available).
- Event counting rules:
  - Distinct fiscal-pressure events require at least two years of no fiscal distress between events.
  - Only the start year of the event is considered the fiscal distress year.
- Resulting crisis counts:
  - 41 fiscal distress events in advanced economies.
  - 135 fiscal distress events in emerging economies.
- Sensitivity notes:
  - Starting in 2009, many high-access programs exceed the 100 percent of quota threshold; changing this threshold does not alter the number of fiscal distress events significantly.
  - Excluding precautionary arrangements does not change results significantly.
  - Lowering thresholds for identifying crisis episodes worsens predictive ability, particularly for emerging economies.
  - Alternative inflation thresholds (e.g., 100 percent for emerging economies) do not significantly affect results.
  - The 1,000 basis point sovereign spread threshold is not binding for advanced economies.

### Summary of events (counts, types, duration)
- Advanced Economies (total):
  - Fiscal Stress Events: 41
  - Default or Restructuring: 0
  - IMF-Supported Program: 6
  - High Inflation: 5
  - Bond Yield Pressure: 29
  - Duration of Fiscal Stress (in years): 2.5
- Advanced Economies by decade:
  - 1970-79: Fiscal Stress Events 14; Default or Restructuring 0; IMF-Supported Program 1; High Inflation 4; Bond Yield Pressure 7; Duration 2.6
  - 1980-89: Fiscal Stress Events 8; Default or Restructuring 0; IMF-Supported Program 2; High Inflation 1; Bond Yield Pressure 5; Duration 2.5
  - 1990-99: Fiscal Stress Events 8; Default or Restructuring 0; IMF-Supported Program 2; High Inflation 0; Bond Yield Pressure 6; Duration 2.3
  - 2000-10: Fiscal Stress Events 11; Default or Restructuring 0; IMF-Supported Program 1; High Inflation 0; Bond Yield Pressure 11; Duration 2.6
- Emerging Economies (total):
  - Fiscal Stress Events: 135
  - Default or Restructuring: 52
  - IMF-Supported Program: 79
  - High Inflation: 6
  - Bond Yield Pressure: 15
  - Duration of Fiscal Stress (in years): 3.6
- Emerging Economies by decade:
  - 1970-79: Fiscal Stress Events 15; Default or Restructuring 8; IMF-Supported Program 9; High Inflation 1; Bond Yield Pressure 0; Duration 3.1
  - 1980-89: Fiscal Stress Events 41; Default or Restructuring 26; IMF-Supported Program 22; High Inflation 0; Bond Yield Pressure 0; Duration 6.6
  - 1990-99: Fiscal Stress Events 37; Default or Restructuring 12; IMF-Supported Program 20; High Inflation 5; Bond Yield Pressure 5; Duration 2.6
  - 2000-10: Fiscal Stress Events 42; Default or Restructuring 6; IMF-Supported Program 28; High Inflation 0; Bond Yield Pressure 10; Duration 1.6
- Additional observations:
  - About 60 percent of the cases relate to IMF-supported programs (79 events) and a third to outright defaults and restructuring (each 52 events).
  - In the last decade, fiscal stress events increasingly identified through severe bond yield spikes.
  - The appendix’s approach identifies more crisis events than other studies.

### Signaling methodology, threshold selection, and index construction
- Signaling approach adopted for transparency and accommodation of data availability differences.
- Indicator signaling dummy:
  - d_t = 1 if xt relative to cut-off C meets the risk condition (indicator crosses cut-off in period t-1 signals crisis in period t), otherwise 0.
- Two methods to determine optimal C:
  - Total misclassified errors (TME): TME(C) = FP(C)/NN + FN(C)/NC. Optimal C* minimizes TME(C). TME yields relatively conservative thresholds.
  - Signal-to-noise ratio (SNR): SNR(C) = (1 - type II errors)/type I errors. Optimal C* maximizes SNR(C). SNR can force very high type II errors.
- Aggregation (two-step):
  - Step 1: For each cluster of fiscal indicators, if an indicator crosses its threshold assign value 1 and weigh proportionately to its predictive power.
  - Step 2: Aggregate cluster indices into overall fiscal stress index using indicators' predictive power and cluster predictive power.
- Estimation practice:
  - TME performs better than SNR; adjustments to TME made to avoid corner solutions, account for trends/structural breaks, and limited reliable data since mid-1990s.
  - Thresholds estimated separately for advanced and emerging economies with constraint they lie on the risk-prone side relative to 1995-2010 median; outliers removed to improve robustness.

### Estimated thresholds, signaling power, and index weights (selected entries)
- Advanced Economies (cluster signaling power totals and selected indicator weights)
  - Basic Fiscal Variables (signaling power cluster total 31.59):
    - r-g (5 year average): Direction to be safe <2; Crisis obs. 167; Non-crisis obs. 03.60; Threshold 0.48; Type 1 error 0.29; Type 2 error 0.14; Index Weight 14.9
    - General government gross debt (percent of GDP): Direction to be safe <155; Crisis obs. 6172.20; Non-crisis obs. 0.22; Threshold 0.67; Type 1 error 0.67; Type 2 error 7.3
    - Cyclically adjusted primary balance (percent of potential GDP): Direction to be safe >5; Crisis obs. 137-4.20; Non-crisis obs. 0.45; Threshold 0.40; Type 1 error 9.4
  - Asset and Liability Management (cluster total 42.56):
    - Gross financing needs (percent of GDP): Direction to be safe <62; Crisis obs. 1917.20; Non-crisis obs. 0.45; Threshold 0.17; Type 1 error 24.6
    - Share of short term debt as a ratio of total debt: Direction to be safe <1550; Crisis obs. 69.1; Non-crisis obs. 0.96; Threshold 0.00; Type 1 error 2.8
    - Debt held by nonresidents (percent of total debt): Direction to be safe <71; Crisis obs. 6383.60; Non-crisis obs. 0.06; Threshold 0.79; Type 1 error 10.1
    - Weighted average maturity of general government debt (years): Direction to be safe >18; Crisis obs. 5353.90; Non-crisis obs. 0.09; Threshold 0.83; Type 1 error 5.0
  - Long-Term Fiscal Trend (cluster total 25.85):
    - Fertility rate (deviation from 2.1): Direction to be safe <3; Crisis obs. 19500.64; Non-crisis obs. 0.25; Threshold 0.71; Type 1 error 12.4
    - Long term projections of public health expenditure (percent of GDP): Direction to be safe <28; Crisis obs. 7494.50; Non-crisis obs. 0.53; Threshold 0.32; Type 1 error 9.4
    - Long term projections of public pension expenditure (percent of GDP): Direction to be safe <21; Crisis obs. 6006.20; Non-crisis obs. 0.09; Threshold 0.76; Type 1 error 9.6
    - Dependency ratio: Direction to be safe <37; Crisis obs. 100636.00; Non-crisis obs. 0.06; Threshold 0.86; Type 1 error 4.5
- Emerging Economies (cluster signaling power totals and selected indicator weights)
  - Basic Fiscal Indicators (cluster total 23.72):
    - r-g (5 year average): Direction to be safe <5; Crisis obs. 24711.10; Non-crisis obs. 0.58; Threshold 0.25; Type 1 error 11.3
    - General government gross debt (percent of GDP): Direction to be safe <20; Crisis obs. 24542.80; Non-crisis obs. 0.61; Threshold 0.35; Type 1 error 2.5
    - Cyclically adjusted primary balance (percent of potential GDP): Direction to be safe >6; Crisis obs. 0592-0.50; Non-crisis obs. 0.33; Threshold 0.52; Type 1 error 9.9
  - Asset and Liability Management (cluster total 43.56):
    - Gross financing needs (percent of GDP): Direction to be safe <29; Crisis obs. 28720.60; Non-crisis obs. 0.27; Threshold 0.69; Type 1 error 2.8
    - Share of short term debt as a ratio of total debt: Direction to be safe <45; Crisis obs. 43044.00; Non-crisis obs. 0.15; Threshold 0.71; Type 1 error 9.2
    - Debt denominated in foreign currencies (percent of total debt): Direction to be safe <52; Crisis obs. 55540.30; Non-crisis obs. 0.70; Threshold 0.15; Type 1 error 9.9
    - Weighted average maturity of general government debt (years): Direction to be safe >40; Crisis obs. 3702.30; Non-crisis obs. 0.11; Threshold 0.85; Type 1 error 2.6
    - Short term external debt (percent of gross international reserves): Direction to be safe <10; Crisis obs. 1110461.80; Non-crisis obs. 0.43; Threshold 0.28; Type 1 error 19.1
  - Long-Term Fiscal Outlook (cluster total 32.73):
    - Fertility rate (deviation from 2.1): Direction to be safe <60; Crisis obs. 6591.30; Non-crisis obs. 0.87; Threshold 0.05; Type 1 error 5.2
    - Long term projections of public health expenditure (percent of GDP): Direction to be safe <28; Crisis obs. 2452.70; Non-crisis obs. 0.38; Threshold 0.50; Type 1 error 8.2
    - Long term projections of public pension expenditure (percent of GDP): Direction to be safe <54; Crisis obs. 74.00; Non-crisis obs. 0.60; Threshold 0.20; Type 1 error 13.4
    - Dependency ratio: Direction to be safe <81; Crisis obs. 83016.10; Non-crisis obs. 0.40; Threshold 0.51; Type 1 error 15.9

