## wpiea2020220-print-pdf

## Source details

**Canonical URL:** [wpiea2020220-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020220-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020220-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020220-print-pdf.pdf.json)

---

### I. Introduction — scope and objectives
- Assesses short- and medium-run impacts on fiscal fluctuations of key determinants: global growth, commodity prices, world risk-appetite (VIX), political shocks, financial shocks, demographic shocks, and domestic growth surprises.
- Fiscal outcomes analyzed: ratios to GDP of overall fiscal deficit, expenditure, revenue, gross government debt, and net international reserves.
- Emphasis on heterogeneity by country characteristics (AE, EMDC, commodity-exporter/importer) and special focus on Latin America (LAC).
- Quantifies fiscal costs of shocks by combining impacts and volatility; decomposes contributions to fiscal-balance volatility via variance decomposition.
- Evaluates potential of state-contingent instruments (e.g., GDP-indexed bonds) to smooth fiscal fluctuations and insure government liabilities from domestic shocks.

### II. Data, sample, and variable treatment
- Sample: unbalanced panel up to 194 countries, yearly frequency 1970–2017; most specifications focus post-1986; typical coverage about 140 countries with about 25 years of data.
- Variable availability notes:
  - VIX from 1986; elections from 1995; commodity TOT shocks from 1980; fiscal variables and domestic growth often from 1970.
- Outlier exclusions: years with annual GDP growth > 100 percent or inflation/exchange rate variation > 200 percent.
- Transformations:
  - real GDP, consumer price inflation, commodity price indices, exchange rates, terms-of-trade indices: yearly percentage changes;
  - population, employment, life expectancy: deviations from country-specific quadratic time trends;
  - expenditure, revenue, fiscal balance measures: percentage of GDP.
- Fiscal data sources: IMF World Economic Outlook; HPDD for gross government debt; international reserves from External Wealth of Nations.

### III. Estimation strategy and identification
- Methods:
  - OLS Local Projection (Jordà, 2005) for horizon analysis (year of impact to H up to 4), controls for lags and polynomials A_h(L), B_h(L), β_h(L); robust country-clustered SEs; 90 percent and 95 percent CIs plotted.
  - Panel fixed effects and Arellano-Bond system-GMM for dynamic-panel bias correction.
- Treatment of shocks:
  - Exogenous shocks analyzed without controlling for contemporary domestic growth to capture total fiscal effect.
  - When estimating impact of domestic real GDP growth on fiscal outcomes, use LP-IV: instrument domestic growth with world growth interacted with trade openness.

### IV. Overview of fiscal volatility (select statistics)
- Average fiscal deficit in sample: 3% of GDP, standard deviation about 14%.
- Excluding bottom and top 1% observations: average deficit 2.7% and standard deviation 5.5%.
- Within-country std. dev. of overall fiscal deficit: 12% (4.6% excluding top/bottom 1%).
- Within-country std. dev.: expenditure 12.5%; revenue 7%.
- Selected table entries (preserved exactly):
  - All economies: Mean -3.0%, Std. dev 13.8%, (Within-country) std. dev 12%, ρ [AR(1)] 0.62
  - Advanced economies: Mean -1.7%, Std. dev 4.9%, (Within-country) std. dev 3%, ρ 0.75
  - EM and developing: Mean -3.3%, Std. dev 15.4%, (Within-country) std. dev 13%, ρ 0.62
  - LAC countries: Mean -3.0%, Std. dev 4.9%, (Within-country) std. dev 4%, ρ 0.76
  - Commodity exporters: Mean -3.2%, Std. dev 23.7%, (Within-country) std. dev 20%, ρ 0.63
- LAC subsample (32 countries, on average 40 years): within-country std. dev. of overall fiscal deficit 4.6% of GDP; mean deficit 3.1% of GDP.

### V. Dynamic impacts of exogenous global and domestic shocks — main empirical findings
- World growth (PPP-weighted, annual real GDP growth, percent):
  - A 1 percentage point increase in world growth → improvement in fiscal balance by somewhat more than ½ of a percentage point of GDP on impact.
  - EMDC: effect mainly via higher revenue ratios; AE: via lower expenditure ratios (denominator effect).
  - Persistence: in emerging economies, improvement ≈ 0.3% of GDP persists four years after impact.
  - Panel coefficient: annual world real GDP growth on Fiscal balance: 0.396 **** [0.0966].
- Terms-of-trade and commodity price shocks:
  - One-standard-deviation TOT increase: notable GDP increase (especially EMDC); initial fiscal balance improvement ≈ ½ percent of GDP; expenditure increases erode fiscal space over time.
  - Export commodity price one-standard-deviation increase (commodity exporters): initial fiscal balance improvement ~1–1½ percentage points of GDP; effect declines over time; reserves increase in commodity exporters.
  - Import commodity price increases: negative effect on fiscal balance for advanced or emerging importers of about ½ percentage point at peak.
  - Oil trade balance: a 1 percentage point of GDP improvement in oil trade balance → 0.3 percentage point increase in fiscal balance (LAC result) and long-term decrease in debt ≈ 2 percentage points.
- Global risk appetite (standardized VIX):
  - A one standard deviation increase in VIX → worsens fiscal balance by about 0.5 percent of GDP on impact, rising after one year to around 1.5 percentage points of GDP, fading after 4 years.
  - After two years, gross debt to GDP increases by four percentage points in AE and six percentage points in EMDC.
  - Reserves decline in EMDC and increase in AE.
  - Panel coefficient: Standardized VIX on Fiscal balance: -0.244 **** [0.0607].
- US 10-year yield (percent):
  - A one point rise in US long-term yields (controlling for Fed funds) → GDP increases in AE and smaller in EMDC; progressive improvement in fiscal balance in both (about 1 percentage point after 2 years); EMDC gross debt tends to rise (higher interest burden).
  - Panel coefficients on Revenue/Expenditure: US 10-year on Revenue -0.133 *** [0.0461]; on Expenditure -0.146 *** [0.0468].
- Political shocks:
  - Election year dummy (shorter sample): associated with higher spending, worse fiscal balance, rising debt in EMDC; Election year on Fiscal balance: -0.387 ** [0.174].
  - Transition year dummy (PolityIV) on Fiscal balance: -3.759 * [1.956] (panel main); transitions out of democracy in EMDC produce very large persistent negative fiscal impacts (debt rises around 20 percentage points after two years in some estimates).
- Banking crises:
  - Larger fiscal impact in AE than EMDC: AE banking crises heavily worsen expenditure, revenues, fiscal balance, debt ratios and real GDP; EMDC effects more muted.
  - Systemic banking crisis on Gross debt: 5.839 ** [2.278] (panel main).
- Demographic shocks (population deviation):
  - AE: population increases → persistent deterioration of fiscal balance and debt via higher expenditure and lower GDP; Log population deviation on Fiscal balance: -0.140 ** [0.0546].
  - EMDC: population increase → positive GDP effect, declines in both revenue and expenditure ratios, fiscal balance roughly unchanged.

### VI. Domestic growth (LP-IV) — magnitudes and debt implications
- LP-IV instruments: world growth × trade openness.
- Main LP-IV findings (All countries; coefficient and SE preserved exactly):
  - Annual real GDP growth on Fiscal balance: 0.756**** [0.0700]
  - Revenue: 0.266**** [0.0609]
  - Expenditure: -0.486**** [0.0495]
  - Gross debt: -1.774**** [0.186]
  - Net debt: -0.986**** [0.233]
  - Reserves to GDP: -0.424**** [0.100]
- By group (selected):
  - Advanced economies:
    - Fiscal balance: 0.522**** [0.0514]; Revenue: 0.0658 [0.0415]; Expenditure: -0.470**** [0.0447]; Gross debt: -0.888**** [0.0978]
  - Emerging economies:
    - Fiscal balance: 0.798**** [0.0937]; Revenue: 0.331**** [0.0830]; Expenditure: -0.457**** [0.0658]; Gross debt: -2.233**** [0.269]
- Interpretation:
  - EMDC: a 1 percentage point domestic growth increase improves fiscal balance by around 1 percentage point of GDP on impact (earlier text).
  - AE: improvement ≈ 0.5 percentage points of GDP on impact.
  - Debt-to-GDP drops: EMDC close to 2.2 percent (LP-IV) or close to 4 percent over time per other decompositions; AE about 0.9–1 percent after three years.
- Panel IV robustness: domestic growth improves fiscal balance by around 0.8% of GDP per additional point of growth; split ~0.5 percentage points decrease in expenditure-to-GDP and ~0.3 percentage points increase in revenue-to-GDP on impact.

### VII. State-contingency parameters (implications for GDP-indexed debt)
- Translate LP-IV into debt-growth elasticities (real level debt response to a 1% growth shock):
  - Full sample:
    - 휂_D,0 = −2% (impact)
    - 휂_D,1 = −3.8% (after one year; stabilizes afterwards)
    - Interpretation: embedding state-contingency would imply offsetting a 4% reduction after two years in principal value of debt for each 1% unexpected growth shock to stabilize debt ratios on average.
  - By group (after two years):
    - Advanced economies: 휂_D,1_AD = −1.2%
    - Emerging and developing economies: 휂_D,1_EMDC = −4.9%
    - LAC countries: 휂_D,1_FAL = −3.7%

### VIII. Drivers of fiscal fluctuations and variance decomposition
- Principal drivers: very rare events (political transitions, banking crises) and global factors (world growth, VIX, commodity prices).
- Examples of magnitudes:
  - Three years after a 1 standard deviation global growth shock → debt to GDP of both AE and EMDC increase by about 3 percentage points.
  - Three years after a VIX shock of similar magnitude → debt to GDP increase by about 5 percentage points.
- Decomposition patterns:
  - Most shocks explain smaller share of fiscal balance variation in AE than in EMDC.
  - Explanatory role of most shocks grows over time except global growth (role declines after one year).
  - Most important shocks: VIX (especially for financially open countries), commodity prices, global growth.
  - Expenditure ratio explains a larger share of short-term fiscal fluctuations than revenue, especially in EMDC and LAC.

