## 1. Fiscal Aggregates

## Source details

**Canonical URL:** [1. Fiscal Aggregates](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09160.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09160.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09160.pdf.json)

---

### Introduction and context
- Study focus: fiscal policy effectiveness during systemic banking crisis episodes (Laeven and Valencia, 2008) using a database of over 100 banking crisis episodes between 1980 and 2008.
- Context: financial crisis beginning in 2007 triggered a global recession in 2009; constrained monetary policy elevated fiscal policy as principal stimulus tool.
- Fiscal responses considered: discretionary budget measures and automatic stabilizers; effectiveness depends on initial fiscal conditions and composition of fiscal measures.

### Research questions
- What is the effectiveness of fiscal policy in shortening the duration of systemic banking crisis episodes and strengthening economic growth in the medium term?
- Does the composition of the fiscal policy response matter for crisis duration and post-shock growth performance?

### Data and stylized facts
- Panel: 182 countries, 1980–2008; identify 118 episodes of banking crises in 99 countries (Laeven and Valencia, 2008 criteria); complemented with WEO, GFS, and GFD data.
- Crisis end definition: two consecutive years of real GDP growth above ½ percentage points per year (robustness: stock market‑based duration).
- Duration statistics:
  - Average crisis duration: 2½ years.
  - 85 percent of episodes last between one to four years.
  - Only one episode lasted eight years.
- Peak-to-trough and period changes:
  - Peak-to-trough: average GDP growth rate fell by more than 5 percentage points during the crisis.
  - Peak-to-trough: general government debt increased by 39 percentage points of GDP.
  - Peak-to-trough: budget deficits increased by 6.9 percentage points of GDP.
  - Pre-crisis to crisis: real GDP growth fell from 0.4 percent (year prior) to an average -2.2 percent during the crisis; in the two-year period following the crisis annual GDP growth rebounded to 4.9 percent per year.
- Two-year changes of fiscal aggregates (all figures as percent of GDP):
  - Debt: Before (t-2; t-1) -9.2; During (t) 27.1; After (t+1; t+2) -7.2
  - Budget balance: Before -0.1; During -5.9; After 1.5
  - Primary budget balance: Before 0.3; During -4.9; After 2.8
  - Total revenues: Before 0.8; During -3.7; After 4.9
  - Total expenditures: Before 0.9; During 2.3; After 2.6
- Crisis period averages interpretation:
  - Fiscal deficits increased by almost 6 percentage points (more than 2 percent of GDP per year).
  - Public debt worsened by 27 percentage points of GDP (about ⅓ of the preexisting average debt level of 78 percent of GDP).
  - Total revenues fell by more than 3½ percentage points of GDP; government expenditures rose by more than 2 percentage points of GDP.
- Revenue composition (two-year windows, as percent of GDP):
  - Taxes: Before 0.5; During -2.3; After 4.2
  - Income, profits, capital gains: Before 0.2; During -1.2; After 3.8
  - Payroll and workforce: Before 0.1; During -0.3; After 0.0
  - Goods and services: Before 0.1; During -0.5; After 0.4
  - International trade: Before 0.1; During -0.3; After 0.0
  - Social contributions: Before 0.2; During -1.2; After 0.2
  - Other revenues: Before 0.1; During -0.2; After 0.5
- Expenditure composition (two-year windows, as percent of GDP):
  - Current expenditure: Before 0.9; During 2.2; After 0.1
    - Goods and services: Before -0.1; During 0.6; After -0.5
    - Employee compensation: Before 0.1; During 0.2; After 0.1
    - Transfers: Before 0.1; During 0.6; After 0.3
    - Interest payments: Before 0.4; During 1.0; After 2.3
    - Other expenses: Before 0.4; During -0.2; After -0.1
  - Public Investment: Before 0.0; During 0.1; After 2.5
- Stylized policy interpretation:
  - Crisis fiscal outcomes reflect automatic stabilizers and discretionary measures (including measures to strengthen the financial system).
  - Public consumption rises immediately and contributes to aggregate demand; public investment often faces implementation lags.