### Main predictors and regression evidence
- Top predictors by country group:
  - Advanced economies: government rollover pressures, size of gross financing needs, and fiscal solvency concerns.
  - Emerging economies: liquidity constraints and public debt structure (currency composition).
- Logit regression findings:
  - Fiscal stress index and components are all significant determinants of fiscal stress episodes.
  - Correlation higher with basic fiscal variables component in advanced economies; asset and liability management component is primary driver in emerging economies.
  - Multivariate logit regressions confirm: basic fiscal variables largest marginal effect in advanced economies; asset and liability management largest marginal effect in emerging market economies.

### Correlations with market indicators (2011) and index trends
- Correlations (2011):
  - Advanced economies:
    - Correlation with sovereign bond yields as of March 2011: 0.4.
    - Correlation with credit ratings: 0.24.
    - Correlation of basic fiscal variables component with bond yields: 0.5.
    - Correlation of basic fiscal variables component with credit ratings: 0.4.
  - Emerging economies:
    - Lower correlations between fiscal stress index and market risk prices/ratings.
    - Asset and liability management component has the strongest correlation among subcomponents.
- Aggregate trends (1995-2011):
  - Fiscal stress increased more rapidly in advanced than in emerging economies.
  - In 2011 the fiscal stress index—weighted with countries’ PPP-GDP—is higher in advanced countries.
  - In advanced economies the fiscal stress index has doubled since 2006 and is at record-high levels.
  - In emerging economies the fiscal stress index is elevated but slightly below late 1990s financial crisis peaks.
- Component contributions:
  - Advanced economies: increase since mid-2000s driven by deterioration in basic fiscal variables; asset and liability management component contributed about half of the increase; long-term fiscal indicators exerted continuous pressure.
  - Emerging economies: main factors are basic fiscal variables followed by long-term fiscal trends; asset and liability management component (mostly due to declining short-term debt to international reserves) mitigated further increase.
- Regional patterns (2011, unweighted):
  - Advanced economies: highest fiscal stress index in North America; peak levels in peripheral euro countries.
  - Emerging economies: fiscal stress index markedly higher in Emerging Europe, followed by Middle East and North Africa.
- Regional changes (1996-2011 and subperiods):
  - Last five years up to 2011: index increased sharply in North America (deterioration in cyclically adjusted primary balance, increases in debt and gross financing needs).
  - Asia and the Pacific: index increased the least but on upward trend due to demographic pressures and rising debt/gross financing needs.
  - Emerging markets: index increased most in Latin America over last five years; Emerging Europe remained elevated 1996-2011.

### Data coverage and summary statistics (1995-2010 sample)
- Sample: 1995-2010.
- Selected indicator means and standard deviations:
  - r-g (5-year average)
    - AE: Obs. 435; Mean 0.20; St. Dev 2.50
    - EM: Obs. 591; Mean -4.69; St. Dev 6.11
  - General government gross debt
    - AE: Obs. 443; Mean 60.38; St. Dev 34.13
    - EM: Obs. 629; Mean 48.81; St. Dev 31.10
  - Cyclical adjusted primary balances
    - AE: Obs. 446; Mean 0.34; St. Dev 3.42
    - EM: Obs. 654; Mean 0.43; St. Dev 9.07
  - Current gross financing need
    - AE: Obs. 450; Mean 11.28; St. Dev 10.74
    - EM: Obs. 595; Mean 10.85; St. Dev 10.54
  - Share of short-term debt as a ratio of total debt
    - AE: Obs. 450; Mean 24.35; St. Dev 16.11
    - EM: Obs. 492; Mean 22.05; St. Dev 20.83
  - Debt denominated in foreign currencies
    - Obs. 712; Mean 54.88; St. Dev 27.67
  - Debt held by non-residents as a proportion of total debt
    - Obs. 228; Mean 42.10; St. Dev 23.92
  - Weighted average maturity of general government debt
    - AE: Obs. 444; Mean 7.08; St. Dev 3.37
    - EM: Obs. 668; Mean 6.38; St. Dev 4.45
  - Short-term external debt (percent of international reserves)
    - Obs. 705; Mean 73.22; St. Dev 87.49
  - Total fertility rate
    - AE: Obs. 441; Mean 0.52; St. Dev 0.28
    - EM: Obs. 822; Mean 0.78; St. Dev 0.76
  - Old age dependency ratio
    - AE: Obs. 448; Mean 31.53; St. Dev 6.65
    - EM: Obs. 832; Mean 17.33; St. Dev 8.06
  - Long-term projections of the change in public pension expenditure
    - AE: Obs. 385; Mean 2.22; St. Dev 2.00
    - EM: Obs. 89; Mean 2.07; St. Dev 2.45
  - Long-term projections of the change in public health expenditure
    - AE: Obs. 396; Mean 3.69; St. Dev 1.30
    - EM: Obs. 336; Mean 1.74; St. Dev 1.14

