## _wp15166

## Source details

**Canonical URL:** [_wp15166](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15166.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15166.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15166.pdf.json)

---

### I. Introduction — context and motivation
- Since 2007, there have been 25 new systemic and borderline systemic banking crises, mostly in advanced economies.
- Examples and headline statistics:
  - In Iceland and Ireland, the cost of government intervention amounted to more than 40 percent of GDP.
  - Public debt increased by more than 70 percent of GDP in five years in some cases.
  - Median direct fiscal cost of government intervention during crises between 1980 and 2011: about 6 percent of GDP.
  - About one third of crisis episodes recorded direct fiscal costs exceeding 10 percent of GDP.
  - Median increase in public debt during the four years that followed crises over 1980–2011: more than 14 percent of GDP.
  - For the 11 most costly crises the increase exceeded 40 percent of GDP.
  - The most recent wave of crises had a median increase in public debt of about 24 percent of GDP.
- Purpose of the analysis:
  - Empirically analyze factors associated with direct fiscal costs and public debt dynamics.
  - Examine how precrisis banking sector characteristics, regulatory/supervisory frameworks, and policy responses help explain fiscal costs.
  - Address limitations of prior literature by considering both direct and overall fiscal costs, recent banking sector complexities, and interactions among clusters of risk factors.

### II. Conceptual framework — definitions and channels
- Definitions preserved exactly:
  - Direct fiscal costs: result from direct governmental interventions in the banking sector during a crisis (recapitalizations, asset purchases, compensating depositors, calls on guarantees, central bank recapitalization for losses).
  - Indirect fiscal costs: incurred through crisis effects on the real economy (interest rates, GDP growth, asset prices) that reduce revenue and increase spending and interest expenditures.
  - Overall fiscal costs: captured by changes in gross public debt ratios, reflecting both direct and indirect effects (but may include effects of other events around a banking crisis).
- Measurement window:
  - Changes in public debt ratios are measured over [T-1; T+4], where T is the starting year of the banking crisis.
- Theoretical/empirical channels and risk factors:
  - Precrisis macroeconomic conditions: credit booms, inflated asset prices, growth, fiscal performance; better external sector performance provides resilience.
  - Precrisis banking sector features: size of banking system, bank leverage, reliance on external funding, international interconnectedness, and nonfinancial sector leverage.
  - Institutional setting: strength of regulatory and supervisory frameworks, resolution frameworks, deposit insurance schemes.
- Policy-response trade-offs:
  - Accommodating policies (blanket guarantees, open-ended liquidity support, repeated recapitalizations, debtor bail-outs, regulatory forbearance) tend to add to fiscal costs.
  - Costly initial interventions may smooth macroeconomic impact and lead to smaller overall costs; low initial-cost measures may generate higher future costs.

### III. Data and stylized facts
- Datasets combined: Laeven and Valencia (2013), World Bank (2012) financial variables, Abbas and others (2011) and Mauro and others (2013) for public debt, IMF April 2014 WEO, Abiad and others (2008) banking supervision index; Laeven and Valencia (2013) on deposit insurance and peak NPLs.
- Direct fiscal costs in dataset: sum of governments’ recapitalization costs and asset purchases and, where relevant, central banks’ recapitalizations and loans to banks. Direct fiscal costs are defined on a gross basis.
- Sample: cross-sectional set of 65 banking crisis episodes from 1980 to 2011 involving 56 high- and middle-income economies.
- Key stylized facts:
  - Sample median direct fiscal cost: about 6 percent of GDP.
  - Sample median increase in public debt: more than 14 percent of GDP.
  - Emerging economies had on average direct fiscal costs twice as high as advanced economies, but their increase in public debt-to-GDP ratios was about half that of advanced economies.
  - Direct fiscal costs exceeded 40 percent of GDP in Argentina, Chile, Iceland, Indonesia, Ireland, Jamaica, and Thailand.
  - Increase in public debt around banking crises exceeded 80 percent of GDP in Argentina and Chile in the sample.
- Recovery rates (sample of 38 countries):
  - Median recovery rate: about 7 percent of gross fiscal costs.
  - Maximum recovery rate: 94 percent (Sweden).
  - No recoveries in about one third of cases.
  - Recovery rates defined as recovery proceeds during the period T to T+5 divided by the gross direct fiscal cost.