### IX. LAC-specific annex findings (selected)
- Imported commodity price shock (one standard deviation) for LAC importers:
  - Decreases fiscal balance by about 0.7% of GDP after a year, persisting for three years.
  - LAC importers experience GDP drop close to 1 percentage point after the shock.
- Export commodity price shock (one standard deviation) for LAC exporters:
  - Increases fiscal balance by about ½ percentage point of GDP after a year; raises GDP by similar amount after 2–3 years.
- Domestic growth shock in LAC:
  - Fiscal balance response somewhat smaller but more persistent than other EMDC; muted early revenue and expenditure responses; medium-term fiscal balance remains positive; gross debt response similar to other EMDC.

### X. Policy implications and recommendations
- Macroeconomic insurance and risk-pooling:
  - Global factors materially affect fiscal risk → demand for insurance against global shocks; scope for efficient risk-pooling across countries (not only public-private insurance).
  - Domestic shocks also matter → supports cross-country risk pooling.
- Debt-contract design:
  - GDP-indexed bonds or other state-contingent instruments could smooth fiscal impacts.
  - Estimates suggest a one percentage point negative domestic growth shock should be followed by about a 4% reduction in the value of debt in the medium-term to stabilize debt ratios on average.
  - Need for state-contingency: about 1% in AE versus about 5% in EMDC.
- Further research priorities:
  - Distinguish automatic stabilizers from discretionary fiscal reactions.
  - Identify existing financial instruments that hedge exogenous shocks.
  - Define potential new optimal state-contingent debt contracts.

*Source: wpiea2020220-print-pdf (excerpt provided).*

### REFERENCES _________________________________________________________________ 40

### wpiea2020220-print-pdf - REFERENCES _________________________________________________________________ 40

### I. Introduction — Motivation
- Government fiscal balances vary significantly over time, especially in emerging economies, due to global factors and a changing domestic economic environment.
- Fiscal volatility creates risks including heightened borrowing costs, lack of fiscal space, foregone investment opportunities, and costly defaults.
- Headline fiscal variables used for assessment include the budget deficit and the debt level; these are also metrics for budget rules.
- This paper assesses the short and medium-run impact on fiscal fluctuations of key determinants: global growth, commodity prices, world risk-appetite conditions, political shocks, financial shocks, demographic shocks, and domestic growth surprises.
- Fiscal outcomes measured: ratios of fiscal deficit, expenditure, revenue, and gross debt to GDP; also net international reserves to GDP.
- Focus on heterogeneity by country characteristics (level of development, commodity-exporter/importer status) and special attention to Latin America within emerging and developing countries.
- The paper quantifies fiscal costs associated with shocks by combining impacts and volatility of shocks, and decomposes contributions of shocks to fiscal-balance volatility via variance decomposition.
- The analysis is positioned to inform the potential design of synthetic assets or existing financial instruments to hedge fiscal loadings on exogenous global risk factors; detailed analysis of hedging instruments is left for future work.
- The paper also quantifies the extent to which state-contingent fiscal instruments (such as GDP-indexed bonds) could smooth fiscal fluctuations and insure government liabilities from domestic shocks.

### I. B. Literature Review — Key strands
- Three major strands of relevant literature:
  - Theoretical tax smoothing literature: optimal taxation/debt paths for exogenous spending or macro shocks (references: Barro, 1979; Mendoza and Oviedo, 2006; Marcet and Scott, 2009; Bohn, 1990; Angeletos, 2002).
  - Empirical “sources of volatility” literature: determinants of fiscal outcomes and macroeconomic volatility (references include Raddatz, 2008; Henzel and Rengel, 2017; Agnello and Sousa, 2009; Tujula and Wolswijk, 2004; Lusinyan et al., 2009; Cerovic et al., 2018).
  - “Macroeconomic hedging” and “sovereign insurance” literature: statistical properties of macro risk and design of assets for insurance (references include Caballero and Panageas, 2004; Ma and Valencia, 2018; Bali et al., 2014; IMF, 2017; Alfaro and Kanczuk, 2005; Barr, Bush, and Pienkowski, 2014; Borensztein and Mauro, 2002; Durdu, 2009; Hatchondo and Martinez, 2012; Sandleris et al., 2011; Crucini, 1999; Callen et al., 2015; Ghesquiere and Mahul, 2007).
- Empirical focus on Latin America reflects region-specific features: history of fiscal crises, limited domestic financial depth, weaker institutions, and integration into world production and financial networks (references: Caballero, 2000; Edwards, 2006; Raddatz, 2008; Klemm, 2014; BIS, 2016).

### Our contribution
- Quantifies sources of fiscal risk in the short and medium term across advanced, emerging, and developing countries, with specific focus on Latin America.
- Separately analyzes contributions of fiscal-balance subcomponents: expenditure, revenue, and of debt.
- Provides estimates of fiscal variable sensitivity to exogenous global macroeconomic risk, linking to macro risk hedging literature.
- Uses external instruments to quantify the impact of GDP growth on debt paths at various horizons, parameterizing potential for state-contingency to insulate government liabilities from temporary macro shocks.

### II. Data and Methodology — Data
- Fiscal outcomes: ratios to GDP of overall fiscal deficit, expenditure, revenue, gross government debt, and net international reserves.
- Shocks classified into thematic blocks covering world macroeconomic fluctuations, trade-related variables, global financial conditions, banking crises, and other domestic shocks (political developments, demographic shocks, natural disasters).
- Thematic blocks described:
  1. World macroeconomic block:
     - World growth and world inflation, both weighted by purchasing power parity GDP.
     - World inflation used as a control in the linear projections panel estimation.
     - Average level of the VIX volatility index over a year as an indicator of global risk appetite.
     - Average US 10-year yield as a proxy for global growth expectations.
  2. Trade block:
     - Change in the goods and services terms of trade.
     - Change in export and import commodity prices (calculated as a trade weighted average).
     - The oil trade balance.
  3. “Financial conditions” block:
     - Global financial conditions as experienced by a country proxied by the interaction of the VIX with financial creditor or debtor status defined as the value of net foreign assets to GDP.
     - Domestic systemic banking crises.

*Source: wpiea2020220-print-pdf - REFERENCES _________________________________________________________________ 40*

### 4.      an “other domestic  shocks”  block:  this  includes  major political  regime

### wpiea2020220-print-pdf - 4.      an “other domestic  shocks”  block:  this  includes  major political  regime

### Data, sample, and variable treatment
- Unbalanced sample of a maximum of 194 countries, yearly frequency spanning from 1970 to 2017; most specifications focus on the post-1986 period.
- Typical sample coverage: about 140 countries with about 25 years of data.
- Variables availability notes:
  - VIX index available from 1986.
  - Elections available from 1995.
  - Commodity terms-of-trade shocks available from 1980.
  - Fiscal variables and domestic growth available starting from 1970 for many countries.
- Outlier exclusions: any year with annual GDP growth higher than 100 percent or inflation or exchange rate variation against the dollar larger than 200 percent (results virtually unchanged when using full sample).
- Subsamples explored: advanced economies (AE), emerging and developing countries (EMDC), Latin American and Caribbean countries (LAC); also splits by external debtors/creditors, small versus large states (population larger than 1mn people), and more versus less open countries (above/below median trade openness).
- Transformations for stationarity:
  - real GDP, consumer price inflation, commodity price indices, exchange rates, terms-of-trade indices: yearly percentage changes;
  - population, employment, life expectancy: deviations from country-specific quadratic time trends;
  - expenditure, revenue, fiscal balance measures: percentage of GDP.
- Data coverage summary:
  - headline deficit, expenditure and revenue: 195 countries, on average 35 years;
  - public debt ratios: 176 countries for 36 years on average.

### Overview of fiscal volatility (key statistics)
- Average fiscal deficit in sample: 3% of GDP, standard deviation about 14%.
- Excluding bottom and top 1% observations: average deficit falls to 2.7% and standard deviation to 5.5%.
- Within-country standard deviation: 12% (4.6% excluding top and bottom 1% observations).
- Within-country standard deviation of expenditure: 12.5%; of revenue: 7%.
- Table (selected entries preserved exactly):
  - All economies: Mean -3.0%, Std. dev 13.8%, (Within-country) std. dev 12%, ρ [AR(1)] 0.62
  - Advanced economies: Mean -1.7%, Std. dev 4.9%, (Within-country) std. dev 3%, ρ 0.75
  - EM and developing: Mean -3.3%, Std. dev 15.4%, (Within-country) std. dev 13%, ρ 0.62
  - LAC countries: Mean -3.0%, Std. dev 4.9%, (Within-country) std. dev 4%, ρ 0.76
  - Commodity exporters: Mean -3.2%, Std. dev 23.7%, (Within-country) std. dev 20%, ρ 0.63
- LAC subsample (32 countries, on average 40 years):
  - within-country standard deviation of overall fiscal deficit: 4.6% of GDP;
  - mean deficit: 3.1% of GDP;
  - corresponding AE values: about 5% standard deviation and 1.7% of GDP deficit;
  - all EMDC: 15 (standard deviation?) and 3.4% deficit (text preservation: "they are 15   and 3.4% deficit").
- In LAC: expenditure to GDP within-country std. dev 6.4%; revenue to GDP within-country std. dev 5.2%.