### Conclusions and suggested extensions
- Role and limitations:
  - Fiscal stress index is a signaling tool to assess exposure to fiscal sustainability risks and to identify underlying factors; it does not attempt to predict crises, which are typically triggered by combinations of economic, financial, or political shocks.
  - Signaling tools should be complemented by judgment-based approaches.
- Key results summary:
  - Thresholds identify likelihood of fiscal stress across a broad set of fiscal variables using EWS methodology.
  - Index uses parsimonious set of fiscal indicators proposed by Baldacci, McHugh and Petrova (2011) and a broader crisis definition consistent with Cottarelli (2011).
  - Top predictors: gross financing needs and solvency risks in advanced countries; public debt structure and exposure to financial spillovers in emerging markets.
  - Fiscal stress risk increased dramatically worldwide as consequence of global financial crisis; risks higher in advanced economies than in emerging economies but remain elevated in both. North America and Europe show highest risks.
- Proposed further work:
  - Use bootstrapping methods to gauge uncertainty around point estimates.
  - Conduct analysis using thresholds based on country-specific distributions instead of an overall threshold to control for country-specific characteristics.
  - Incorporate time-specific effects to account for common factors during periods of global contagion.

*Source: _wp11100 - References*

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

### _wp11100 - References .............................................................................................................

### Introduction
- Objective: build a new index of fiscal stress that provides early warning signals of fiscal sustainability problems for advanced and emerging economies.
- Innovations:
  - Fiscal crisis definition broadened beyond sovereign debt default/near-default to include: public debt default/restructuring, need to access large-scale official/IMF support, hyperinflation, and extreme financing problems (spikes in sovereign bond spreads).
  - Fiscal stress index based on indicators measuring current fiscal variables and their baseline projections using the conceptual framework in Baldacci, McHugh and Petrova (2011).
  - For each indicator, thresholds are estimated via a univariate procedure that maximizes the likelihood of predicting a fiscal crisis.
  - The index measures the number of indicators exceeding thresholds, weighted by relative signaling power.
- Key empirical findings overview:
  - Fiscal stress risks remain elevated in advanced economies and well above pre-crisis years.
  - Fiscal stress is lower for emerging economies due to rebuilding of fiscal buffers and more positive growth prospects, but remains higher than pre-crisis years and signals continued vulnerabilities to shocks.

### Literature review
- EWS model variation dimensions: (i) crisis event definition; (ii) methodology; (iii) indicator set; country coverage often limited by data quality.
- Crisis definitions in prior fiscal-crisis studies typically cover tail events (outright default/rescheduling, large-scale official financing).
- Rationale for broader crisis definition: extreme rollover problems and severe financing difficulties can occur without outright default; broader definition captures these sustainability risks.
- Two common empirical approaches:
  - Univariate “signaling” approach (Kaminsky, Lizondo and Reinhart, 1998): identify critical thresholds for each indicator, average number of indicators exceeding thresholds into composite index, weight by signaling power.
  - Multivariate regression (panel probit/logit): binary dependent variable for crisis occurrence, coefficients test conditional significance.
- Comparative evidence:
  - Berg and Patillo (1999) and Berg, Borensztein and Patillo (2005): multivariate probit outperforms signaling in-sample and in cross-country predictions; signaling can have better out-of-sample performance.
  - No clear winner; performance depends on crisis type.
- This paper uses the “signaling” approach:
  - Rationale: transparency, accommodates differences in data availability, avoids limiting predictive variables due to data gaps.
  - Limitation: individual variables cannot be tested for conditional statistical significance; contribution reflected via weights proportional to predictive power.
- Fiscal indicators literature:
  - External sector variables matter for financial crises: reserve levels, real exchange rate, current account, external debt relative to exports.
  - Fiscal variables less powerful predictors historically; best fiscal predictors include short-term public debt and foreign-currency debt.
  - Indicator set here is parsimonious and follows Baldacci, McHugh and Petrova (2011), covering solvency risks (current deficit and debt levels, projected growth-adjusted interest rate), long-term demographic-related budget pressures (projected change in health care and pension expenditures), sovereign asset/liability composition, and financing requirements.

### Methodology
- Fiscal crisis episode definition: period of extreme government funding difficulties (Cottarelli, 2011), arising from public debt build-up, contingent liabilities, negative revenue shocks, demographic spending pressures, or adverse debt composition.
- Four types of criteria to capture fiscal distress: (i) debt default or restructuring; (ii) implicit default; (iii) recourse to exceptional official financing; (iv) sharp deterioration in market access.
- Specific operational criteria:
  - Public debt default or restructuring: S&P definition (failure to service debt as payments come due; restructuring/rescheduling alters original debt terms); includes commercial and official creditors.
  - Large financing: large IMF-supported program defined as access to 100 percent of quota or more.
  - Implicit/internal public debt default via high inflation:
    - Advanced economies: inflation greater than 35 percent per annum.
    - Emerging economies: inflation greater than 500 percent per annum.
  - Extreme sovereign financing constraint via sovereign yield pressure:
    - Sovereign spreads greater than 1,000 basis points or 2 s.d. from the country average (applies to both advanced and emerging economies).
    - For emerging economies, include periods when bond yield spreads exceed 1,000 basis points even if not exceeding 2 s.d.
- Data and sample:
  - Annual data for 29 advanced economies and 52 emerging economies covering 1970-2010 used to identify fiscal stress events.
  - Data sources: Standard and Poor (debt default/restructuring), IMF Finance Department database (exceptional IMF programs), IMF IFS / Bloomberg / Datastream (sovereign yields and CDS where available).
  - Where available, long-term domestic bond spreads and 5-year CDS spreads used for advanced economies; long-term domestic bond spreads relative to comparable U.S. bonds used for emerging economies.
- Event counting and separation:
  - Distinct fiscal-pressure events require at least two years of no fiscal distress between events.
  - Only the start year of the event is considered the fiscal distress year.
- Resulting crisis counts (based on the paper’s definition and sample):
  - 41 fiscal distress events in advanced economies.
  - 135 fiscal distress events in emerging economies.
- Sensitivity notes:
  - Starting in 2009, many high-access programs exceed the 100 percent of quota threshold; changing this threshold does not alter the number of fiscal distress events significantly.
  - Excluding precautionary arrangements does not change results significantly.
  - Lowering thresholds for identifying crisis episodes worsens predictive ability, particularly for emerging economies.
  - Alternative inflation thresholds (e.g., 100 percent for emerging economies) do not significantly affect results.
  - The 1,000 basis point sovereign spread threshold is not binding for advanced economies due to traditionally lower sovereign spreads.