### IV. Main empirical findings (summary)
- Precrisis banking-sector characteristics:
  - Fiscal costs are higher where the banking sector is larger, more leveraged, or more reliant on external funding.
- Institutions and supervision:
  - Fiscal costs tend to be lower where banking supervision is stronger and deposit insurance coverage is broader.
- Policy responses:
  - Bank guarantees appear to increase both direct and overall fiscal costs.
  - Asset purchases and recapitalizations—though they have initial direct fiscal costs—are not necessarily associated with larger increases in public debt over time.
- Trade-offs:
  - Evidence consistent with a trade-off between costly short-term policy interventions and the overall increase in public debt: swift interventions may be initially costly but could lead to better macroeconomic outcomes and smaller increases in public debt.

### V. Box 1 — Recording Direct Fiscal Costs of Banking Crises under the Government
- Recording framework and valuation (GFSM 2001):
  - Government intervention recording depends on whether the government acquires an effective claim and on the mode of financing.
  - Valuation at acquisition determines immediate loss or not; subsequent market-value changes are recorded as other economic flow.
- Three main intervention types (definitions preserved exactly):
  - (1) Requited recapitalization: government takes an equity stake, extends a loan or purchases a bad asset at the market value.
  - (2) Unrequited recapitalization: government injects capital in a bank but does not effectively receive any claim.
  - (3) The purchase of a bad asset at a premium: government acquires an asset in a bank at a cost higher than market value.
- Immediate fiscal-account impacts (symbols preserved):
  - Requited recapitalization: Net lending/Borrowing: ✕; Overall Balance: ↓; Gross Public Debt: ✕ (and ↑ in alternate financing row); Net Public Debt: ✕.
  - Unrequited recapitalization: Net lending/Borrowing: Net lending/Borrowing ↓; Overall Balance: ↓; Gross Public Debt: ✕ (and ↑ in alternate financing row); Net Public Debt: ↑.
  - Bad asset purchase at a premium: Net lending/Borrowing: Net lending/Borrowing ↓ 3/; Overall Balance: Overall Balance ↓ 4/; Gross Public Debt: Gross Public Debt ✕ (and ↑ in alternate financing row) 4/; Net Public Debt: Net Public Debt ↑ 3/.
- Notes:
  - 1/ Does not include the secondary impact of these operations (e.g. the impact of revenue through dividends received from an acquired asset).
  - 2/ Includes taking an equity stake in a financial institution, extending a loan or purchasing a bad asset at the market value.
  - 3/ Changes by the amount of the difference between the market value and the purchase price.
  - 4/ Changes by the amount of the purchase price.

### VI. Econometric evidence — direct fiscal costs (key results preserved)
- Basic model (Tables A1a, A1b):
  - Dependent variable: log of direct fiscal costs in percent of GDP.
  - Banking supervision index coefficients: e.g., -0.443***; -0.428***; -0.403***; -0.545**; -0.615*** (standard errors reported in source).
  - Deposit insurance coverage coefficients: -0.00186**; -0.00191**; -0.00348***; -0.00279** (standard errors reported).
  - Interconnectedness measures:
    - Offshore deposits/domestic deposits: 0.0230**; 0.0295*** (where reported).
    - Non-resident loans/GDP: 0.00704**; 0.00666*.
  - Size and leverage:
    - Non-financial sector leverage (private credit/GDP): 0.00871** (where included).
    - Size of financial sector (assets/GDP): 0.00771** (where included).
- Extended models and robustness (Tables A2–A5, A4 accounting for Peak NPLs):
  - Peak NPLs coefficient: 0.0238**; 0.0233**; 0.0296*; 0.0236* in Table A4.
  - Bank guarantee (dummy) coefficients across specifications: 0.380; 0.388; 0.625**; 0.743**; 0.615*; and in GMM: 0.812**; 0.883***; 1.189***; 0.707***.
  - GMM results largely confirm OLS findings for key precrisis variables and strengthen significance of bank guarantee coefficients in some specifications.
  - Interaction tests for recent crises (2007–11) yield mostly insignificant interaction terms; exceptions noted for supervision quality and international interconnectedness in the recent wave.
- Summary of statistical significance:
  - Banking supervision index: consistently negative and statistically significant in many specifications (e.g., -0.461*** in Table A2).
  - Interconnectedness and funding-mix indicators: positive and often significant.
  - Income per capita often negative and significant in some specifications (e.g., -0.00320* in Table A2).