### Main empirical findings — crisis duration
- Fiscal expansions shorten the duration of systemic banking crises; average fiscal policy response in the sample would reduce crisis length by more than two quarters.
- Key bivariate and multivariate results (selected coefficients reported verbatim where available):
  - Change in total public expenditures over crisis episode: Coefficient 0.103***; T 4.5; R-squared 0.1118
  - Change public investment (percent of total public expenditure): Coefficient 0.266***; T 6.8; R-squared 0.3118 (bivariate highlights)
- Duration regression baseline (118 episodes; OLS and Ordered Logit):
  - Budget Balance (in percent of GDP): Coefficient 0.072***; t-statistic (3.73)
  - Expansionary fiscal policy dummy: Coefficient -0.626*** (Model 2 OLS); -1.023*** (Model 4 Ord. Logit)
  - Previous credit boom: positive and significant (e.g., 0.690*** in Model 1)
  - Deposit freeze or guarantee: negative and significant (e.g., -0.522** in Model 1)
  - Number of banks closed: negative and significant (e.g., -0.168*** in Model 1)
  - Government intervention: negative and significant (e.g., -0.721*** in Model 1)
  - Adj. R-squared / Pseudo R-squared: 0.435 (Model 1) and 0.211 (Model 3)
- Budget composition effects on crisis length (equation (2); Table 6 highlights):
  - Expansionary fiscal policy: coefficients around -0.522** to -0.601** (OLS) and around -0.945** to -1.049** (Ord. Logit).
  - Public consumption (percent of total expenditures): negative and significant (e.g., -0.035*** in Model 1 OLS).
    - A 5 percentage point increase in public consumption share reduces crisis length by almost three months (interpreted from estimated coefficient magnitudes).
  - Public investment (percent of total expenditures): negative but weaker significance (e.g., -0.027*).
  - Income tax revenue (percent of total revenues): positive and significant in some specifications (e.g., 0.076***), implying reducing income-tax share helps shorten crises.
  - Goods & services tax revenue (percent of total revenues): positive and significant in some specifications (e.g., 0.119***).
  - Policy controls retain expected signs: credit booms lengthen crises; deposit guarantees, number of banks closed, and government intervention shorten crises.
- Containment and resolution policy estimates (OLS, Observation 118; verbatim highlights):
  - Credit Boom (t-1): 0.933*** 3.750.11 118
  - Deposit Freeze: -0.951*** -3.830.12 118
  - Bank Holiday: -0702** -3.130.06 118
  - Blanket Guarantee: -0.721** -3.210.07 118
  - Liquidity Support: 0.769** 3.440.08 118
  - Forbearance: 0.787** 3.510.08 118
  - Government Intervention: -0.973*** -3.910.11 118
  - Number of Bank Closures: -0.289*** -5.450.19 118
  - Fresh Private Capital: -0.321 -1.480.01 118
  - Number Foreign Sales: 0.482*** 4.810.16 118
  - Bank Agency: -0.932*** -4.240.11 118
  - Public Recapitalization: 0.803** 3.490.08 118

### Main empirical findings — post-crisis growth (five-year horizon)
- Countercyclical fiscal expansions as a general category: no significant effect on post-crisis output growth (coefficients e.g., 0.262, 0.251 not statistically significant).
- Composition matters for medium-term recovery:
  - Public investment (percent of total expenditures): positive and significant (e.g., 0.229***; T 4.94).
    - A one percent increase in the share of capital outlays in the budget raised post-crisis growth by about ⅓ of one percent per year (text summary).
  - Income tax revenue (percent of total revenues): negative and significant for growth (e.g., -0.177**).
  - Goods & services tax revenue: positive and significant for growth (e.g., 0.402***).
  - Private sector and financial variables:
    - Private Investment (percent of total investment): large positive coefficients (e.g., 7.530**).
    - Cost of financing: negative and significant (e.g., -0.121***).
    - Fresh capital injections into financial sector: positive and significant in several specifications (e.g., 1.453**).

### Initial conditions and interaction effects
- Dummies:
  - Highly Indebted: initial public debt above sample average.
  - High GDP per capita: initial GDP per capita above sample average.
- Key interaction results:
  - High initial public debt increases crisis duration (e.g., Highly Indebted (t-1) coefficient 0.798** in Table 8).
  - Expansionary fiscal policy benefits are weaker when initial conditions are poor; in high‑debt countries crises lasted almost one year longer—effect offsets benefits of expansionary policy.
  - High GDP per capita reduces crisis duration (Table 9: High GDP per capita (t-1) coefficients negative, e.g., -0.345***).
  - Interaction: expansionary fiscal policy * High GDP per Capita (t-1) negative and significant in some specifications (e.g., -0.876**), indicating stronger benefits where per capita income is higher.
  - For post-crisis growth: public investment and reductions in income taxes more effective for growth when initial conditions (debt, GDP per capita) are stronger (Tables 10 and 11).

### Robustness and endogeneity checks
- Alternative duration definition (stock-market-recovery): episodes shorter; baseline results robust (Appendix Table A5).
  - Expansionary fiscal policy: -0.709*** (-3.91); -0.756*** (-4.12); -0.740*** (-4.12); -0.751*** (-4.19) across four models.
  - Public consumption (percent of total expenditures): -0.018* (-1.89) in Model 1.
- Alternative discretionary fiscal policy indicator (Blanchard (1990)-style): baseline results consistent (Appendix Table A6).
  - Discretionary Expansionary fiscal policy: -0.370* (-1.76); -0.418* (-1.94); -0.437** (-2.08); -0.458** (-2.19).
  - Public consumption (percent of total expenditures): -0.377*** (-3.25) in Model 1.
- Endogeneity: 2SLS with instrument Liquidity Support for Expansionary fiscal policy (Appendix Table A7):
  - Expansionary fiscal policy coefficients: Model 1: -0.522**  (-2.45); Model 2: -0.572**  (-2.61); Model 3: -0.581**  (-2.74); Model 4: -0.602**  (-2.85)
  - Public consumption (percent of total expenditures): Model 1: -0.035***  (-3.12)
  - Public investment (percent of total expenditures): Model 2: -0.027*  (-1.82)
  - Income tax revenue (percent of total revenues): Model 3: 0.076***  (3.07)
  - Goods & services tax revenue (percent of total revenues): Model 4: 0.119***  (3.18)
  - Previous Credit boom: positive coefficients across models (e.g., 0.568**  (2.80) to 0.621***  (2.99))
  - Deposit freeze or guarantee: negative across models (e.g., -0.555**  (-2.76))
  - Number of banks closed and Government intervention: negative and significant across models.
  - Observations: 118; Adj. R-squared: Model 1 0.450; Model 2 0.419; Model 3 0.449; Model 4 0.452.