*Source: _wp11100 - References*

### Appendix Tables 3a and 3b report the duration of the event and the fiscal stress criterion that was used to

### Appendix Tables 3a and 3b report the duration of the event and the fiscal stress criterion that was used to identify the episode.

### Summary of Events Across Advanced and Emerging Economies
- Table 2 reports counts by start of crisis period, type of event, and duration of fiscal stress (in years).
- Advanced Economies (total):
  - Fiscal Stress Events: 41
  - Default or Restructuring: 0
  - IMF-Supported Program: 6
  - High Inflation: 5
  - Bond Yield Pressure: 29
  - Duration of Fiscal Stress (in years): 2.5
- Advanced Economies by decade:
  - 1970-79: Fiscal Stress Events 14; Default or Restructuring 0; IMF-Supported Program 1; High Inflation 4; Bond Yield Pressure 7; Duration 2.6
  - 1980-89: Fiscal Stress Events 8; Default or Restructuring 0; IMF-Supported Program 2; High Inflation 1; Bond Yield Pressure 5; Duration 2.5
  - 1990-99: Fiscal Stress Events 8; Default or Restructuring 0; IMF-Supported Program 2; High Inflation 0; Bond Yield Pressure 6; Duration 2.3
  - 2000-10: Fiscal Stress Events 11; Default or Restructuring 0; IMF-Supported Program 1; High Inflation 0; Bond Yield Pressure 11; Duration 2.6
- Emerging Economies (total):
  - Fiscal Stress Events: 135
  - Default or Restructuring: 52
  - IMF-Supported Program: 79
  - High Inflation: 6
  - Bond Yield Pressure: 15
  - Duration of Fiscal Stress (in years): 3.6
- Emerging Economies by decade:
  - 1970-79: Fiscal Stress Events 15; Default or Restructuring 8; IMF-Supported Program 9; High Inflation 1; Bond Yield Pressure 0; Duration 3.1
  - 1980-89: Fiscal Stress Events 41; Default or Restructuring 26; IMF-Supported Program 22; High Inflation 0; Bond Yield Pressure 0; Duration 6.6
  - 1990-99: Fiscal Stress Events 37; Default or Restructuring 12; IMF-Supported Program 20; High Inflation 5; Bond Yield Pressure 5; Duration 2.6
  - 2000-10: Fiscal Stress Events 42; Default or Restructuring 6; IMF-Supported Program 28; High Inflation 0; Bond Yield Pressure 10; Duration 1.6
- Source noted in table: Authors’ calculations.

### Incidence and Timing of Fiscal Crises
- Incidence of new fiscal stress events is clustered around specific periods.
- Advanced economies:
  - Fiscal stress episodes prior to the recent financial crisis associated with the oil boom of 1973 and the recession of the early 1990s.
  - Many countries entered fiscal distress after the onset of the recent crisis in 2008, with additional new crises in 2009-10.
- Emerging economies:
  - Fiscal stress events clustered around the public debt crises in the early 1980s, the Latin American and Asian crises of the 1990s, and the recent global financial crisis.
- Figure 1 sources: IMF International Financial Statistics; Bloomberg; Standard and Poor’s; and authors’ calculations.

### Duration and Countries in Fiscal Stress
- Average length of fiscal stress:
  - Advanced economies: 2½ years
  - Emerging economies: 3½ years
- Because fiscal stress episodes last multiple years, the incidence of fiscal crises in a year may not equal the number of countries experiencing fiscal stress in that year.
- Analysis presents both the number of crisis events (incidence) and the number of countries in fiscal stress in parallel.
- Figure 2 sources: IMF International Financial Statistics; Bloomberg; Standard and Poor’s; and authors’ calculations.

### Key Observations
- About 60 percent of the cases relate to IMF-supported programs (79 events) and a third to outright defaults and restructuring (each 52 events).
- In the last decade, fiscal stress events in these economies were increasingly identified through severe bond yield spikes.
- The appendix’s approach identifies more crisis events than other studies (reference to Table 3).

*Source: Authors’ calculations; sources cited within figures: IMF International Financial Statistics; Bloomberg; Standard and Poor’s.*

### Appendix Tables 4a and 4b). This stems from a more comprehensive definition of crisis

### _wp11100 - Appendix Tables 4a and 4b). This stems from a more comprehensive definition of crisis

### Comparison of events across studies
- Table 3 summary (Fiscal Stress Index; IMF EWS; RR; HKS; MRS; LV):
  - Number of events: 176; 28; 48; 16; 32; 22
  - Number of common events: (IMF EWS vs others) — 20; 30; 13; 33; 33
  - Number of countries: 81; 48; 66; 29; 47; 102
- Notes:
  - Events dated differently by only one year are considered common events. Events within the identified crisis durations are also considered common.
  - IMF EWS refers to IMF (2007), RR to Reinhart and Rogoff (2010), HKS to Hemming, Kell and Schimmelpfenning (2003), MRS to Manasse, Roubini and Schimmelpfenning (2003), LV to Laeven and Valencia (2008).
  - The number of total events and missed events in other studies do not add to 164 due to differences in the sample of countries and dates covered.

### Fiscal stress thresholds (methodology)
- Signaling approach (IMF, 2007; IMF, 2010):
  - Define cut-off values C for each fiscal indicator xt such that indicator signals a crisis if xt is on the risk-prone side of C.
  - Signaling window j set to one year.
- Definition of indicator dummy d_t:
  - d_t = 1 if xt relative to C meets the risk condition (indicator crosses cut-off in period t-1 signals crisis in period t), otherwise 0.
- Two methods to determine optimal C:
  - Total misclassified errors (TME): TME(C) = FP(C)/NN + FN(C)/NC (sum of type I and type II errors normalized by totals). Optimal C* minimizes TME(C).
    - TME places greater weight on misclassifying fiscal crisis events given small number of crisis events relative to non-crisis periods, yielding relatively conservative thresholds.
  - Signal-to-noise ratio (SNR): SNR(C) = (1 - type II errors)/type I errors expressed with normalization. Optimal C* maximizes SNR(C).
    - SNR approach can force very high type II errors, yielding thresholds with very high percentages of missed crises and misclassified non-crises.

### Fiscal stress index construction
- Two-step aggregation:
  - Step 1: For each cluster of fiscal indicators, if an indicator crosses its threshold assign value 1 and weigh proportionately to its predictive power.
  - Step 2: Aggregate cluster indices into overall fiscal stress index using indicators' predictive power and cluster predictive power.
- Formal aggregation formula provided in source (weights for individual indicators and groups, and dummy d taking 1 when indicator is above/below threshold).