### VII. Econometric evidence — overall fiscal costs (debt changes)
- Dependent variable: change in public debt ratios over [T-1; T+4].
- Key drivers (Tables A6–A9, A8 accounting for NPLs and depositor losses):
  - Change in Primary Balance (t, t+3): large, robust negative coefficients across almost all specifications (e.g., -0.905***; -0.972***; -1.046***; -1.073***; and many more in extended tables).
  - Size and leverage effects:
    - Non-Financial Sector Leverage (Private Credit/GDP) positive and significant in many specifications (e.g., 0.113**; 0.148**).
    - Size of Financial Sector (Assets/GDP) positive and significant in some specifications (e.g., 0.095*; 0.136**).
    - Financial Sector Leverage (Loans/Deposits) positive and significant in several specifications (e.g., 0.049***).
  - Bank guarantees: associated with higher overall fiscal costs in multiple specifications (e.g., coefficients such as 9.223; 11.224*; 12.532**; 11.993*; 13.260*).
  - Asset purchases and bank recapitalizations: not consistently associated with higher increases in public debt; some specifications show significance for recapitalization or asset purchases in GMM variants.
  - Robustness: results stable across alternative horizons (4, 5, 6 years), controlling for peak NPLs and depositor losses, and GMM instrumentation.
- Findings on depositor-loss and NPLs (Table A8):
  - Peak NPL (% total loans): coefficients reported across many specifications (e.g., 0.350; 0.366; 0.272; ...).
  - Loss imposed on depositors (dummy): coefficients vary widely with very large standard errors in several columns; some specifications show large positive coefficients (e.g., 22.703** in a column) but overall noisy estimates.

### VIII. Policy implications and conclusions
- Risk monitoring and prevention:
  - Identified risk indicators (precrisis macro, financial, institutional conditions) can help explain fiscal costs and should be monitored.
- Institutional reforms:
  - Strengthening regulatory and supervisory frameworks can mitigate fiscal risks; better banking supervision consistently associated with lower direct fiscal costs.
- Crisis management:
  - Acting early in crises may be beneficial; costly short-term interventions (asset purchases, recapitalizations) can limit longer-term increases in public debt.
  - Low-upfront-cost measures (guarantees) may raise both direct and overall costs and can create moral hazard.
- Suggested further research:
  - Develop a more refined measure of overall fiscal cost and isolate the debt change portion specifically due to banking crises.
  - Study determinants of recovery rates and crisis duration.
  - Investigate whether banks’ ownership structure (foreign or sovereign shareholders) affects fiscal costs.

### IX. Data sources and appendix notes
- Data sources (Table A10) include:
  - Direct fiscal costs — Laeven and Valencia (2013).
  - Income per capita — WEO.
  - Public debt — Abbas and others (2010) spliced with Mauro and others (2013).
  - Current account balance — WEO.
  - Banking supervision index — Abiad and others (2008).
  - Deposit insurance coverage and Peak NPLs — Laeven and Valencia (2013).
  - Financial sector leverage, size, and interconnectedness — World Bank (2012).
  - Policy response variables — Laeven and Valencia (2013).
- Systemic crises listing (Table A11): country-year episodes from Laeven and Valencia (2013) (selected examples preserved in source).