### Estimation strategy (methods and identification)
- Two main methodologies used:
  - OLS Local Projection (LP) Method of Jorda (2005) for dynamic short- and medium-term responses (year of impact to next 3 years; horizons up to H=4 shown).
    - Controls for lags of dependent and independent variables, 2-lags polynomials A_h(L), B_h(L), β_h(L).
    - Inference via dynamic panel OLS variance-covariance matrices; robust standard errors clustered at country level; 90 percent and 95 percent confidence intervals plotted.
    - Local projection does not by itself deliver causal identification beyond plausibly exogenous regressors.
  - Panel fixed-effects estimation (within-country variation) and Arellano-Bond generalized method of moments to correct bias from lagged dependent variables.
- Treatment of exogenous shocks:
  - Exogenous shocks analyzed without controlling for contemporary domestic growth (to capture total fiscal effect including via domestic growth).
  - When estimating the impact of domestic real GDP growth on fiscal outcomes, separate regressions use instrumental variables to account for endogeneity (growth instrumented via global dynamics).
  - Instrument for domestic growth in LP-IV: world growth interacted with trade openness.

### Dynamic impact of exogenous shocks — main empirical findings

- World growth (annual world real GDP growth, percent):
  - A 1 percentage point increase in PPP GDP weighted world growth is associated on the year of impact with an improvement in the fiscal balance by somewhat more than ½ of a percentage point of GDP.
  - Mechanisms:
    - EMDC: effect operates mainly via higher revenue ratios.
    - AE: effect operates mainly via lower expenditure ratios (denominator effect from stronger GDP response).
  - Persistence: in emerging economies, improvement around 0.3% of GDP persists four years after initial impact.
  - Table coefficient (panel): annual world real GDP growth, percent shows coefficient 0.396 **** [0.0966] on Fiscal balance (main spec).

- Terms-of-trade and commodity price shocks:
  - One-standard-deviation terms-of-trade increase:
    - notable increase in GDP, especially in EMDC;
    - EMDC: more positive effects on revenue and expenditure ratios than AE; initial fiscal balance improvement about ½ percent of GDP but expenditure increases erode fiscal space over time.
    - AE: moderate positive effect on fiscal balance mainly via GDP-driven lower expenditure ratio; revenue-to-GDP reaction in AE is flat.
    - Reserves increase mainly for EMDC and not for AE.
  - Export commodity price shock (one standard deviation increase, export-weighted):
    - For commodity exporters initial impact ~1-1½ percentage points of GDP improvement in fiscal balance (roughly equally from revenue increase and expenditure/GDP decline).
    - Effect declines over time, revenue effect dissipates; negative effect on debt ratio strong and persistent; reserves increase in commodity exporters.
    - For non-exporters, effect is very small.
  - Import commodity price increases:
    - Negative effect on fiscal balance for advanced or emerging importers of about ½ percentage point at peak.
    - AE: effect mainly via higher expenditure ratios and stronger GDP decline (swells debt ratios).
    - EM importers: effect mainly via lower revenue ratios; emerging importers experience declining debt ratios and reserves decrease.
  - Oil trade balance shock:
    - Fiscal balance improves in both exporters and importers; effect stronger in EMDC due to revenue response and GDP increase; reserves increase more persistently in EMDC.

- Global risk appetite (VIX, standardized):
  - A one standard deviation increase in the VIX worsens the fiscal balance by about 0.5 percent of GDP on impact, growing after one year to around 1.5 percentage point of GDP, fading only after 4 years.
  - Channels: lower revenue and higher expenditure ratios; expenditure effect stronger in AE due to larger GDP drop (denominator effect).
  - After two years, gross debt to GDP ratio increases by four percentage points of GDP in AE and six percentage points in EMDC.
  - Reserves decline in EMDC and increase in AE (capital flows to safer countries).
  - Effect more pronounced and persistent among external debtors versus external creditors.

- US interest rates (US government 10-year bond yield, percent):
  - Following a rise of one point in US long-term yields (controlling for Fed funds), GDP increases in AE (and smaller in EMDC).
  - Progressive improvement in fiscal balance in both AE and EMDC (about 1 percentage point after 2 years).
  - Despite fiscal balance improvement, gross debt in EMDC tends to rise, likely reflecting higher interest burden.

- Political shocks and conflicts:
  - Election years (dummy):
    - Associated with higher spending, worse fiscal balance, and rising debt in EMDC—effect mainly significant in medium term—but not in AE.
    - Elections tested separately in shorter sample (post-1995); election year dummy coefficient in panel-with-elections: Fiscal balance -0.387 ** [0.174]; Expenditure 0.263 ** [0.129].
  - Political transition years (Polity IV transition dummy):
    - In EMDC, associated with large decline in GDP, lower revenue, higher expenditure, lower overall fiscal balance, and increased debt to GDP ratio; estimates imprecise due to limited events.
    - Transitions out of democracy in EMDC: very large and persistent negative impact on fiscal balance mainly via substantially higher spending and lower GDP; persistent higher public debt ratios (magnitudes rising around 20 percentage points of GDP increases after two years).
    - Transitions into democracy in EMDC: improvement in fiscal position, associated with increase in GDP and revenue ratios; ensuing decline in debt possibly reflecting debt restructuring/relief.
  - Military spending increases in EMDC:
    - Similar qualitative effect to transitions out of democracy: rising expenditure ratios, worsening revenue and fiscal balance ratios, declining GDP, rising public debt ratios.
    - In AE, military spending associated with increase in growth over time and a stronger decline in expenditure ratios (reflecting better growth and fiscal discipline).

- Domestic financial shocks:
  - Banking crisis:
    - Larger impact on fiscal variables in AE than EMDC.
    - In AE: banking crisis heavily worsens expenditure, revenues, fiscal balance, debt ratios, and significantly affects real GDP (partly due to bank recapitalizations).
    - In EMDC: effects more muted for fiscal ratios and GDP; effect on revenues similar to AE; public debt effect less visible (possible sovereign restructurings), fiscal balance effect less persistent.

- Demographic shocks (population deviations from quadratic trend):
  - In AE: population increases associated with persistent and significant deteriorations of fiscal balance and debt to GDP ratios, mostly via higher expenditure ratios (despite higher revenue ratios) and lower GDP (aging), reserves decline in AE.
  - In EMDC: population increase associated with positive GDP effect, declines in both revenue and expenditure ratios, leaving fiscal balance roughly unchanged.

### Panel regression main results (selected coefficients and diagnostics)
- Main specification (All countries, main specification) — selected coefficients preserved exactly:
  - L.Overall fiscal balance, % of GDP: lag coefficient 0.640 **** [0.0828]
  - Annual world real GDP growth, percent on Fiscal balance: 0.396 **** [0.0966]
  - Standardized VIX on Fiscal balance: -0.244 **** [0.0607]
  - Commodity export price change, normalized x exporter on Fiscal balance: 1.390 **** [0.314]
  - Commodity import price change, normalized on Fiscal balance: 0.627 *** [0.215]
  - US government 10-year bond yield, percent on Revenue: -0.133 *** [0.0461]; on Expenditure: -0.146 *** [0.0468]
  - Transition year dummy, from PolityIV on Fiscal balance: -3.759 * [1.956]
  - Systemic banking crisis from Laeven-Valencia on Gross debt: 5.839 ** [2.278]
  - Log population, deviation from quadratic trend on Fiscal balance: -0.140 ** [0.0546]
  - Damages from natural disasters, USD bn on Fiscal balance: -0.226 ** [0.0949]
  - Adj. R-Square for Fiscal balance equation: 0.373; Observations 4791; Countries 175.
- All countries, main specification, with elections (shorter sample) — selected coefficients preserved exactly:
  - L.Overall fiscal balance, % of GDP lag: 0.540 **** [0.104]
  - Standardized VIX on Fiscal balance: -0.245 *** [0.0732]
  - Commodity export price change x exporter on Fiscal balance: 1.299 **** [0.326]
  - US government 10-year bond yield, percent on Fiscal balance: 0.153 ** [0.0706]
  - Transition year dummy, from PolityIV on Fiscal balance: -4.532 [2.775]
  - Election year dummy on Fiscal balance: -0.387 ** [0.174]
  - Adj. R-Square for Fiscal balance equation: 0.304; Observations 3494; Countries 170.

### Robustness and dynamic panel corrections
- Arellano-Bond system-GMM implemented to correct dynamic-panel bias; results broadly similar in sign and magnitude to main specifications though some persistence coefficients reduced.
- Arellano-Bond (selected):
  - L.Overall fiscal balance, % of GDP lag: 0.582 **** [0.0849]
  - Annual world real GDP growth, percent on Fiscal balance: 0.365 *** [0.130]
  - Standardized VIX on Fiscal balance: -0.267 *** [0.100]
  - Commodity export price change x exporter on Fiscal balance: 1.469 **** [0.404]
  - Transition year dummy, from PolityIV on Fiscal balance: -2.081 ** [0.840]
  - Damages from natural disasters x log population on Net debt: -0.0855 **** [0.0248]
  - Observations 4629 (varies by equation); standard errors in brackets; significance markers preserved.