### Trade-offs, constraints, and caveats
- Potential conflict between short-term gains and medium-term growth depending on fiscal instrument choice.
- Limited fiscal space and public debt sustainability concerns can restrict effectiveness of fiscal expansions.
- Fiscal multipliers vary across instruments: largest for government consumption, public investment, and transfers; relatively smaller for indirect taxes.
- Results sensitive to initial fiscal conditions and macro-financial context; effectiveness depends on accompanying macroeconomic policies and stimulus design.

### Policy implications and recommendations
- Prioritize composition of fiscal stimulus during banking crises, balancing short- and medium-term objectives:
  - Short-term: use measures with timely demand impact (e.g., public consumption, consumption tax cuts) to shorten crises.
  - Medium-term: scale up public investment and favor income tax reductions over consumption tax cuts to support post-crisis growth.
- Build fiscal space and ensure public debt sustainability prior to crises to enable effective countercyclical responses.
- Address financial sector vulnerabilities concurrently with fiscal measures; crisis resolution that restructures private sector balance sheets can have lasting fiscal costs.
- Anticipate implementation lags for public investment—prepare shovel-ready projects and improve execution to realize medium-term gains.
- Strengthen fiscal institutions, reduce political risks, and improve budget execution capacity—especially in emerging markets—to reap benefits of countercyclical fiscal policy.

### Conclusion — succinct empirical messages
- Timely countercyclical fiscal responses, when accompanied by actions addressing financial sector weaknesses, shorten crisis duration.
- Expansionary fiscal policies reduced crisis duration by almost one year on average (text summary).
- Composition matters:
  - Government consumption is more effective in shortening crisis duration than public investment for immediate impact.
  - Scaling up public investment and reducing income taxes are associated with better medium-term growth.
  - Reductions in consumption taxes during crises can undermine future economic performance.
- Initial conditions (high precrisis public debt, low per capita income) limit the scope and effectiveness of countercyclical fiscal policy.
- Private sector recovery conditions (private investment share, cost of financing, fresh banking-sector capital) materially affect medium-term recovery.

*Source: _wp09160 - 1. Fiscal Aggregates (IMF Working Paper excerpt).*

### 1. Fiscal Aggregates....................................................................................................

### 1. Fiscal Aggregates

### Introduction and context
- The paper studies fiscal policy effectiveness during systemic banking crisis episodes, focusing on episodes originating in the banking sector (Laeven and Valencia, 2008).
- The financial crisis that began in 2007 triggered a global recession in 2009; constrained monetary policy raised the prominence of fiscal policy as the principal tool for stimulating recovery.
- Fiscal responses include discretionary budget measures and automatic stabilizers; their effectiveness depends on initial fiscal conditions and the composition of fiscal measures.
- The analysis uses a database of banking crises comprising over 100 banking crisis episodes that occurred in the world between 1980 and 2008.

### Research questions
- What is the effectiveness of fiscal policy in shortening the duration of systemic banking crisis episodes and strengthening economic growth in the medium term?
- Does the composition of the fiscal policy response matter, both in terms of crisis duration and post-shock growth performance?

### Main findings
- Fiscal expansions shorten the duration of systemic banking crises.
- The composition of fiscal expansion packages is key to success during crisis episodes:
  - Increasing public consumption is more effective than increasing public investment in reducing crisis duration, attributed to its timeliness in boosting aggregate demand.
  - Cutting consumption taxes is correlated with shorter crises than cutting income taxes, as tax reductions on consumption have a more widespread immediate impact on consumers.
- Countercyclical fiscal expansions, taken as a general category, have no significant effect on post-crisis output growth.
- Composition matters for medium-term recovery:
  - Increasing the share of public investment during shock periods is effective in improving medium-term output performance.
  - Government (public) consumption has no significant positive effect on medium-term output growth.
  - Cutting the share of income taxes is associated with removing distortions that hamper medium-term economic growth.
  - Reductions in consumption taxes during crises undermine future economic performance.
- These results indicate a potential trade-off between short-term gains (shortening crisis duration) and medium-term growth objectives depending on fiscal instrument choice.

### Policy implications and recommendations
- Policymakers should prioritize the composition of fiscal stimulus during banking crises, balancing:
  - Short-term objectives: use measures with timely demand impact (e.g., public consumption, consumption tax cuts) to shorten crises.
  - Medium-term objectives: scale up public investment and favor income tax reductions over consumption tax cuts to support post-crisis growth.
- Ensure fiscal space and public debt sustainability prior to crises to increase the ability to implement effective countercyclical fiscal policies.
- Address financial sector vulnerabilities concurrently with fiscal measures to avoid long-term negative impacts on public debt and growth; crisis resolution that restructures private sector balance sheets can have lasting fiscal costs.