*Source: _wp15166 - APPENDIX TABLES*

### 1. Direct Fiscal Costs across Selected Vulnerability Indicators for Recent Banking Crisis, 2007–11. 17

### 1. Direct Fiscal Costs across Selected Vulnerability Indicators for Recent Banking Crisis, 2007–11. 17

### I. Introduction — context and motivation
- Since 2007, there have been 25 new systemic and borderline systemic banking crises, mostly in advanced economies, which have often carried significant fiscal costs.
- Examples: In Iceland and Ireland, the cost of government intervention amounted to more than 40 percent of GDP, and public debt increased by more than 70 percent of GDP in five years.
- Median direct fiscal cost of government intervention during crises that occurred between 1980 and 2011 was about 6 percent of GDP; about one third of crisis episodes recorded direct fiscal costs exceeding 10 percent of GDP.
- Median increase in public debt during the four years that followed crises over 1980–2011 was more than 14 percent of GDP; the increase exceeded 40 percent of GDP for the 11 most costly crises.
- The most recent wave of crises had a median increase in public debt of about 24 percent of GDP; the increase in some countries more than doubled this amount.
- Purpose of the analysis:
  - Empirically analyze factors associated with direct fiscal costs and public debt dynamics.
  - Examine how precrisis banking sector characteristics, regulatory/supervisory frameworks, and policy responses help explain fiscal costs.
  - Address limitations of prior literature by considering both direct and overall fiscal costs, recent banking sector complexities (e.g., cross-border linkages), and interactions among clusters of risk factors.

### II. Conceptual framework — definitions and channels
- Definitions:
  - Direct fiscal costs: result from direct governmental interventions in the banking sector during a crisis (recapitalizations, asset purchases, compensating depositors, calls on guarantees, central bank recapitalization for losses).
  - Indirect fiscal costs: incurred through crisis effects on the real economy (interest rates, GDP growth, asset prices) that reduce revenue and increase spending and interest expenditures.
  - Overall fiscal costs: captured by changes in gross public debt ratios, reflecting both direct and indirect effects (but may include effects of other events around a banking crisis).
- Time window for measuring changes in public debt: changes in public debt ratios are measured over [T-1; T+4], where T is the starting year of the banking crisis.
- Main theoretical/empirical factors associated with the probability and size of crises and with fiscal costs:
  - Precrisis macroeconomic conditions: credit booms, inflated asset prices, growth, and fiscal performance; better external sector performance provides resilience.
  - Precrisis banking sector features: size of banking system, bank leverage, reliance on external funding, international interconnectedness, and nonfinancial sector leverage.
  - Institutional setting: strength of regulatory and supervisory frameworks, resolution frameworks, deposit insurance schemes; weaker institutions can lead to higher fiscal costs via less efficient crisis management.
- Policy response trade-offs:
  - Accommodating policies (blanket guarantees, open-ended liquidity support, repeated recapitalizations, debtor bail-outs, regulatory forbearance) tend to add to fiscal costs.
  - Costly initial interventions may smooth macroeconomic impact and lead to smaller overall costs; conversely, low initial-cost measures may generate higher future costs (e.g., guarantees, forbearance).

### III. Data and stylized facts
- Datasets combined: Laeven and Valencia (2013) dataset of banking crises, World Bank (2012) financial variables, Abbas and others (2011) and Mauro and others (2013) for public debt, IMF April 2014 WEO, Abiad and others (2008) banking supervision index; Laeven and Valencia (2013) on deposit insurance and peak NPLs.
- Direct fiscal costs definition in dataset: sum of governments’ recapitalization costs and asset purchases and, where relevant, central banks’ recapitalizations and loans to banks. Direct fiscal costs are defined on a gross basis.
- Sample: cross-sectional set of 65 banking crisis episodes from 1980 to 2011 involving 56 high- and middle-income economies.
- Key stylized facts:
  - Sample median direct fiscal cost: about 6 percent of GDP.
  - Sample median increase in public debt: more than 14 percent of GDP.
  - Emerging economies had on average direct fiscal costs twice as high as advanced economies, but their increase in public debt-to-GDP ratios was about half that of advanced economies.
  - Direct fiscal costs exceeded 40 percent of GDP in Argentina, Chile, Iceland, Indonesia, Ireland, Jamaica, and Thailand.
  - Increase in public debt around banking crises exceeded 80 percent of GDP in Argentina and Chile in the sample.