### Data and coverage
- Indicators and data sources:
  - Uses 12 fiscal indicators (Baldacci, McHugh and Petrova, 2011) classified into three clusters: basic fiscal variables, long-term fiscal trends, and asset and liability management (Appendix Table 5).
  - Data sources: IMF’s Fiscal Monitor, IMF’s World Economic Outlook (WEO), Bank of International Settlements (BIS), and United Nations databases.
- Sample period and focus:
  - Some data available 1970-2010, most series start in 1980s; complete country coverage only by mid-1990s.
  - Thresholds estimated using complete dataset, analysis focuses on period after 1995.
- Observed patterns:
  - Advanced economies: global financial crisis starting in 2008 triggered pronounced deterioration in basic fiscal variables (e.g., public debt to GDP and cyclically adjusted primary balance) and sharp upturn in gross financing needs. Long-term pension and health expenditure costs trending up.
  - Emerging economies: deterioration in cyclically adjusted primary balance began before crisis; public debt to GDP remained lower than historical levels. Asset and liability management variables deteriorated since 2008 due to large deficit financing needs; financing conditions worsened with short-term debt reaching mid-1990s crisis levels. Long-term fiscal challenge indicators trending up but less than in advanced economies.

### Indicator thresholds and weights (estimation practice)
- Estimation approach:
  - TME method performs better than SNR; however, adjustments to TME are made to avoid median- or tail-based corner solutions, account for trends/structural breaks, and limited reliable data since mid-1990s.
  - Thresholds estimated separately for advanced and emerging economies with constraint they lie on the risk-prone side relative to 1995-2010 median; outliers removed to improve robustness.
- Adjustments and rationale:
  - Outliers defined via standardized scores (high absolute standardized value).
  - Some indicators required outlier removal: basic fiscal variables, gross financing needs, fertility rate indicator.
  - In some cases (weighted average maturity and debt held by nonresidents for advanced economies) threshold located on crisis-prone side and selected without data adjustment.

### Estimated thresholds, signaling power, and index weights (selected reporting from Tables 5a and 5b)
- Advanced Economies: Thresholds and relative weights (signaling power shown as Index Weight)
  - Basic Fiscal Variables (signaling power cluster total 31.59):
    - r-g (5 year average): Direction to be safe <2; Crisis obs. 167; Non-crisis obs. 03.60; Threshold 0.48; Type 1 error 0.29; Type 2 error 0.14; Index Weight 14.9
    - General government gross debt (percent of GDP): Direction to be safe <155; Crisis obs. 6172.20; Non-crisis obs. 0.22; Threshold 0.67; Type 1 error 0.67; Type 2 error 7.3
    - Cyclically adjusted primary balance (percent of potential GDP): Direction to be safe >5; Crisis obs. 137-4.20; Non-crisis obs. 0.45; Threshold 0.40; Type 1 error 9.4
  - Asset and Liability Management (cluster total 42.56):
    - Gross financing needs (percent of GDP): Direction to be safe <62; Crisis obs. 1917.20; Non-crisis obs. 0.45; Threshold 0.17; Type 1 error 24.6
    - Share of short term debt as a ratio of total debt: Direction to be safe <1550; Crisis obs. 69.1; Non-crisis obs. 0.96; Threshold 0.00; Type 1 error 2.8
    - Debt held by nonresidents (percent of total debt): Direction to be safe <71; Crisis obs. 6383.60; Non-crisis obs. 0.06; Threshold 0.79; Type 1 error 10.1
    - Weighted average maturity of general government debt (years): Direction to be safe >18; Crisis obs. 5353.90; Non-crisis obs. 0.09; Threshold 0.83; Type 1 error 5.0
  - Long-Term Fiscal Trend (cluster total 25.85):
    - Fertility rate (deviation from 2.1): Direction to be safe <3; Crisis obs. 19500.64; Non-crisis obs. 0.25; Threshold 0.71; Type 1 error 12.4
    - Long term projections of public health expenditure (percent of GDP): Direction to be safe <28; Crisis obs. 7494.50; Non-crisis obs. 0.53; Threshold 0.32; Type 1 error 9.4
    - Long term projections of public pension expenditure (percent of GDP): Direction to be safe <21; Crisis obs. 6006.20; Non-crisis obs. 0.09; Threshold 0.76; Type 1 error 9.6
    - Dependency ratio: Direction to be safe <37; Crisis obs. 100636.00; Non-crisis obs. 0.06; Threshold 0.86; Type 1 error 4.5
  - (Source: Authors’ calculations.)

- Emerging Economies: Thresholds and relative weights (signaling power shown as Index Weight)
  - Basic Fiscal Indicators (cluster total 23.72):
    - r-g (5 year average): Direction to be safe <5; Crisis obs. 24711.10; Non-crisis obs. 0.58; Threshold 0.25; Type 1 error 11.3
    - General government gross debt (percent of GDP): Direction to be safe <20; Crisis obs. 24542.80; Non-crisis obs. 0.61; Threshold 0.35; Type 1 error 2.5
    - Cyclically adjusted primary balance (percent of potential GDP): Direction to be safe >6; Crisis obs. 0592-0.50; Non-crisis obs. 0.33; Threshold 0.52; Type 1 error 9.9
  - Asset and Liability Management (cluster total 43.56):
    - Gross financing needs (percent of GDP): Direction to be safe <29; Crisis obs. 28720.60; Non-crisis obs. 0.27; Threshold 0.69; Type 1 error 2.8
    - Share of short term debt as a ratio of total debt: Direction to be safe <45; Crisis obs. 43044.00; Non-crisis obs. 0.15; Threshold 0.71; Type 1 error 9.2
    - Debt denominated in foreign currencies (percent of total debt): Direction to be safe <52; Crisis obs. 55540.30; Non-crisis obs. 0.70; Threshold 0.15; Type 1 error 9.9
    - Weighted average maturity of general government debt (years): Direction to be safe >40; Crisis obs. 3702.30; Non-crisis obs. 0.11; Threshold 0.85; Type 1 error 2.6
    - Short term external debt (percent of gross international reserves): Direction to be safe <10; Crisis obs. 1110461.80; Non-crisis obs. 0.43; Threshold 0.28; Type 1 error 19.1
  - Long-Term Fiscal Outlook (cluster total 32.73):
    - Fertility rate (deviation from 2.1): Direction to be safe <60; Crisis obs. 6591.30; Non-crisis obs. 0.87; Threshold 0.05; Type 1 error 5.2
    - Long term projections of public health expenditure (percent of GDP): Direction to be safe <28; Crisis obs. 2452.70; Non-crisis obs. 0.38; Threshold 0.50; Type 1 error 8.2
    - Long term projections of public pension expenditure (percent of GDP): Direction to be safe <54; Crisis obs. 74.00; Non-crisis obs. 0.60; Threshold 0.20; Type 1 error 13.4
    - Dependency ratio: Direction to be safe <81; Crisis obs. 83016.10; Non-crisis obs. 0.40; Threshold 0.51; Type 1 error 15.9
  - (Source: Authors’ calculations.)