### Estimating the impact of domestic growth (instrumental variable analysis)
- LP-IV setup:
  - Instruments: world growth interacted with trade openness (plausibly exogenous under the assumption channel is via domestic GDP growth).
  - LP-IV regressions estimate dynamic response of fiscal deficit to instrumented domestic growth for horizons h=(0,...,4).
- Main LP-IV findings:
  - Domestic growth has a strong positive impact on the fiscal balance.
  - Magnitudes:
    - EMDC: a 1 percentage point increase in real GDP growth improves the fiscal balance by around 1 percentage point of GDP on impact.
    - AE: a 1 percentage point increase in real GDP growth improves the fiscal balance by around 0.5 percentage points of GDP on impact.
  - Decomposition:
    - Initial negative effect on expenditure to GDP ratios about 0.5 percentage points of GDP (similar in AE and EMDC).
    - Revenue to GDP: virtually unchanged in AE on impact; rises by about 0.5 percentage points in EMDC.
  - Persistence:
    - Domestic growth shocks persist over time in EMDC but dwindle after one year in AE.
    - Resulting debt-to-GDP responses: EMDC experience much larger reductions in debt over time (close to 4 percent of GDP) than AE (close to 1 percent after three years).
- Panel IV estimates (robustness):
  - Consistent with LP-IV: domestic growth improves fiscal balance by around 0.8% of GDP per additional point of domestic real GDP growth.
  - Effect split between ~0.5 percentage points decrease in expenditure-to-GDP and ~0.3 percentage points increase in revenue-to-GDP on impact (text preserved up to where source ends).

*Italic: Source: wpiea2020220-print-pdf (excerpt provided).*

### 0.3 percentage points of GDP.

### wpiea2020220-print-pdf - 0.3 percentage points of GDP.

### Key empirical findings: response of fiscal aggregates to domestic GDP shocks
- Positive domestic growth shocks substantially improve the fiscal balance, mainly by lowering expenditure ratios, and—in EMDC—by raising revenue ratios.
- Effect sizes (local projection IV estimates; coefficient and standard error in brackets):
  - All countries (Annual real GDP growth, percent):
    - Fiscal balance: 0.756**** [0.0700]
    - Revenue: 0.266**** [0.0609]
    - Expenditure: -0.486**** [0.0495]
    - Gross debt: -1.774**** [0.186]
    - Net debt: -0.986**** [0.233]
    - Reserves to GDP: -0.424**** [0.100]
  - Advanced economies:
    - Fiscal balance: 0.522**** [0.0514]
    - Revenue: 0.0658 [0.0415]
    - Expenditure: -0.470**** [0.0447]
    - Gross debt: -0.888**** [0.0978]
    - Net debt: -0.746**** [0.109]
    - Reserves to GDP: -0.391*** [0.146]
  - Emerging economies:
    - Fiscal balance: 0.798**** [0.0937]
    - Revenue: 0.331**** [0.0830]
    - Expenditure: -0.457**** [0.0658]
    - Gross debt: -2.233**** [0.269]
    - Net debt: -1.235**** [0.368]
    - Reserves to GDP: -0.386**** [0.112]
- Cross-group comparisons:
  - The effect on expenditure is similar in AE and EMDC, but the impact on revenue to GDP ratios is virtually zero in AE and sizable in EMDC, yielding a smaller overall impact on the fiscal balance in AE (about 0.5 percent) than EMDC (about 0.8 percent).
  - Gross debt to GDP drops by about 2.2 percent of GDP in emerging economies following a one percentage point domestic growth shock, versus 0.9 percent in advanced economies.

### State-contingency parameters for public debt (GDP-indexed implications)
- Method: translate LP IV coefficients into real debt-growth elasticities (state-contingent reduction in real debt value required to neutralize a 1 percentage point growth shock on debt ratios).
- Key elasticity estimates (real level of debt response to a one percent growth shock):
  - Full sample:
    - 휂_D,0 = −2% (impact)
    - 휂_D,1 = −3.8% (after one year; stabilizes afterwards)
    - Interpretation: embedding state-contingency would imply offsetting a 4% reduction after two years in the principal value of the debt for each 1% unexpected growth shock to stabilize debt ratios on average.
  - By country group and horizon (same shock and horizon referenced):
    - Advanced economies: 휂_D,1_AD = −1.2% (after two years)
    - Emerging and developing economies: 휂_D,1_EMDC = −4.9% (after two years)
    - LAC countries: 휂_D,1_FAL = −3.7% (after two years)

### Drivers of fiscal fluctuations and variance decomposition
- Largest drivers of fiscal fluctuations across countries: very rare events (political transitions, banking crises) and global factors (world growth, VIX, commodity prices).
- Magnitudes and examples:
  - Three years after a 1 standard deviation global growth shock, debt to GDP of both AE and EMDC would increase by about 3 percentage points.
  - Three years after a VIX shock of similar magnitude, debt to GDP would increase by about 5 percentage points.
  - Banking crises and political transitions produce much larger effects, highly dependent on country type and transition type.
- Variance decomposition findings:
  - Most shocks account for a smaller share of fiscal balance variation in advanced economies than in EMDC.
  - The role of most shocks in explaining fiscal deficit variance tends to grow over time, except the global growth shock whose role tends to decline after one year.
  - The most important shocks in explaining fiscal variance are the VIX (especially for financially open countries), commodity prices, and global growth.
  - Expenditure ratio is a much more important predictor of short-term fiscal fluctuations and explains a relatively larger share of forecast errors in emerging economies.

### Time profile and heterogeneous effects across shocks and country groups
- Time patterns:
  - Strong initial impacts for most shocks.
  - Some shocks die off over time (world growth); others persist (commodity prices, political shocks); financial shocks (VIX, banking crises) often show hump-shaped patterns peaking after one or two years; demographic shocks have limited initial impact but grow over time.
- Heterogeneous responses:
  - World growth or positive terms-of-trade shocks: improve fiscal balance—via lower expenditure in AE and higher revenue in EMDC; lower debt while increasing GDP.
  - Global financial shocks (VIX increase): worsen fiscal balance via lower revenue and higher expenditure; adversely affect debt ratios and GDP.
  - EMDC-specific: transitions out of democracy, increases in military spending, or elections tend to worsen fiscal balance and debt ratios mainly via higher spending and lower GDP; transition into democracy tends to have the opposite effect.
  - Domestic banking crisis: heavily worsens expenditure, revenue, fiscal balance and debt ratios, and adversely affects GDP; effects stronger in AE than EMDC.
  - Population increases (relative to trend): in AE associated with persistent and significant deteriorations of fiscal balance and debt ratios (higher expenditure and lower GDP); in EMDC GDP effect is positive but revenue and expenditure ratios decline, leaving fiscal balance roughly unchanged.
  - Natural disasters: worsen fiscal balance and debt ratios as well as GDP, mainly for smaller countries.
  - Reserves: increase substantially after favorable TOT shocks in EMDC (especially commodity exporters); reserves decline in EMDC and increase in AE after an adverse VIX shock.

### Policy implications and recommendations
- Macroeconomic insurance considerations:
  - Global factors matter for fiscal risk, implying demand for insurance against global shocks; only small countries may be able to acquire such insurance.
  - Differential fiscal responses across countries imply scope for efficient risk-pooling across countries (not only public-private insurance).
  - Domestic shocks are relevant, further supporting scope for risk pooling across countries.
- Debt-contract design:
  - GDP-indexed bonds or other state-contingent debt instruments could help smooth fiscal impacts: estimates suggest a one percentage point negative domestic growth shock should be followed by about a 4% reduction in the value of debt in the medium-term to stabilize debt ratios on average.
  - Need for state-contingency is smaller in advanced economies (about 1%) than in EMDC (about 5%).
- Further research areas noted:
  - Distinguish automatic stabilizers from discretionary fiscal reactions to shocks.
  - Identify existing financial instruments that hedge against exogenous shocks.
  - Define potential new optimal state-contingent debt contracts.