### Trade-offs, constraints, and caveats highlighted
- There is a potential conflict between the size of countercyclical fiscal expansions during downturns and their medium-term growth implications.
- Limited fiscal space and concerns about public debt sustainability can restrict the effectiveness of fiscal expansions during crises.
- The effectiveness of fiscal policy depends on accompanying macroeconomic policies and the specific design of stimulus packages, given that fiscal multipliers vary across instruments (largest for government consumption, public investment, and transfers; relatively smaller for indirect taxes).

### Structure note
- The remainder of the paper reviews relevant literature and presents empirical analysis exploring crisis duration and post-crisis growth as functions of fiscal policy size and composition.

*Source: _wp09160 - 1. Fiscal Aggregates (IMF Working Paper excerpt).*

### Section III describes the data and the econometric approach. Section IV presents the

### _wp09160 - Section III describes the data and the econometric approach. Section IV presents the

### II. Literature Review
- Two major reference datasets and findings:
  - Reinhart and Rogoff (2008a, 2008b, 2009):
    - Identify 66 banking crisis cases between 1945 and 2007.
    - Find government debt rises by 86 percent during the three years following a banking crisis.
    - Growth resumes slowly to reach an average annual rate of 2½ ercent in the third year after the crisis.
  - Laeven and Valencia (2008):
    - Identify 124 systemic banking crises over 1970–2007.
    - Estimate fiscal costs net of recoveries average about 13.3 percent of GDP.
    - Estimate output losses average 20 percent of GDP.
- Origins and consequences of crises:
  - Crises tend to erupt with weak macroeconomic environment: low growth, high inflation and interest rates (Demirgüc-Kunt and Detragiache, 1998; Collyns and Kincaid, 2003).
  - Recessions associated with credit crunches and house price busts tend to be deeper and longer (Claessens, Kose, and Terrones, 2008).
- Policy response literature:
  - Monetary and fiscal policy tend to be countercyclical during recessions, credit contractions, and asset price declines (Claessens, Kose, and Terrones, 2008; IMF, 2009b).
  - Expansionary fiscal policy (proxied by discretionary government consumption) tends to shorten the duration of recessions.
  - Prescriptive advice: optimal fiscal package should be large, lasting, diversified, contingent, collective, and sustainable (Spilimbergo et al., 2008).
- Financial market perceptions:
  - Ardagna (2009): financial markets value fiscal discipline; interest rates on long-term government bonds and stock market prices worsen during fiscal expansion periods.
  - Akitoby and Stratmann (2008): revenue-based adjustments lower government spreads more than expenditure-based ones; debt-financed spending increases sovereign risks.
  - Baldacci, Gupta, and Mati (2008): spending on public investment contributes to lower government bond spreads provided fiscal position remains sustainable.

### III. Fiscal Policy During Banking Crises — data and stylized facts
- Dataset:
  - Panel of 182 countries between 1980 and 2008.
  - Follow Laeven and Valencia (2008) criteria; identify 118 episodes of banking crises in 99 countries (some countries up to four times).
  - Complemented with WEO, GFS, and GFD data.
- Crisis duration definition:
  - Crisis ends after two consecutive years of real GDP growth above ½ percentage points per year.
  - Robustness tested using a stock market‑based duration definition in Section V.
- Duration statistics:
  - Average crisis duration: 2½ years.
  - 85 percent of crisis episodes last between one to four years.
  - Only one episode lasted eight years.
- Peak-to-trough and period changes (selected figures):
  - Peak-to-trough: average GDP growth rate fell by more than 5 percentage points during the crisis.
  - Peak-to-trough: general government debt increased by 39 percentage points of GDP.
  - Peak-to-trough: budget deficits increased by 6.9 percentage points of GDP.
  - Pre-crisis to crisis: real GDP growth fell from 0.4 percent (year prior) to an average -2.2 percent during the crisis; in the two-year period following the crisis annual GDP growth rebounded to 4.9 percent per year (level still below pre-crisis).
- Two-year changes of fiscal aggregates (Table 1; all figures as percent of GDP):
  - Debt: Before (t-2; t-1) -9.2; During (t) 27.1; After (t+1; t+2) -7.2
  - Budget balance: Before -0.1; During -5.9; After 1.5
  - Primary budget balance: Before 0.3; During -4.9; After 2.8
  - Total revenues: Before 0.8; During -3.7; After 4.9
  - Total expenditures: Before 0.9; During 2.3; After 2.6
- Crisis period averages and interpretations:
  - During banking crises fiscal deficits increased by almost 6 percentage points (more than 2 percent of GDP per year).
  - Public debt worsened by 27 percentage points of GDP (about ⅓ of the preexisting average debt level of 78 percent of GDP).
  - Total revenues fell by more than 3½ percentage points of GDP; government expenditures rose by more than 2 percentage points of GDP.
- Revenue composition (Table 2; as percent of GDP, two-year windows):
  - Taxes: Before 0.5; During -2.3; After 4.2
  - Income, profits, capital gains: Before 0.2; During -1.2; After 3.8
  - Payroll and workforce: Before 0.1; During -0.3; After 0.0
  - Goods and services: Before 0.1; During -0.5; After 0.4
  - International trade: Before 0.1; During -0.3; After 0.0
  - Social contributions: Before 0.2; During -1.2; After 0.2
  - Other revenues: Before 0.1; During -0.2; After 0.5
- Expenditure composition (Table 3; as percent of GDP, two-year windows):
  - Current expenditure: Before 0.9; During 2.2; After 0.1
    - Goods and services: Before -0.1; During 0.6; After -0.5
    - Employee compensation: Before 0.1; During 0.2; After 0.1
    - Transfers: Before 0.1; During 0.6; After 0.3
    - Interest payments: Before 0.4; During 1.0; After 2.3
    - Other expenses: Before 0.4; During -0.2; After -0.1
  - Public Investment: Before 0.0; During 0.1; After 2.5
- Stylized policy interpretation:
  - Crisis fiscal outcomes reflect automatic stabilizers and discretionary measures (including measures to strengthen the financial system).
  - Public consumption rises immediately and contributes to aggregate demand; public investment often faces implementation lags.