### Main predictors and regression results
- Top predictors differ by country group:
  - Advanced economies: government rollover pressures, size of gross financing needs, and fiscal solvency concerns are strongest predictors.
  - Emerging economies: liquidity constraints and public debt structure (currency composition) are primary signals of fiscal stress.
- Logit regression findings:
  - Fiscal stress index and components are all significant determinants of fiscal stress episodes.
  - Correlation higher with basic fiscal variables component in advanced economies; asset and liability management component is primary driver in emerging economies.
  - Multivariate logit regressions confirm: basic fiscal variables largest marginal effect in advanced economies; asset and liability management largest marginal effect in emerging market economies.

### Correlations with market indicators (2011)
- Advanced economies (fiscal stress index estimated for 2011):
  - Correlation with sovereign bond yields as of March 2011: 0.4 (significantly different from zero).
  - Correlation with credit ratings: 0.24 (lower but still significant).
  - Correlation of basic fiscal variables component with bond yields: 0.5.
  - Correlation of basic fiscal variables component with credit ratings: 0.4.
- Emerging economies:
  - Lower correlations between fiscal stress index and market risk prices/ratings.
  - Asset and liability management component has the strongest correlation among subcomponents.

### Fiscal stress index trends and regional patterns
- Aggregate trends (1995-2011):
  - Fiscal stress increased more rapidly in advanced than in emerging economies.
  - In 2011 the fiscal stress index—weighted with countries’ PPP-GDP—is higher in advanced countries.
  - In advanced economies the fiscal stress index has doubled since 2006 and is at record-high levels.
  - In emerging economies the fiscal stress index is elevated but slightly below late 1990s financial crisis peaks.
- Component contributions (advanced economies):
  - Increase since mid-2000s driven by sharp deterioration in basic fiscal variables (debt to GDP; cyclically adjusted primary balance).
  - Asset and liability management component peaked and contributed about half of the increase.
  - Long-term fiscal indicators exerted continuous pressure.
- Component contributions (emerging economies):
  - Main factors: basic fiscal variables followed by long-term fiscal trends.
  - Asset and liability management component (mostly due to declining short-term debt to international reserves) mitigated further increase.
- Regional levels (2011, unweighted):
  - Advanced economies: highest fiscal stress index in North America; peak levels in peripheral euro countries.
  - Emerging economies: fiscal stress index markedly higher in Emerging Europe, followed by Middle East and North Africa.
- Regional changes (1996-2011 and subperiods):
  - Last five years up to 2011: index increased sharply in North America (deterioration in cyclically adjusted primary balance, increases in debt and gross financing needs).
  - Asia and the Pacific: index increased the least but on upward trend over 15 years due to demographic pressures and rising debt/gross financing needs.
  - Emerging markets: index increased most in Latin America over last five years (peaking cyclically adjusted primary deficits; in some cases declining debt maturity and reserve coverage). Emerging Europe remained elevated 1996-2011 due to solvency indicators, asset and liability risks, and long-term fiscal outlook concerns.

### Conclusions and suggested extensions
- Role and limitations:
  - The fiscal stress index is a signaling tool to assess exposure to fiscal sustainability risks and to identify underlying factors; it does not attempt to predict crises, which are typically triggered by combinations of economic, financial, or political shocks.
  - Signaling tools should be complemented by judgment-based approaches.
- Key results summary:
  - Thresholds identify likelihood of fiscal stress across a broad set of fiscal variables using EWS methodology.
  - Index uses parsimonious set of fiscal indicators proposed by Baldacci, McHugh and Petrova (2011) and a broader crisis definition consistent with Cottarelli (2011).
  - Top predictors: gross financing needs and solvency risks in advanced countries; public debt structure and exposure to financial spillovers in emerging markets.
  - Fiscal stress risk increased dramatically worldwide as consequence of global financial crisis; risks higher in advanced economies than in emerging economies but remain elevated in both. North America and Europe show highest risks.
- Proposed further work:
  - Use bootstrapping methods to gauge uncertainty around point estimates.
  - Conduct analysis using thresholds based on country-specific distributions instead of an overall threshold to control for country-specific characteristics.
  - Incorporate time-specific effects to account for common factors during periods of global contagion.

*Source: Authors’ calculations and content from _wp11100 - Appendix Tables 4a and 4b). This stems from a more comprehensive definition of crisis (IMF working paper content).*

### REFERENCES

### _wp11100 - REFERENCES

### References (Selected works cited)
- Abiad, Abdul, 2003, “Early Warning Systems: A Survey and a Regime-Switching Approach,” IMF Working Paper 03/32.
- Baldacci, Emanuele, James McHugh and Iva Petrova, 2011, “Measuring Fiscal Vulnerability and Fiscal Stress: A Proposed Set of Indicators,” IMF Working Paper 11/94.
- Beim, David, and Charles Calomiris, 2001, “Emerging Financial Markets.” Appendix to Chapter 1. New York: McGraw-Hill/Irwin Publishers.
- Berg, Andrew and Catherine Patillo, 1999, “Predicting Currency Crises: The Indicators Approach and an Alternative,” Journal of International Money and Finance, Vol.18, pp. 561-586.
- Berg, Andrew, Eduardo Borensztein and Catherine Patillo, 2005, “Assessing Early Warning Systems: How Have They Worked in Practice?” IMF Staff Papers, Vol. 52, No. 3, pp. 462-502.
- Bussiere, Matthieu, and Marcel Fratzscher, 2006, “Towards a New Early Warning System of Financial Crises” Journal of International Money and Finance, Vol. 25, pp. 953-973.
- Candelon, Bertrand, Elena-Ivona Dumitrescu and Christophe Hurlin, 2010, “How to Evaluate an Early Warning System,” IMF Working Paper.
- Chamon, Marcos, Paolo Manasse and Alessandro Prati, 2007, “Can We Predict the Next Capital Account Crisis?” IMF Staff Papers, Vol. 54, No. 2, pp. 270-305.
- Čihák, Martin and Klaus Schaeck, 2010, “How Well do Aggregate Prudential Ratios Identify Banking System Problems?” Journal of Financial Stability, Volume 6, Issue 3, September 2010, pp. 130-144.
- Cottarelli, Carlo, 2011, “The Risk Octagon: A Comprehensive Framework for Assessing Sovereign Risks” presented in University of Rome “La Sapienza,” January.
- Detragiache, Enrica, and Antonio Spilimbergo, 2001, ”Crises and Liquidity: Evidence and Interpretation,” IMF Working Paper No. 01/2.
- Edison, Hali J., 2000, “Do Indicators of Financial Crises Work? An Evaluation of an Early Warning System,” International Finance Discussion Paper Nº675.
- Eichengreen, B., R. Hausmann and U. Panizza, 2002, “Original Sin: the Pain, the Mystery, and the Road to Redemption,” Mimeo.
- Finger, Harald and Mauro Mecagni, 2007, An Analysis of Recent Cross-Country Experience, IMF Occasional Paper.
- Frankel, Jeffrey and George Saravelos, 2010, “Are Leading Indicators of Financial Crises Useful for Assessing Country Vulnerability? Evidence from the 2008-09 Global Crisis,” NBER Working Paper 16047, June.
- Hemming, Richard, Michael Kell, and Axel Schimmelpfennig, 2003, Fiscal Vulnerability and Financial Crises in Emerging Market Economies, IMF Occasional Paper 218.
- Hemming, Richard, and Murray Petrie, 2000, “A Framework for Assessing Fiscal Vulnerability,” IMF Working Paper 00/52.
- International Monetary Fund, 2007, “Assessing Underlying Vulnerabilities and Crisis Risks in Emerging Market Countries—A New Approach.”
- International Monetary Fund, 2010a, Updating the VEE Methodology, unpublished paper.
- International Monetary Fund, 2010b, “Fiscal Exit: From Strategy to Implementation,” Fiscal Monitor, November 2010.
- International Monetary Fund, 2011, “Shifting Gears: Tackling Challenges of the Road to Fiscal Adjustment,” Fiscal Monitor, April 2011.
- Kaminsky, Graciela, Saul Lizondo and Carmen Reinhart, 1998, “Leading Indicators of Currency Crisis,” IMF Staff Papers, Vol. 45, No. 1, pp. 1-48.
- Laeven, Luc, and Fabian Valencia, 2008, “Systemic Banking Crises: A New Database,” IMF Working Paper 08/224.
- Manasse, Paolo, Nouriel Roubini and Axel Schimmelpfennig, 2003, ‘‘Predicting Sovereign Debt Crises,” IMF Working Paper 03/221.
- Kumar, Manmohan S., Uma Moorthy and W. R. M. Perraudin, 2002, “Predicting Emerging Market Currency Crashes,” Working Paper No. 02/7.
- Mauro, Paolo, Nathan Sussman, and Yishay Yafeh, 2002, ‘‘Emerging Market Spreads: Then vs. Now,’’ Quarterly Journal of Economics, Vol. 117 (May), pp. 695–733.
- Nag, Ashok, and Amit Mitra, 1999, “Neural Networks and Early Warning Indicators of Currency Crises,” Reserve Bank of India Occasional Papers 20(2), pp. 183-222.
- Pescatori, Andrea and Amadou Sy, 2007, “Are Debt Crises Adequately Defined?,” IMF Staff Papers, Vol. 54(2).
- Reinhart, Carmen and Kenneth Rogoff, 2009, “The Aftermath of Financial Crises,” American Economic Review, Vol. 99 No. 2, May 2009, 466-472.
- Reinhart, Carmen and Kenneth Rogoff, 2010, “From Financial Crash to Debt Crises,” NBER Working Paper 15795.
- Sturzenegger, Federico and Jeromin Zettelmeyer, 2006, Debt Defaults and Lessons from a Decade of Crises. Table 1 in Chapter 1, Cambridge: MIT Press.
- Wong, J., E. Wong and P. Leung, 2007, A Leading Indicator Model of Banking Distress—Developing an Early Warning System for Hong Kong and Other EMEAP Economies, Working Paper 07_22, Hong Kong Monetary Authority.