*Source: wpiea2020220-print-pdf - 0.3 percentage points of GDP.*

### REFERENCES

### REFERENCES

### Scholarly and working-paper sources
- Abbas, S. A., Belhocine, N., El-Ganainy, A., & Horton, M. (2011). Historical patterns and dynamics of public debt—evidence from a new database. IMF Economic Review, 59(4), 717-742.
- Agnello, L., & Sousa, R. M. (2009). The determinants of public deficit volatility, ECB Working Paper Series, n. 1042.
- Aguiar, M., & Gopinath, G. (2007). Emerging market business cycles: The cycle is the trend. Journal of political Economy, 115(1), 69-102.
- Alberola, E., Kataryniuk, I., Melguizo, A., & Orozco, R. (2016). Fiscal policy and the cycle in Latin America: the role of financing conditions and fiscal rules, Banco de España Working Paper No. 1604.
- Alfaro, L. and F. Kanczuk (2005). “Sovereign debt as a contingent claim: a quantitative approach”. In: Journal of International Economics 65.2, pp. 297–314.
- Ang, A., Piazzesi, M., & Wei, M. (2006). What does the yield curve tell us about GDP growth?. Journal of econometrics, 131(1-2), 359-403.
- Angeletos, G. M. (2002). Fiscal policy with noncontingent debt and the optimal maturity structure. The Quarterly Journal of Economics, 117(3), 1105-1131.
- Aslam, A., Beidas-Strom, S., Bems, R., Celasun, O., Çelik, S. K., & Koczan, Z. (2016). Trading on Their Terms? Commodity Exporters in the Aftermath of the Commodity Boom. International Monetary Fund Working Paper 16-27.
- Bali, T. G., Brown, S. J., & Caglayan, M. O. (2014). Macroeconomic risk and hedge fund returns, Journal of Financial Economics, 114(1), 1-19.
- Barr, D., O. Bush, and A. Pienkowski (2014). “GDP-linked bonds and sovereign default”. Bank of England Working Paper 484.
- Barro, R. J. (1979). On the determination of the public debt. Journal of political Economy, 87 (5, Part 1), 940-971.
- Blanchard, O. (1990). Suggestions for a New Set of Fiscal Indicators. OECD Working Paper n. 79.
- Bohn, H. (1990). Tax smoothing with financial instruments. The American Economic Review, 1217-1230.
- Borensztein, E. and P. Mauro (2002). “Reviving the case for GDP-indexed bonds”. IMF Policy discussion paper 02/10.
- Borensztein, E., Jeanne, O., & Sandri, D. (2013). Macro-hedging for commodity exporters. Journal of Development Economics, 101, 105-116.
- Brender, A., & Drazen, A. (2005). Political budget cycles in new versus established democracies. Journal of monetary Economics, 52(7), 1271-1295.
- Caballero, R. J., & Panageas, S. (2008). Hedging sudden stops and precautionary contractions. Journal of Development Economics, 85(1-2), 28-57.
- Caballero, R. J., & Panageas, S. (2004). Contingent reserves management: an applied framework, NBER Working Paper No. w10786.
- Callen, M., Imbs, J. and Mauro P. (2015), Pooling risk across countries. Journal of International Economics, 96 (2015) 88–99.
- Canova, F., & Ciccarelli, M. (2013). Panel Vector Autoregressive Models: A Survey, in VAR Models in Macroeconomics–New Developments and Applications: Essays in Honor of Christopher A. Sims (pp. 205-246). Emerald Group Publishing Limited.
- Cerovic, S., K. Gerling, A. Hodge, and P. Medas (2018), Predicting fiscal crises, IMF Working Paper 18/181.
- Crucini, M. (1998), On international and national dimensions of risk sharing, The Review of Economics and Statistics, 81(1): 73–84.
- Cuadra, Gabriel and Horacio Sapriza (2008). “Sovereign default, interest rates and political uncertainty in emerging markets”. Journal of International Economics 76.1, pp. 78–88.
- Durdu, Ceyhun Bora (2009). “Quantitative implications of indexed bonds in small open economies”. Journal of Economic Dynamics and control 33.4, pp. 883–902.
- Em-dat, C. R. E. D. (2010). The OFDA/CRED international disaster database. Université catholique de Louvain.
- Ghesquiere, F. and Mahul, O. (2007), Sovereign Natural Disaster Insurance for Developing Countries: A Paradigm Shift in Catastrophe Risk Financing, World Bank Policy Research Working Paper 4345.
- Gorodnichenko, Y., & Lee, B. (2017). A Note on Variance Decomposition with Local Projections, NBER Working Paper No. w23998.
- Gruss, B. (2014). After the boom–commodity prices and economic growth in Latin America and the Caribbean, IMF Working Paper 14/154.
- Hadzi-Vaskov, M., & Ricci, M. L. A. (2017). Does Gross or Net Debt Matter More for Emerging Market Spreads? IMF Working Paper 16/246.
- Hatchondo, J. C. and L. Martinez (2012). “On the Benefits of GDP-Indexed Government Debt: Lessons from a Model of Sovereign Defaults”. Economic Quarterly 98.2, pp. 139–157.
- Henzel, S. R., & Rengel, M. (2017). Dimensions of macroeconomic uncertainty: A common factor analysis. Economic Inquiry, 55(2), 843-877.
- Jaimovich, D. & Panizza, U. (2007). Procyclicality or Reverse Causality? Inter-American Development Bank Working paper n. 599.
- Jorda, O., Schularick, M., Taylor, A. M. (2018). The Effects of Quasi-Random Monetary Experiments, FRB San Francisco Working Paper 2017-02.
- Jordà, Ò. (2005). Estimation and inference of impulse responses by local projections. American Economic Review, 95(1), 161-182.
- Khandelwal, P., & Roitman, A. (2013). The economics of political transitions: implications for the Arab Spring, IMF Working Paper No. 13-69.
- Klemm, M. A. (2014). Fiscal policy in Latin America over the cycle. IMF Working Paper No. 14-59.
- Laeven, M. L. & Valencia, F. (2008). Systemic banking crises: a new database. IMF Working Paper No. 8-224.
- Lane, Philip R. et Milesi-Ferretti, Gian Maria. The external wealth of nations mark II: Revised and extended estimates of foreign assets and liabilities, 1970–2004. Journal of international Economics, 2007, vol. 73, no 2, p. 223-250.
- Lee, M., Bai, J., Zhang, F., & Bachmann, R. (2014). The Welfare Costs of Fiscal Uncertainty: a Quantitative Evaluation. 2014 Meeting Papers (No. 744). Society for Economic Dynamics.
- Lusinyan, L., Cebotari, A., Velloso, R., Davis, J. M., Mati, A., Petrie, M., & Mauro, P. (2009). Fiscal Risks; Sources, Disclosure, and Management. IMF Working Paper No. 09/01.
- Ma, C., & Valencia, M. F. (2018). Welfare Gains from Market Insurance: The Case of Mexican Oil Price Risk, IMF Working Paper 18/35.
- Marcet, A., & Scott, A. (2009). Debt and deficit fluctuations and the structure of bond markets. Journal of Economic Theory, 144(2), 473-501.
- McHugh, M. J., Petrova, I., & Baldacci, M. E. (2011). Measuring fiscal vulnerability and fiscal stress: a proposed set of indicators. IMF Working Paper No. 11-94.
- Mendoza, E. G., & Oviedo, P. M. (2006). Fiscal policy and macroeconomic uncertainty in developing countries: The tale of the tormented insurer. NBER Working Paper No. w12586.
- Noy, I., & Nualsri, A. (2011). Fiscal storms: public spending and revenues in the aftermath of natural disasters. Environment and Development Economics, 16(1), 113-128.
- Pedroni, P. (2013). Structural Panel VARs. Econometrics, 1(2), 180-206.
- Plagborg-Møller, M., & Wolf, C. K. (2017). Instrumental Variable Identification of Dynamic Variance Decompositions, manuscript.
- Quinn, D. P., & Toyoda, A. M. (2008). Does capital account liberalization lead to growth? The Review of Financial Studies, 21(3), 1403-1449.
- Raddatz, C. (2008). External Shocks and Macroeconomic Volatility in Latin America. World Bank Policy Research Working Paper 4345.
- Ricci, M. L. A., Chamon, M. M., & Zhang, M. Y. S. (2011). Country Insurance Using Financial Instruments. IMF Working Paper No. 11-169.
- Sahay, M. R., Arora, M. V. B., Arvanitis, M. A. V., Faruqee, M. H., N'Diaye, M. P., & Griffoli, M. T. M. (2014). Emerging market volatility: Lessons from the taper tantrum. IMF Working Paper No. 14-19.
- Sandleris, G., H. Saprizza, and G. Taddei (2011). Indexed Sovereign Debt: An Applied Framework. The Colegio Alberto Notebooks Working Paper 104.
- Shambaugh, J. C. (2004). The effect of fixed exchange rates on monetary policy. The Quarterly Journal of Economics, 119(1), 301-352.
- Stock, J. H., & Watson, M. W. (2018). Identification and Estimation of Dynamic Causal Effects in Macroeconomics Using External Instruments. The Economic Journal, 128(610), 917-948.
- Tujula, M., & Wolswijk, G. (2004). What determines fiscal balances? An empirical investigation in determinants of changes in OECD budget balances, ECB Working paper series N. 422.