### IV. The Effectiveness of Fiscal Response — empirical strategy and main results
- Research aims:
  - Assess effectiveness of fiscal policy response in (i) reducing duration of banking crises and (ii) promoting post-crisis economic growth.
- Preliminary bivariate evidence (Table 4 highlights):
  - Change in total public expenditures over crisis episode: Coefficient 0.103***; T 4.5; R-squared 0.1118
  - Change in public consumption (percent of total public expenditure): Coefficient 0.063*; T 1.7; R-squared 0.0118
  - Change public investment (percent of total public expenditure): Coefficient 0.266***; T 6.8; R-squared 0.3118
  - Change in total public revenues over crisis episode: Coefficient 0.086***; T 3.9; R-squared 0.1118
  - Change tax from Income (percent of total public revenue): Coefficient -0.265***; T -3.3; R-squared 0.1118
  - Change tax from Goods & Services (percent of total public revenue): Coefficient 0.319**; T 2.6; R-squared 0.0118
  - Note: Post-crisis growth defined as average GDP growth rate during the next five years after end of crisis.
- Duration regression baseline model (equation (1)):
  - Dependent variable: discrete duration from 1 to 8 years; estimation via OLS and Ordered Logit on 118 episodes.
  - Key findings (Table 5):
    - Budget Balance (in percent of GDP): Coefficient 0.072*** (Model 1 OLS); t-statistic (3.73)
    - Expansionary fiscal policy dummy: Coefficient -0.626*** (Model 2 OLS) and -1.023*** (Model 4 Ord. Logit)
    - Previous credit boom: positive and significant (e.g., 0.690*** in Model 1)
    - Deposit freeze or guarantee: negative and significant (e.g., -0.522** in Model 1)
    - Number of banks closed: negative and significant (e.g., -0.168*** in Model 1)
    - Government intervention: negative and significant (e.g., -0.721*** in Model 1)
    - Adj. R-squared / Pseudo R-squared: 0.435 (Model 1) and 0.211 (Model 3)
  - Magnitudes and interpretation:
    - Fiscal expansions reduce crisis length: average fiscal policy response in the sample would reduce crisis length by more than two quarters.
    - Deposit guarantees reduce crisis length by two to four quarters.
    - Closing failed banks reduces crisis length by about one year.
    - Crises preceded by credit booms last about one year longer.
- Budget composition and crisis length (equation (2); Table 6 highlights):
  - Expansionary fiscal policy: coefficients around -0.522** to -0.601** (OLS) and around -0.945** to -1.049** (Ord. Logit).
  - Public consumption (percent of total expenditures): negative and significant (e.g., -0.035*** in Model 1 OLS).
    - A 5 percentage point increase in public consumption share reduces crisis length by almost three months (interpreted from estimated coefficient magnitudes).
  - Public investment (percent of total expenditures): negative but weaker significance (e.g., -0.027*).
  - Income tax revenue (percent of total revenues): positive and significant in some specifications (e.g., 0.076***), implying reducing income-tax share helps shorten crises.
  - Goods & services tax revenue (percent of total revenues): positive and significant in some specifications (e.g., 0.119***).
  - Policy controls retain expected signs: credit booms lengthen crises; deposit guarantees, number of banks closed, and government intervention shorten crises.
- Post-crisis growth regressions (equation (3); Table 7 highlights):
  - Expansionary fiscal policy: not statistically significant for average GDP growth in the five years post-crisis (coefficients e.g., 0.262, 0.251).
  - Public investment (percent of total expenditures): positive and significant (e.g., 0.229***; T 4.94).
    - A one percent increase in the share of capital outlays in the budget raised post-crisis growth by about ⅓ of one percent per year (text summary).
  - Income tax revenue (percent of total revenues): negative and significant for growth (e.g., -0.177**).
  - Goods & services tax revenue: positive and significant for growth (e.g., 0.402***).
  - Private sector variables matter:
    - Private Investment (percent of total investment): large positive coefficients (e.g., 7.530**).
    - Cost of financing: negative and significant (e.g., -0.121***).
    - Fresh capital injections into financial sector: positive and significant in several specifications (e.g., 1.453**).
- Initial conditions and interactions:
  - Two dummies: Highly Indebted (initial public debt above sample average) and High GDP per capita (initial GDP per capita above sample average).
  - Key effects:
    - High initial public debt increases crisis duration (e.g., Highly Indebted (t-1) coefficient 0.798** in Table 8).
    - Expansionary fiscal policy benefits are weaker when initial conditions are poor; in high‑debt countries crises lasted almost one year longer—effect offsets benefits of expansionary policy.
    - High GDP per capita reduces crisis duration (Table 9: High GDP per capita (t-1) coefficients negative, e.g., -0.345***).
    - Interaction terms: expansionary fiscal policy * High GDP per Capita (t-1) negative and significant in some specifications (e.g., -0.876**), indicating stronger benefits where per capita income is higher.
  - For post-crisis growth, interaction results show:
    - Public investment and reductions in income taxes more effective for growth when initial conditions (debt, GDP per capita) are stronger (Tables 10 and 11).