### Appendix Table 1 — Pros and Cons of Early Warning Statistical Methodologies
- Methodologies listed:
  - Signaling Approach
    - Studies: Kaminsky, Lizondo, and Reinhart, 1998; IMF (2007); Allen and Johnson (2007); Hemming, Kell and Schimmelpfennig (2003); Edison (2000).
    - Pros:
      - Easily accommodate differences in data availability across variables
      - Permit a transparent mapping from indicators to aggregate indices
    - Cons:
      - Ignore interactions among variables
      - Cannot use standard statistical tests to check for robustness
  - Logit/Probit Models
    - Studies: Berg and Patillo (1999); Berg, Borensztein and Patillo (2005); Bussiere and Fratzscher (2006); Hemming, Kell and Schimmelpfennig (2003); Kumar, Moorthy and Perraudin (IMF 2002); Cihak and Schaeck (2010); Wong, Wong and Leung (2010).
    - Pros:
      - Take into account correlations across variables
      - Easily test for statistical significance of individual variables
    - Cons:
      - Constrain usable dataset when data availability varies across variables
      - Number of variables that can be meaningfully considered simultaneously is limited
      - Changes in the variables themselves and changes in their contribution to the crisis prediction not always transparent

### Appendix Table 2 — Comparison of Statistical Methodologies (Highlights)
- Berg and Patillo (1999)
  - Comparison Criteria: In-sample, Out-of-sample, Cross-country
  - Evaluation Criteria: Percent of pre-crisis periods correctly called; Percent of tranquil periods correctly called; False alarms as percent of total alarms; QPS and LPS
  - Findings:
    - Signaling methodologies outperform probit out-of-sample
    - Probit outperform signaling methodologies in cross-country studies
- Berg, Borensztein and Patillo (2005)
  - Evaluation Criteria includes: Probability of crisis given signal; Probability of crisis given no signal; Loss function = weighted sum of false alarms (as a % of total non-crisis) and missed crises (as a % of total crisis)
  - Findings:
    - Signaling methodologies perform better out-of-sample than in-sample
    - Probit’s performance deteriorates substantially over the sample
- Davis and Karim (2008)
  - Findings:
    - Logit most appropriate for global EWS; signaling approach for country-specific EWS.

### Appendix Tables 3a & 3b — Fiscal Stress Events (Selected entries)
- Advanced Economies: Fiscal Stress Events (columns include Start of Crisis; Sovereign Default or Restructuring; First Year of IMF-Supported Program; Inflation Pressures (35 percent); Bond Yield Pressure; Duration of Fiscal Crisis)
  - Examples:
    - Australia: 1986 (duration indicator 2); 1989 (1); 2008 (3)
    - Canada: 1990 (1)
    - Greece: 1993 (3); 2008 (6)
    - Iceland: 1974 (2); 1978 (6); 2008 (4)
    - Ireland: 1974 (1); 1976 (1); 2008 (6)
    - Japan: 1974 (1); 2009 (2)
    - Korea, Republic of: 1974 (8); 1983 (2); 1997 (4)
    - United Kingdom: 1970 (1); 1974 (6)
- Emerging Market Economies: Fiscal Stress Events (columns include Start of Crisis; Sovereign Default or Restructuring; First Year of IMF-Supported Program; Inflation Pressures (500 percent); Bond Yield Pressure; Duration of Fiscal Crisis)
  - Examples:
    - Argentina: 1982 (12); 1998 (1); 2000 (6)
    - Brazil: 1983 (12); 1998 (1); 2001 (2)
    - Ecuador: 1982 (14); 1999 (2); 2005 (1)
    - Mexico: 1977 (1); 1982 (9); 1995 (3); 1999 (1); 2009 (2)
    - Nigeria: 1982 (11); 2001 (1); 2004 (2)
    - Russia: 1991 (10); 2009 (1)
    - Venezuela: 1983 (23); 2008 (1); 2010 (1)
  - Sources: IMF International Financial Statistics; Standard and Poor’s; and authors' calculations.