### Key data statistics (summary tables — All countries, Selected groups)
- All countries (Obs. correspond to sample sizes in table rows):
  - Overall fiscal balance, % of GDP: Mean -3.0; Median -2.6; 25th percentile -5.1; 75th percentile 0.0; Std. dev. 13.8; Obs. 6840
  - Overall government revenue, % of GDP: Mean 28.7; Median 25.8; 25th percentile 18.5; 75th percentile 36.9; Std. dev. 14.7; Obs. 6856
  - Overall government expenditure, % of GDP: Mean 31.7; Median 29.0; 25th percentile 20.8; 75th percentile 39.7; Std. dev. 18.8; Obs. 6851
  - Public debt, % of GDP: Mean 56.9; Median 44.5; 25th percentile 26.4; 75th percentile 70.2; Std. dev. 54.5; Obs. 6287
  - Change in debt to GDP ratio: Mean -0.0; Median 0.3; 25th percentile -2.8; 75th percentile 3.7; Std. dev. 26.7; Obs. 6051
  - Annual real GDP growth, percent: Mean 3.8; Median 3.8; 25th percentile 1.5; 75th percentile 6.3; Std. dev. 5.9; Obs. 7817
  - Annual inflation, percent: Mean 10.7; Median 5.7; 25th percentile 2.3; 75th percentile 11.6; Std. dev. 18.5; Obs. 7694
  - Commodity export price change, percent: Mean 0.7; Median -0.1; 25th percentile -8.7; 75th percentile 10.0; Std. dev. 16.0; Obs. 6392
  - Commodity import price change, percent: Mean 0.5; Median 0.2; 25th percentile -7.4; 75th percentile 10.3; Std. dev. 15.1; Obs. 6392
  - Change in unemployment rate: Mean -0.1; Median -0.1; 25th percentile -0.6; 75th percentile 0.4; Std. dev. 1.4; Obs. 3515
  - Transition year dummy, from PolityIV: Mean 0.0; Median 0.0; 25th percentile 0.0; 75th percentile 0.0; Std. dev. 0.2; Obs. 9302
  - Election year dummy: Mean 0.3; Median 0.0; 25th percentile 0.0; 75th percentile 1.0; Std. dev. 0.4; Obs. 4168
  - Democracy index, from PolityIV: Median 4.0; 25th percentile 0.0; 75th percentile 9.0; Std. dev. 17.0; Obs. 6802 (note: mean is uninformative due to special coding)
  - Log population, deviation from quad. Trend: Mean -0.0; Median -0.0; 25th percentile -0.8; 75th percentile 0.7; Std. dev. 3.1; Obs. 7892
  - Oil trade balance, % of GDP: Mean 1.0; Median -2.1; 25th percentile -4.4; 75th percentile -0.0; Std. dev. 12.9; Obs. 6071
  - Average of imports and exports, % of GDP: Mean 39.0; Median 34.2; 25th percentile 22.7; 75th percentile 49.4; Std. dev. 25.1; Obs. 6808
  - Absolute value of NFA to GDP: Mean 69.3; Median 39.6; 25th percentile 17.9; 75th percentile 77.1; Std. dev. 128.3; Obs. 7100

- All Countries (alternate summary block):
  - Overall fiscal balance, % of GDP: Mean -1.7; Median -2.1; 25th percentile -4.3; 75th percentile 0.7; Std. dev. 4.9; Obs. 1479
  - Overall government revenue, % of GDP: Mean 37.5; Median 38.1; 25th percentile 31.1; 75th percentile 44.8; Std. dev. 10.5; Obs. 1479
  - Overall government expenditure, % of GDP: Mean 39.3; Median 40.7; 25th percentile 33.4; 75th percentile 47.6; Std. dev. 11.6; Obs. 1489
  - Public debt, % of GDP, HPDD: Mean 53.1; Median 46.6; 25th percentile 28.1; 75th percentile 68.7; Std. dev. 36.2; Obs. 1460
  - Change in debt to GDP ratio: Mean 1.2; Median 0.6; 25th percentile -1.4; 75th percentile 3.4; Std. dev. 5.3; Obs. 1332
  - Annual real GDP growth, percent: Mean 3.3; Median 3.0; 25th percentile 1.5; 75th percentile 4.9; Std. dev. 3.8; Obs. 1603
  - Annual inflation, percent: Mean 5.5; Median 3.0; 25th percentile 1.6; 75th percentile 6.4; Std. dev. 9.2; Obs. 1598
  - Commodity export price change: Mean 0.5; Median 0.3; 25th percentile -6.8; 75th percentile 8.5; Std. dev. 13.2; Obs. 1293
  - Commodity import price change: Mean 0.5; Median 0.6; 25th percentile -6.4; 75th percentile 9.6; Std. dev. 15.1; Obs. 1293
  - Change in unemployment rate, WDI: Mean -0.0; Median -0.1; 25th percentile -0.7; 75th percentile 0.5; Std. dev. 1.4; Obs. 962
  - Transition year dummy, from PolityIV: Mean 0.0; Median 0.0; 25th percentile 0.0; 75th percentile 0.0; Std. dev. 0.1; Obs. 1869
  - Election year dummy: Mean 0.4; Median 0.0; 25th percentile 0.0; 75th percentile 1.0; Std. dev. 0.5; Obs. 828
  - Democracy index, from PolityIV: Mean 8.5; Median 10.0; 25th percentile 9.0; 75th percentile 10.0; Std. dev. 7.2; Obs. 1398
  - Log population, deviation from quadratic trend: Mean 0.0; Median 0.0; 25th percentile -0.7; 75th percentile 0.7; Std. dev. 1.5; Obs. 1745
  - Oil trade balance, % of GDP: Mean -1.9; Median -1.8; 25th percentile -3.3; 75th percentile -0.9; Std. dev. 4.1; Obs. 1383
  - Average of imports and exports, % of GDP: Mean 47.6; Median 35.2; 25th percentile 27.0; 75th percentile 53.5; Std. dev. 37.5; Obs. 1365
  - Absolute value of NFA to GDP: Obs. 1499 (value entries truncated in source)

- Emerging and Developing Countries only:
  - Overall fiscal balance, % of GDP: Mean -3.3; Median -2.7; 25th percentile -5.4; 75th percentile -0.2; Std. dev. 15.4; Obs. 5325
  - Overall government revenue, % of GDP: Mean 26.1; Median 23.3; 25th percentile 17.0; 75th percentile 31.7; Std. dev. 14.3; Obs. 5341
  - Overall government expenditure, % of GDP: Mean 29.4; Median 26.3; 25th percentile 19.8; 75th percentile 34.6; Std. dev. 19.6; Obs. 5326
  - Public debt, % of GDP, HPDD: Mean 58.3; Median 44.1; 25th percentile 26.1; 75th percentile 71.8; Std. dev. 58.7; Obs. 4889
  - Change in debt to GDP ratio: Mean -0.4; Median 0.2; 25th percentile -3.4; 75th percentile 3.9; Std. dev. 30.2; Obs. 4704
  - Annual real GDP growth, percent: Mean 3.9; Median 4.1; 25th percentile 1.5; 75th percentile 6.5; Std. dev. 6.3; Obs. 6145
  - Annual inflation, percent: Mean 12.2; Median 6.7; 25th percentile 2.9; 75th percentile 13.0; Std. dev. 20.0; Obs. 6062
  - Commodity export price change, percent: Mean 0.7; Median -0.3; 25th percentile -9.3; 75th percentile 10.4; Std. dev. 16.6; Obs. 5099
  - Commodity import price change, percent: Mean 0.5; Median 0.1; 25th percentile -7.7; 75th percentile 10.4; Std. dev. 15.1; Obs. 5099
  - Change in unemployment rate: Mean -0.1; Median -0.1; 25th percentile -0.7; 75th percentile 0.4; Std. dev. 1.6; Obs. 1990
  - Transition year dummy, from PolityIV: Mean 0.0; Median 0.0; 25th percentile 0.0; 75th percentile 0.0; Std. dev. 0.2; Obs. 7289
  - Election year dummy: Mean 0.3; Median 0.0; 25th percentile 0.0; 75th percentile 1.0; Std. dev. 0.4; Obs. 3340
  - Democracy index, from PolityIV: Median 1.0; 25th percentile 0.0; 75th percentile 7.0; Std. dev. 18.2; Obs. 5404 (note: mean is uninformative)
  - Log population, deviation from quadratic trend: Mean -0.0; Median -0.0; 25th percentile -0.8; 75th percentile 0.8; Std. dev. 3.1; Obs. 6214
  - Oil trade balance, % of GDP: Mean 1.9; Median -2.2; 25th percentile -4.8; 75th percentile 0.2; Std. dev. 14.4; Obs. 4688
  - Average of imports and exports, % of GDP: Mean 36.7; Median 33.7; 25th percentile 21.9; 75th percentile 48.4; Std. dev. 20.2; Obs. 5412
  - Absolute value of NFA to GDP: Mean 76.5; Median 43.4; 25th percentile 21.5; 75th percentile 84.4; Std. dev. 140.5; Obs. 5601