### V. Robustness Analysis
- Three robustness checks reported:
  - Alternative duration definition:
    - Define end of crisis as first year stock market index returns to precrisis level.
    - Episodes tend to be shorter; baseline results robust (see Annex Table A5).
  - Alternative discretionary fiscal policy index:
    - Construct Blanchard (1990)-style discretionary fiscal indicator.
    - Baseline results consistent with this new formulation (see Annex Table A6).
  - Endogeneity between duration and fiscal policy:
    - Use Two-Stage Least Squares (TSLS) with instruments (all other independent variables and a measure of liquidity support).
    - Main findings hold (see Annex Table A7).

### VI. Conclusion — summary of policy-relevant findings and implications
- Main empirical conclusions:
  - Timely countercyclical fiscal responses (discretionary measures and automatic stabilizers), when accompanied by actions addressing financial sector weaknesses, shorten crisis duration.
  - Expansionary fiscal policies reduced crisis duration by almost one year on average (text summary).
  - Composition matters for crisis length:
    - Stimulus relying on government consumption more effective in shortening crisis duration than those relying on public investment.
    - A 10 percentage point increase in the share of public consumption in the budget reduces crisis length by three to four months.
    - Reducing the share of income taxes is less effective than consumption tax cuts in shortening crisis length.
  - Post-crisis growth:
    - Fiscal expansions per se do not have a significant impact on medium-term GDP growth.
    - Scaling up public investment and reducing income taxes are associated with better medium-term growth outcomes.
    - A one percent increase in the share of capital outlays in the budget raised post-crisis growth by about ⅓ of one percent per year.
  - Initial conditions matter:
    - High precrisis public debt and low per capita income limit the scope and effectiveness of countercyclical fiscal policy.
    - In high-debt countries, crises last almost one year longer; fiscal expansion benefits are offset when initial debt is high.
  - Private sector and resolution policies matter for growth and recovery:
    - Higher private investment share, lower cost of financing, and fresh capital injections into the banking sector positively affect medium-term growth.
- Policy implications:
  - In crises caused by financial sector distress, timely countercyclical fiscal responses should be considered as part of the policy toolkit to shorten crisis duration.
  - Fiscal space and strong initial fiscal positions are essential to enable effective countercyclical responses—build fiscal space in tranquil times.
  - Design stimulus composition carefully:
    - For short-term exit from crisis, government consumption increases have immediate effects.
    - For medium-term growth, prioritize scaling up public investment and consider income tax reductions.
    - Anticipate implementation lags for public investment; ensure shovel-ready projects and efficient execution to realize medium-term gains.
  - Strengthen fiscal institutions, reduce political risks, and improve budget execution capacity—especially in emerging market economies—to reap benefits of countercyclical fiscal policy.
- Further research suggested:
  - Investigate potential crowding-in effects where public investment is associated with increases in private investment during financial stress; preliminary evidence exists but requires dedicated testing.

*Source: Authors’ calculations and estimates as reported in the provided IMF working paper section.*

### Appendix Table A1. Episodes of Banking Crisis in the World, 1980–2008

### _wp09160 - Appendix Table A1. Episodes of Banking Crisis in the World, 1980–2008

### Episodes of Banking Crisis: coverage
- The appendix lists episodes of banking crisis for many countries during 1980–2008. (Individual country entries record the start year of the episode, "Episodes", "Duration (1)", and "Duration (2)".)
- Examples extracted from the table (preserving original numeric entries exactly):
  - Albania: 1994; Episodes 4; Duration (1) 3
  - Argentina: 1980; Episodes 3; Duration (1) 3
  - Argentina: 1989; Episodes 2; Duration (1) 2
  - Argentina: 1995; Episodes 1; Duration (1) 1
  - Argentina: 2001; Episodes 2; Duration (1) 1
  - Mexico: 1981; Episodes 6; Duration (1) 4
  - Mexico: 1994; Episodes 2; Duration (1) 2
  - Paraguay: 1995; Episodes 8; Duration (1) 6
  - United Kingdom: 2007; Episodes 2; Duration (1) 2
  - United States: 1988; Episodes 4; Duration (1) 3
  - United States: 2007; Episodes 2; Duration (1) 2
  - Uruguay: 1981; Episodes 4; Duration (1) 4
  - Uruguay: 2002; Episodes 1; Duration (1) 1
  - Japan: 1997; Episodes 2; Duration (1) 2
  - Thailand: 1997; Episodes 2; Duration (1) 2
- (The full country-level listing appears in the appendix table; above are direct verbatim extracts.)