### Appendix Tables 4a & 4b — Event Comparison Across Studies (Selected concordances and notes)
- Advanced Economies (examples):
  - Australia: Start of Crisis 1986; IMF (2007) n.a.; other studies contain various entries (1989, 2008)
  - Finland: Start of Crisis 1990; Reinhart and Rogoff list 1991
  - Korea: Start of Crisis entries include 1974, 1983, 1997 with cross-study concordance (1997 appears in multiple studies)
  - Japan: 2009 crisis appears as n.a. in several studies
- Emerging Market Economies (examples):
  - Argentina: multiple study dates including 1982, 1995, 1998, 2000, 2001 depending on study
  - Brazil: 1983, 1998, 2001 appear variably across studies
  - Indonesia: 1997 and 2002 appear across studies
  - Mexico: 1982, 1995, 1999, 2009 appear across studies
- Notes:
  - “n.a.” indicates that either the country or the year were not covered in the respective study.
  - Empty cells indicate no crisis.
  - The dating of the crises follows exactly the respective studies — events from other studies included where two consecutive crises are separated by only one year.

### Appendix Table 5 — List of Indicators and Data Sources (Selected indicators and availability)
- Fiscal Stress Variables
  - Debt default: Period of domestic or foreign bonded and bank debt default. Data availability: 1970-2010. Data source: Standard and Poor’s.
  - IMF-Supported programs: Period of IMF-supported program exceeding 100 percent of IMF member’s quota. Data availability: 1970-2010. Data source: IMF.
  - Inflation rate: In percent: period during which inflation exceeds 35 percent per annum for AE, or 500 percent per annum for EMs. Data availability: 1970-2010. Data source: IMF/IFS.
  - Bond yield pressure: Government bond spreads (relative to 10-year US Treasury bond) exceeding 2 standard deviations above the historical annual mean of the country, or 1000 basis points on annual basis; or at least 6 months in a year based on monthly data. Data availability: 1970-2010. Data source: IMF/IFS.
- Basic Fiscal Variables
  - r-g (5-year average): Imputed interest rate on general government debt, deflated by the GDP deflator, minus real GDP growth rate. Five year forward moving average. Data availability: 1971-2010 (AE); 1985-2010 (EM). Data source: WEO.
  - Cyclical adjusted primary balances: Expressed as a percentage of potential GDP. Data availability: 1979-2010 (AE); 1984-2010 (EM). Data source: WEO.
  - General government gross/net debt: Expressed in percent of GDP. Net debt used for Japan and Canada, gross debt for all other countries. Data availability: 1970-2010 (AE); 1982-2010 (EM). Data source: WEO.
- Long-Term Fiscal Trend variables
  - Total fertility rate: The average number of children per woman. Data availability: 1970-2010. Data source: UN.
  - Old age dependency ratio: 20 years ahead projections of the ratio of population over 65, divided by the number of adults. Data availability: 1970-2010. Data source: UN.
  - Long-term projections of the change in public pension expenditure: Expressed in percent of GDP, the change in projected expenditures 30 year ahead relative to the base year. Data availability: 1980-2010. Data source: IMF staff estimates.
  - Long-term projections of the change in public health expenditure: Expressed in percent of GDP, the change in projected expenditures 30 year ahead relative to the base year. Data availability: 1979-2010 (AE); 1995-2010 (EM). Data source: IMF staff estimates.
- Asset and Liability Management
  - Current gross financing need: Projected general government overall balance plus general government debt with a maturity of one year or less. Expressed in percent of GDP. Data availability: 1990-2010 (AE); 1993-2010 (EM). Data sources: WEO, VEE.
  - Share of short-term debt as a ratio of total debt: Short-term debt is defined as general government debt with remaining maturity of one year or less. Total debt is general government gross debt. Data availability: 1989-2010. Data source: BIS.
  - Debt denominated in foreign currencies: General government debt, expressed in terms of total debt. Data availability: 1990-2010 (EM). Data sources: WEO, VEE.
  - Debt held by non-residents as a proportion of total debt: Includes both domestic and foreign currency debt issued. Expressed as a proportion of total debt. Data availability: 1998-2010 (AE). Data sources: JEDH, WEO.
  - Weighted average maturity of general government debt: Historical data calculated by staff; current data available in Bloomberg. Data availability: 1980-2010. Data sources: Bloomberg; Dealogic.
  - Short-term external debt to international reserves: Short-term debt is defined as general government debt with remaining maturity of one year or less. Data availability: 1990-2010 (AE); 1970-2010 (EM). Data sources: WEO, IFS.

### Appendix Table 6 — Summary Statistics of Fiscal Indicators (1995-2010 Sample)
- Sample: 1995-2010
- Advanced Economies / Emerging Market Economies (selected indicators with Obs., Mean, St. Dev)
  - r-g (5-year average)
    - AE: Obs. 435; Mean 0.20; St. Dev 2.50
    - EM: Obs. 591; Mean -4.69; St. Dev 6.11
  - General government gross debt
    - AE: Obs. 443; Mean 60.38; St. Dev 34.13
    - EM: Obs. 629; Mean 48.81; St. Dev 31.10
  - Cyclical adjusted primary balances
    - AE: Obs. 446; Mean 0.34; St. Dev 3.42
    - EM: Obs. 654; Mean 0.43; St. Dev 9.07
  - Current gross financing need
    - AE: Obs. 450; Mean 11.28; St. Dev 10.74
    - EM: Obs. 595; Mean 10.85; St. Dev 10.54
  - Share of short-term debt as a ratio of total debt
    - AE: Obs. 450; Mean 24.35; St. Dev 16.11
    - EM: Obs. 492; Mean 22.05; St. Dev 20.83
  - Debt denominated in foreign currencies
    - Obs. 712; Mean 54.88; St. Dev 27.67
  - Debt held by non-residents as a proportion of total debt
    - Obs. 228; Mean 42.10; St. Dev 23.92
  - Weighted average maturity of general government debt
    - AE: Obs. 444; Mean 7.08; St. Dev 3.37
    - EM: Obs. 668; Mean 6.38; St. Dev 4.45
  - Short-term external debt (percent of international reserves)
    - Obs. 705; Mean 73.22; St. Dev 87.49
  - Total fertility rate
    - AE: Obs. 441; Mean 0.52; St. Dev 0.28
    - EM: Obs. 822; Mean 0.78; St. Dev 0.76
  - Old age dependency ratio
    - AE: Obs. 448; Mean 31.53; St. Dev 6.65
    - EM: Obs. 832; Mean 17.33; St. Dev 8.06
  - Long-term projections of the change in public pension expenditure
    - AE: Obs. 385; Mean 2.22; St. Dev 2.00
    - EM: Obs. 89; Mean 2.07; St. Dev 2.45
  - Long-term projections of the change in public health expenditure
    - AE: Obs. 396; Mean 3.69; St. Dev 1.30
    - EM: Obs. 336; Mean 1.74; St. Dev 1.14
- Sources: IMF WEO and IFS; BIS; Bloomberg; Dealogic; and authors’ calculations.

*Source: _wp11100 - REFERENCES (IMF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11100.pdf_