- LAC countries:
  - Overall fiscal balance, % of GDP: Mean -3.0; Median -2.7; 25th percentile -4.6; 75th percentile -0.5; Std. dev. 4.9; Obs. 1266
  - Overall government revenue, % of GDP: Mean 22.3; Median 22.9; 25th percentile 17.5; 75th percentile 27.1; Std. dev. 7.7; Obs. 1273
  - Overall government expenditure, % of GDP: Mean 25.3; Median 24.9; 25th percentile 19.5; 75th percentile 30.2; Std. dev. 9.0; Obs. 1266
  - Public debt, % of GDP, HPDD: Mean 60.5; Median 42.7; 25th percentile 27.3; 75th percentile 65.8; Std. dev. 74.5; Obs. 1329
  - Change in debt to GDP ratio: Mean -0.8; Median 0.4; 25th percentile -3.1; 75th percentile 4.0; Std. dev. 51.7; Obs. 1281
  - Annual real GDP growth, percent: Mean 3.4; Median 3.7; 25th percentile 1.1; 75th percentile 5.9; Std. dev. 4.3; Obs. 1475
  - Annual inflation, percent: Mean 13.7; Median 7.1; 25th percentile 3.3; 75th percentile 15.8; Std. dev. 20.1; Obs. 1429
  - Commodity export price change, percent: Mean 0.3; Median -0.4; 25th percentile -9.2; 75th percentile 9.9; Std. dev. 15.6; Obs. 1159
  - Commodity import price change, percent: Mean 0.8; Median 0.6; 25th percentile -7.3; 75th percentile 11.4; Std. dev. 15.7; Obs. 1159
  - Change in unemployment rate: Mean -0.1; Median -0.1; 25th percentile -0.9; 75th percentile 0.6; Std. dev. 1.5; Obs. 731
  - Transition year dummy, from PolityIV: Mean 0.0; Median 0.0; 25th percentile 0.0; 75th percentile 0.0; Std. dev. 0.1; Obs. 1507
  - Election year dummy: Mean 0.3; Median 0.0; 25th percentile 0.0; 75th percentile 1.0; Std. dev. 0.4; Obs. 734
  - Democracy index, from PolityIV: Median 7.0; 25th percentile 2.0; 75th percentile 9.0; Std. dev. 14.8; Obs. 1048 (note: mean uninformative)
  - Log population, deviation from quadratic trend: Mean -0.0; Median -0.1; 25th percentile -0.6; 75th percentile 0.5; Std. dev. 1.5; Obs. 1438
  - Oil trade balance, % of GDP: Mean -1.3; Median -2.1; 25th percentile -5.1; 75th percentile 0.3; Std. dev. 7.8; Obs. 1056
  - Average of imports and exports, % of GDP: Mean 34.6; Median 31.4; 25th percentile 21.1; 75th percentile 46.4; Std. dev. 17.8; Obs. 1122
  - Absolute value of NFA to GDP: Mean 59.1; Median 39.6; 25th percentile 19.4; 75th percentile 80.5; Std. dev. 62.3; Obs. 1349

### Data sources and construction (key points)
- Fiscal variables:
  - Yearly fiscal data primarily from the IMF World Economic Outlook, complemented with the IMF Government Finance Statistics database for pre-1995 data.
  - Overall government balance to GDP ratio, government revenues and expenditure (relative to GDP) are focal variables.
  - Gross government debt to GDP ratio from the Historical Public Debt Database of Abbas et al. (2011); note: correlation between HPDD and WEO debt measures is 0.95 in the subsample where both are available.
  - International reserves (excluding gold) to GDP ratios computed from the updated External Wealth of Nations database (Lane and Milesi-Ferretti, 2007).

- Trade and external variables:
  - Terms-of-trade and trade openness (average of exports and imports to GDP) from the World Economic Outlook.
  - Foreign assets and liabilities from the External Wealth of Nations database.
  - Exchange rate pegs classified using Shambaugh (2004) central binary dummy for pegged currency.
  - Export and import commodity price shocks from Gruss (2014) trade-weighted commodity price indices, computed separately for exports and imports, averaged annually with time-varying trade weights.

- Global financial variables:
  - Long-term US 10-year interest rates from World Economic Outlook.
  - US effective fed funds rate from St-Louis Federal Reserve FRED database.
  - VIX volatility index from the Chicago Board of Exchanges.
  - Nominal exchange rate data from the World Economic Outlook.

- Political and other domestic variables:
  - Polity IV database (World Bank) for democratic regime measures and transition periods (transition periods occur in around 3% of country-years observations).
  - Election year dummy based on a national-level election occurrence database (acknowledgement of Karim Youssef’s election year database based on electionguide.org).
  - Natural disasters from EM-DAT (CRED, Em-Dat, 2010); occurrence coded as a dummy if a non-human-made disaster is reported in a country-year. Only categories used: animal accident, drought, earthquake, epidemic, extreme temperature, flood, impact, insect infestation, landslide, mass movement, storm, volcanic activity, wildfire.
  - Military expenditure from SIPRI database (available after 1988 for a subset of countries).
  - Domestic systemic banking crises from an updated Laeven and Valencia (2008) “Systemic banking crises” database (using crisis start year dummy).
  - Population from World Bank WDI.
  - Domestic credit to private sector to GDP, stock market capitalization to GDP, and yearly stock market volatility from World Bank Global Financial Development database and FinStats database.

### Annex findings for LAC countries (selected quantitative results)
- Terms-of-trade (commodity price) shocks — LAC exposure:
  - A one standard deviation increase in trade-weighted imported commodity prices (within-country variation among Latin American countries) decreases the fiscal balance for LAC importers by about 0.7% of GDP after a year, and remains at such level for the first three years; other EMDC importers reach that level after 4 years.
  - LAC importers experience a substantial drop in GDP close to one percentage point (≈ 1 percentage point) after an imported commodity price shock; other emerging commodity importers do not show this drop.
  - A one standard deviation increase in trade-weighted export commodity prices increases the fiscal balance of LAC exporters by about ½ a percentage point of GDP after a year, and raises GDP by a similar amount after 2-3 years; impact similar to other commodity exporting economies.
- Oil trade balance shocks — LAC:
  - A one percentage point of GDP improvement in the oil trade balance is associated with a 0.3 percentage point increase in the fiscal balance and a long-term decrease in the debt to GDP ratio of around 2 percentage points.
- Domestic growth shock — LAC:
  - A one percentage point domestic growth increase produces qualitatively similar effects on fiscal balance, revenue, expenditure, and public debt ratios as in other EMDC, but quantitatively:
    - The fiscal balance response in LAC is somewhat smaller but more persistent.
    - The muted fiscal balance effect in early years stems from both a more muted positive revenue-to-GDP response and a more muted negative expenditure-to-GDP response.
    - Over the medium term, the fiscal balance effect remains positive in LAC (unlike other EMDC) because there is an absence of a pick-up in the expenditure ratio.
    - Net result: impact on gross debt to GDP ratios is similar in LAC to other emerging and developing economies due to offsetting larger persistence of GDP impact and slightly smaller fiscal-balance effect.

*Source: REFERENCES; ANNEX I–III, wpiea2020220-print-pdf - REFERENCES*

### ANNEX IV. PANEL REGRESSIONS FOR DIFFERENT COUNTRY GROUPINGS

### ANNEX IV. PANEL REGRESSIONS FOR DIFFERENT COUNTRY GROUPINGS

### Advanced, main specification — key regression coefficients and model fit
- L.Overall fiscal balance, % of GDP: 0.717 **** [0.0337]
- Annual world real GDP growth, percent:
  - coefficient in first column: 0.697 **** [0.0760]
  - other column coefficients reported include: -0.00347 [0.0555]; -0.690 **** [0.0805]; -1.541 **** [0.160]; -1.232 *** [0.404]; -0.310 ** [0.124]; 1.046 **** [0.117]; -0.0677 [0.215]
- Standardized VIX: -0.239 *** [0.0851]; additional VIX coefficients: -0.243 **** [0.0503]; 0.0257 [0.0697]; 0.0941 [0.239]; 0.214 [0.391]; 0.162 [0.162]; -0.261 *** [0.0727]; -0.00556 [0.176]
- Commodity export price change, normalized: -0.125 [0.222]; other commodity-related coefficients reported across columns (e.g., 0.227 ** [0.110], 0.537 [0.365])
- Commodity import price change, normalized: 0.0280 [0.159]; other reported coefficients include 0.780 * [0.397], -0.307 * [0.160], -0.246 ** [0.114]
- US government 10-year constant maturity bond yield, percent: 0.00463 [0.0396]; in other columns coefficients include -0.419 **** [0.110], -0.441 ** [0.205], 0.300 **** [0.0609]
- Transition year dummy, from PolityIV: 0.697 **** [0.173]; other transition-year related coefficients include -0.448 ** [0.165], -1.365 **** [0.240], -1.721 *** [0.516], 4.042 **** [0.236], 3.899 **** [0.702]
- Systemic banking crisis from Laeven-Valencia: -0.631 [0.689]; additional coefficients across specifications reported (e.g., 2.921 ** [1.350], -0.976 ** [0.441], -1.508 ** [0.659])
- Log population, deviation from quadratic trend: -0.211 * [0.106]; other reported coefficients include 0.0645 * [0.0366], 0.260 *** [0.0909], 0.838 ** [0.395], 0.951 ** [0.460], -0.234 ** [0.101]
- Damages from natural disasters, USD bn: -0.220 [0.182]; other coefficients include 0.239 ** [0.0990], 0.784 *** [0.283], 1.223 *** [0.336], -0.222 **** [0.0545], 0.783 *** [0.261]
- Interaction Damages x log population: 0.0113 [0.00956]; other interaction coefficients include -0.0123 ** [0.00523], -0.0395 ** [0.0148], -0.0634 *** [0.0177], 0.0114 **** [0.00288], -0.0411 *** [0.0137]
- Lagged regressors (selected):
  - L.Overall government revenue, % of GDP: 0.851 **** [0.0303]
  - L.Overall government expenditure, % of GDP: 0.790 **** [0.0291]
  - L.Public debt, % of GDP, HPDD: 0.973 **** [0.0141]
  - L.General govt net debt, % of GDP: 0.975 **** [0.0171]
  - L.International reserves, % of USD GDP, EWN: 0.944 **** [0.0280]
  - L.Annual real GDP growth, percent: 0.336 **** [0.0398]
  - L.Change in the nominal effective exchange rate, percent: 0.319 **** [0.0405]
- Constant terms across columns: -3.051 **** [0.377]; 5.952 **** [1.232]; 11.27 **** [1.251]; 10.32 **** [1.148]; 7.948 **** [1.852]; 2.605 **** [0.565]; -3.385 **** [0.521]; 0.126 [0.791]
- Model fit and sample size:
  - Adj. R-Square: 0.596 0.800 0.732 0.946 0.938 0.852 0.424 0.163
  - Observations: 1000 1000 1000 945 660 1000 994 1002
  - Countries: 35 35 35 34 31 35 35 35