### Definitions of duration measures (notes from the appendix)
- Duration 1: "counts the number of years between the start of the crisis (as identified by Laeven and Valencia, 2008) and the start of two consecutive years of GDP growth above 0.5 percent of per year."
- Duration 2: "counts the number of years between the start of the crisis (as identified by Laeven and Valencia, 2008) and the year in which the stock market (measured by its major index) recovered the level previous to the crisis. Data for these stock market values comes from the Global Financial Database (2009)."

### Budget Composition: Revenues (changes, in percent of total revenue)
- Figures show the change in the variables during the period compared to the precrisis year.
- Aggregated changes (Before Crisis (t-2; t-1), During Crisis (t), After Crisis (t+1; t+2)):
  - Taxes: 1.45; -14.21; 1.84
    - Income, profits, capital gains: 1.39; -8.31; 1.74
    - Payroll and workforce: 0.19; -1.29; 0.09
    - Property: 0.14; -1.16; -0.18
    - Goods and services: 0.15; -2.13; 0.48
    - International trade: -0.37; -1.67; -0.08
    - Other taxes: -0.05; 0.35; -0.21
  - Social contributions: 1.32; -7.83; 1.61
  - Other revenues: 1.67; -2.09; 3.17
- Source/computation: "Source: Own elaboration. Data: WEO and GFS."

### Budget Composition: Expenditures (changes, as percent of total expenditure)
- Figures show the change in the variables during the period compared to the precrisis year.
- Aggregated changes (Before Crisis (t-2; t-1), During Crisis (t), After Crisis (t+1; t+2)):
  - Current Expenditure: 0.10; -9.51; 2.43
    - Goods and services: -0.24; -1.78; 0.35
    - Employee compensation: -1.89; -4.04; -0.53
    - Transfers: 0.11; -2.54; 1.12
    - Interest payments: 0.76; 0.33; 1.46
    - Other expenses: 1.36; -1.47; 0.03
  - Public Investment: -0.24; 0.98; 0.77
- Source/computation: "Source: Own elaboration. Data: WEO and GFS."

### The Relationship Between Containment and Resolution Policies and Crisis Length (OLS estimates)
- Dependent variable: length of banking crisis.
- Sample size: Observation 118 for each reported coefficient.
- Key coefficient estimates (Coefficient; T-stat; R-squared; Observation) — as reported verbatim:
  - Credit Boom (t-1): 0.933*** 3.750.11 118
  - Deposit Freeze: -0.951*** -3.830.12 118
  - Bank Holiday: -0702** -3.130.06 118
  - Blanket Guarantee: -0.721** -3.210.07 118
  - Liquidity Support: 0.769** 3.440.08 118
  - Forbearance: 0.787** 3.510.08 118
  - Government Intervention: -0.973*** -3.910.11 118
  - Number of Bank Closures: -0.289*** -5.450.19 118
  - Fresh Private Capital: -0.321 -1.480.01 118
  - Number Foreign Sales: 0.482*** 4.810.16 118
  - Bank Agency: -0.932*** -4.240.11 118
  - Public Recapitalization: 0.803** 3.490.08 118
- Significance notation (as reported): "*** significant at 1percent; ** significant at 5 percent; * significant at 10 percent"
- Note on data: "Laeven and Valencia (2008) report containment and resolution policies data for 42 cases. We complemented their database using inference analysis where concrete data was not available from IMF country reports."