### Emerging, main specification — key regression coefficients and model fit
- L.Overall fiscal balance, % of GDP: 0.633 **** [0.0884]
- Annual world real GDP growth, percent:
  - first-column coefficient: 0.322 *** [0.115]; other coefficients reported include 0.233 * [0.123], -0.153 ** [0.0693], -0.574 * [0.322], -0.504 [0.499], -0.303 ** [0.129], 0.563 **** [0.0949], -0.444 ** [0.197]
- Standardized VIX: -0.255 *** [0.0760]; other VIX coefficients include -0.158 ** [0.0611], 0.0924 [0.0675], 0.895 **** [0.262], -0.207 ** [0.104], -0.137 [0.0851], -0.534 *** [0.185]
- Commodity export price change, normalized: -0.0404 [0.145]; notable interaction:
  - Commodity export price change, normalized x exporter: 1.430 **** [0.327]; other coefficients 0.901 **** [0.223], -0.573 ** [0.250], -1.339 [1.037], -3.169 ** [1.558]
- Commodity import price change, normalized: 0.662 *** [0.211]; commodity import price change x importer: -0.632 ** [0.245]
- US government 10-year constant maturity bond yield, percent: 0.0258 [0.0697]; other coefficients include -0.172 *** [0.0575], -0.179 *** [0.0566], 0.426 ** [0.166], -0.445 *** [0.158]
- Transition year dummy, from PolityIV: -4.544 [2.929]; other transition-related coefficients include -1.973 *** [0.691], 2.229 [2.370], 11.89 * [6.311], 27.73 ** [13.73], 7.650 [6.679], -5.414 * [3.241], -2.404 [1.486]
- End of transition: into more democracy: 2.455 [2.754]; End of transition: into less democracy: 0.0856 [4.296]; several related coefficients reported (some statistically significant)
- Systemic banking crisis from Laeven-Valencia: -0.662 [0.475]; other columns: 7.102 ** [3.201], 14.25 * [8.215], -1.051 ** [0.509], -2.111 **** [0.622]
- Log population, deviation from quadratic trend: -0.142 *** [0.0522]; other coefficients include 0.106 ** [0.0519], -0.0798 *** [0.0258]
- Damages from natural disasters, USD bn: -0.353 [0.311]; Damages x log population: 0.0172 [0.0147]; other coefficients and interactions reported
- Lagged regressors (selected):
  - L.Overall government revenue, % of GDP: 0.687 **** [0.0655]
  - L.Overall government expenditure, % of GDP: 0.838 **** [0.0617]
  - L.Public debt, % of GDP, HPDD: 0.904 **** [0.0224]
  - L.General govt net debt, % of GDP: 0.935 **** [0.0120]
  - L.International reserves, % of USD GDP, EWN: 0.966 **** [0.115]
  - L.Annual real GDP growth, percent: 0.188 **** [0.0540]
  - L.Change in the nominal effective exchange rate, percent: 0.344 **** [0.0335]
- Constant terms across columns: -2.144 **** [0.575]; 8.135 **** [1.511]; 6.082 **** [1.672]; 4.788 ** [1.960]; 3.464 [2.379]; 2.794 [2.629]; 1.350 *** [0.441]; 2.068 * [1.077]
- Model fit and sample size:
  - Adj. R-Square: 0.361 0.481 0.596 0.866 0.902 0.752 0.0988 0.176
  - Observations: 3791 3791 3792 3369 945 3837 3872 3912
  - Countries: 140 140 140 130 56 140 140 139

### LAC (Latin America and the Caribbean), main specification — key regression coefficients and model fit
- L.Overall fiscal balance, % of GDP: 0.687 **** [0.0276]
- Annual world real GDP growth, percent:
  - first-column coefficient: 0.217 *** [0.0785]; other coefficients include 0.202 * [0.109], -0.0169 [0.118], -0.172 [0.626], -0.676 [0.424], -0.316 ** [0.120], 0.613 **** [0.127], -0.892 ** [0.431]
- Standardized VIX: -0.246 ** [0.0908]; other VIX coefficients include -0.102 [0.0875], 0.152 [0.0921], 1.979 **** [0.421], 0.726 * [0.365], -0.400 *** [0.118], -0.978 ** [0.378]
- Commodity export price change, normalized: 0.238 [0.189]; interaction commodity export x exporter: 0.197 [0.513]; other commodity coefficients across columns reported (including exporter interactions statistically significant)
- Commodity import price change, normalized: 0.373 [0.277]; commodity import price change x importer: -0.424 [0.292]; various other coefficients reported (e.g., -2.431 *** [0.759])
- US government 10-year constant maturity bond yield, percent: 0.127 ** [0.0548]; other coefficients include -0.166 ** [0.0617], -0.318 **** [0.0580], -0.812 ** [0.279], -0.242 **** [0.0550], 0.210 *** [0.0730]
- Transition year dummy, from PolityIV: -0.240 [0.188]; other transition-related coefficients include 0.731 *** [0.206], 1.004 **** [0.206], 1.378 *** [0.388], 5.927 **** [1.087]
- End of transition: into more democracy and into less democracy coefficients reported, some large and statistically significant (e.g., End of transition into less democracy: 25.50 **** [2.606] in one column)
- Systemic banking crisis from Laeven-Valencia: -1.861 ** [0.849]; other columns include 31.54 *** [8.602] (gross debt column), -3.614 ** [1.392]
- Log population, deviation from quadratic trend: -0.0971 [0.0970]; other coefficients reported including 0.0924 [0.0996]
- Damages from natural disasters, USD bn: 0.625 [0.733]; Damages x log population: -0.0349 [0.0430]; specific coefficients show significant impacts in certain columns (e.g., -11.14 **** [2.445])
- Lagged regressors (selected):
  - L.Overall government revenue, % of GDP: 0.750 **** [0.0561]
  - L.Overall government expenditure, % of GDP: 0.710 **** [0.0450]
  - L.Public debt, % of GDP, HPDD: 0.951 **** [0.0115]
  - L.General govt net debt, % of GDP: 0.964 **** [0.0254]
  - L.International reserves, % of USD GDP, EWN: 0.888 **** [0.0238]
  - L.Annual real GDP growth, percent: 0.249 **** [0.0454]
  - L.Change in the nominal effective exchange rate, percent: 0.413 **** [0.0637]
- Constant terms across columns: -2.155 **** [0.457]; 6.015 **** [1.392]; 9.166 **** [1.049]; 4.039 [3.150]; 5.987 *** [1.679]; 4.072 **** [0.613]; -0.709 [0.689]; 2.222 [3.184]
- Model fit and sample size:
  - Adj. R-Square: 0.540 0.634 0.629 0.932 0.870 0.839 0.168 0.234
  - Observations: 909 909 909 894 245 934 934 934
  - Countries: 32 32 32 32 14 32 32 32

### Standard errors and significance notation (applies to regressions above)
- Standard errors are reported in brackets.
- Significance levels:
  - * p < 0.10
  - ** p < 0.05
  - *** p < 0.01
  - **** p < 0.001

---

### ANNEX VII. VARIANCE DECOMPOSITION CHARTS — methodology and principal findings

### Methodology for forecast error variance decomposition (local projections)
- Approach implemented follows Gorodnichenko and Lee (2017) adapted for local projection methods.
- Two-step procedure:
  1. Compute forecast errors of outcome variables at each projection horizon using the local projection method.
  2. Compare the (t−1) forecast error variance with the variance of the best linear predictor of the time t+ h outcome using values of the variable of interest between time of impact and projection horizon.
- The share sh of forecast error variance at horizon h accounted for by variable D is estimated from the ratio of the variance of the predicted contribution of D to the variance of the h-steps ahead forecast error f_{i+h|i-1}.
- Variation in the treatment variable σ^2_D is taken to be only the within-country variation (V̂ar(D_{i,t} − D̄_i)), and similarly for forecast-error variation, with residuals not including country fixed effects.
- Main shocks analyzed: world growth, world growth interacted with openness, change in export prices, change in import prices, election year, political transitions, change in the VIX, change in US 10-year yields.

### Principal empirical findings from the variance decomposition exercise
- Relative explanatory power by country group:
  - Most shocks account for a smaller share of variation of fiscal balances in advanced economies than in emerging markets and developing countries.
- Time profile of shock importance:
  - The explanatory role of most shocks tends to grow over time (longer horizons), except global growth shock, whose role tends to decline after one year.
- Most important shocks identified:
  - The VIX (especially for countries with high financial openness).
  - Commodity prices.
  - Global growth.
- Decomposition between fiscal sub-components:
  - Expenditure shocks are a much more important predictor of short-term fluctuations in the fiscal balance ratio than revenue shocks.
  - Expenditure plays a relatively larger role in explaining forecast errors in emerging economies and Latin American countries (relative to revenue shocks) than in advanced economies.
- Interpretation note:
  - Percentages from the decomposition may not sum to one because of covariance between components; reported percentages are interpretable as the expected reduction in forecast error, at various horizons, if given the sequence of unexpected shocks to one (but not both) sub-components (expenditure or revenue).

*Source: ANNEX IV and ANNEX VII of the provided IMF content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020220-print-pdf.pdf_