### Robustness estimations (alternative crisis-duration definition and discretionary fiscal policy specifications)
- Appendix Table A5 — Duration of Crisis (2) (stock-market-recovery definition). OLS estimates, Observations 118, Adj. R-squared reported by model:
  - Model 1 adj. R-squared 0.335; Model 2 adj. R-squared 0.314; Model 3 adj. R-squared 0.333; Model 4 adj. R-squared 0.334
  - Consistent statistically significant coefficients reported (verbatim entries include coefficients and t-statistics in parentheses):
    - Expansionary fiscal policy: -0.709*** (-3.91); -0.756*** (-4.12); -0.740*** (-4.12); -0.751*** (-4.19)
    - Public consumption (percent of total expenditures): -0.018* (-1.89) in Model 1
    - Public investment (percent of total expenditures): -0.003 (-0.26) in Model 2
    - Income tax revenue (percent of total revenues): 0.038* (1.80) in Model 3
    - Goods & services tax revenue (percent of total revenues): 0.058* (1.84) in Model 4
    - Previous Credit boom: 0.193 (1.12); 0.226 (1.30); 0.205 (1.19); 0.058* (1.84)
    - Deposit freeze or guarantee: -0.287* (1.69); -0.310* (1.77); -0.241 (-1.38); -0.294* (-1.73)
    - Number of banks closed: -0.091** (-2.39); -0.110** (-2.67); -0.101** (-2.48); -0.097** (-2.37)
    - Government intervention: -0.421** (-2.43); -0.474** (-2.71); -0.487*** (2.84); -0.485*** (-2.83)
    - Constant terms: 3.147*** (14.02); 3.215*** (14.30); 3.238*** (14.60); 3.147*** (14.00)
- Appendix Table A6 — Focusing on Discretionary Expansionary Fiscal Policy. OLS estimates, Observations 118, Adj. R-squared reported:
  - Model 1 adj. R-squared 0.436; Model 2 adj. R-squared 0.404; Model 3 adj. R-squared 0.434; Model 4 adj. R-squared 0.436
  - Key coefficients (verbatim, with t-statistics in parentheses):
    - Discretionary Expansionary fiscal policy: -0.370* (-1.76); -0.418* (-1.94); -0.437** (-2.08); -0.458** (-2.19)
    - Public consumption (percent of total expenditures): -0.377*** (-3.25) in Model 1
    - Public investment (percent of total expenditures): -0.030* (-1.98) in Model 2
    - Income tax revenue (percent of total revenues): 0.079*** (3.16) in Model 3
    - Goods & services tax revenue (percent of total revenues): 0.123*** (3.25) in Model 4
    - Previous Credit boom: 0.562** (2.72); 0.615*** (2.90); 0.582** (2.82); 0.583** (2.83)
    - Deposit freeze or guarantee: -0.557** (-2.72); -0.561** (-2.65); -0.457** (-2.17); -0.567** (-2.77)
    - Number of banks closed: -0.148*** (-3.06); -0.163*** (-3.31); -0.155*** (-3.22); -0.148*** (-3.05)
    - Government intervention: -0.675*** (-3.26); -0.741*** (-3.51); -0.805*** (-3.94); -0.800*** (-3.92)
    - Constant terms: 3.637*** (13.83); 3.757*** (14.05); 3.826*** (14.65); 3.633*** (13.80)
- Significance notation repeated in these tables: "*** significant at 1percent; ** significant at 5 percent; * significant at 10 percent"
- Dependent variable for both robustness tables: "length of banking crisis."
- Source for estimates: "Source: Authors’ estimates."

*Source: _wp09160 - Appendix Table A1. Episodes of Banking Crisis in the World, 1980–2008 (content as provided in the supplied appendix pages).*

### Appendix Table A7. Robustness Estimations: Controlling for Endogeneity

### Appendix Table A7. Robustness Estimations: Controlling for Endogeneity (2SLS Estimates)

### Main findings: coefficients on determinants of Duration of Crisis (four models)
- Expansionary fiscal policy:
  - Model 1: -0.522**  (-2.45)
  - Model 2: -0.572**  (-2.61)
  - Model 3: -0.581**  (-2.74)
  - Model 4: -0.602**  (-2.85)

- Public consumption (percent of total expenditures):
  - Model 1: -0.035***  (-3.12)

- Public investment (percent of total expenditures):
  - Model 2: -0.027*  (-1.82)

- Income tax revenue (percent of total revenues):
  - Model 3: 0.076***  (3.07)

- Goods & services tax revenue (percent of total revenues):
  - Model 4: 0.119***  (3.18)

- Previous Credit boom:
  - Model 1: 0.568**  (2.80)
  - Model 2: 0.621***  (2.99)
  - Model 3: 0.590***  (2.91)
  - Model 4: 0.592***  (2.93)

- Deposit freeze or guarantee:
  - Model 1: -0.555**  (-2.76)
  - Model 2: -0.563**  (-2.72)
  - Model 3: -0.461**  (-2.24)
  - Model 4: -0.568***  (-2.84)

- Number of banks closed:
  - Model 1: -0.137**  (-2.86)
  - Model 2: -0.152***  (-3.09)
  - Model 3: -0.143***  (2.99)
  - Model 4: -0.135**  (-2.82)

- Government intervention:
  - Model 1: -0.713***  (-3.48)
  - Model 2: -0.781***  (-3.74)
  - Model 3: -0.841***  (-4.16)
  - Model 4: -0.837***  (-4.16)

- Constant terms:
  - Model 1: 3.737***  (14.12)
  - Model 2: 3.854***  (14.36)
  - Model 3: 3.917***  (14.98)
  - Model 4: 3.731***  (14.12)

### Key statistics and model diagnostics
- Observations: 118 (for each model)
- Adj. R-squared:
  - Model 1: 0.450
  - Model 2: 0.419
  - Model 3: 0.449
  - Model 4: 0.452

- Significance notation:
  - *** significant at 1 percent
  - ** significant at 5 percent
  - * significant at 10 percent

### Dependent variable, instruments, and estimation method
- Dependent variable: length of banking crisis.
- Estimation: 2SLS (instrumental variables).
- Note: Instrumented variable: Expansionary fiscal policy; Instrument: Liquidity Support.

*Source: Authors’ estimates.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09160.pdf_