### IV. Main empirical findings (summary)
- Precrisis banking sector characteristics matter:
  - Fiscal costs of banking crises are higher where the banking sector is larger, more leveraged, or more reliant on external funding.
- Institutions and supervision:
  - Fiscal costs tend to be lower where banking supervision is stronger and deposit insurance coverage is broader.
- Policy responses:
  - Bank guarantees appear to increase both direct and overall fiscal costs.
  - The correlation for other policy measures (recapitalizations and asset purchases) is less clear-cut: although they have initial direct fiscal costs, they do not necessarily add to the overall fiscal cost of crises.
- Trade-offs:
  - Evidence is consistent with a trade-off between costly short-term policy interventions and the overall increase in public debt; swift interventions may be initially costly but could lead to better macroeconomic outcomes and smaller increases in public debt.

### V. Scope and limitations emphasized by the paper
- Change in public debt is an imperfect measure of the overall fiscal costs because it captures debt variations around a banking crisis that may also reflect other events (sovereign and currency crises, fiscal adjustments, inflation, sovereign debt restructuring).
- Direct fiscal costs in the dataset do not include all possible direct costs that materialize later; they are recorded on a gross basis to better gauge immediate financial pressure.
- The analysis focuses on both direct and overall fiscal costs and accounts for interactions among macroeconomic variables, financial sector characteristics, institutions, and policy responses.

*1. Direct Fiscal Costs across Selected Vulnerability Indicators for Recent Banking Crisis, 2007–11. 17*

### Box 1. Recording Direct Fiscal Costs of Banking Crises under the Government

### Box 1. Recording Direct Fiscal Costs of Banking Crises under the Government

### Recording framework and valuation (GFSM 2001)
- Government intervention recording depends on whether the government acquires an effective claim and on the mode of financing.
- Proper valuation of acquired assets is critical to reporting government net worth:
  - Valuation at acquisition determines immediate loss or not.
  - Subsequent market-value changes are recorded as other economic flow (affect net financial worth but not net lending/borrowing).
  - Assessment is challenging during crises due to fire sales, uncertain recovery rates, and regulatory forbearance.

### Types of government interventions and immediate fiscal-account impacts
- Three main intervention types (definitions preserved exactly):
  - (1) Requited recapitalization: government takes an equity stake, extends a loan or purchases a bad asset at the market value.
  - (2) Unrequited recapitalization: government injects capital in a bank but does not effectively receive any claim.
  - (3) The purchase of a bad asset at a premium: government acquires an asset in a bank at a cost higher than market value.
- Summary of impacts on key fiscal indicators (symbols preserved as in source: ✕ - no impact, ↓ decline in value, ↑ increase in value):
  - Requited recapitalization:
    - Net lending/Borrowing: ✕
    - Overall Balance: ↓
    - Gross Public Debt: ✕ (in one row), ↑ (in another row corresponding to cash/issuance distinctions)
    - Net Public Debt: ✕
  - Unrequited recapitalization:
    - Net lending/Borrowing: Net lending/Borrowing ↓ (and repeated ↓ in alternate financing row)
    - Overall Balance: ↓
    - Gross Public Debt: ✕ (and ↑ in alternate financing row)
    - Net Public Debt: ↑
  - Bad asset purchase at a premium:
    - Net lending/Borrowing: Net lending/Borrowing ↓ 3/
    - Overall Balance: Overall Balance ↓ 4/
    - Gross Public Debt: Gross Public Debt ✕ (and ↑ in alternate financing row) 4/
    - Net Public Debt: Net Public Debt ↑ 3/
- Notes tied to table:
  - 1/ Does not include the secondary impact of these operations (e.g. the impact of revenue through dividends received from an acquired asset).
  - 2/ Includes taking an equity stake in a financial institution, extending a loan or purchasing a bad asset at the market value.
  - 3/ Changes by the amount of the difference between the market value and the purchase price.
  - 4/ Changes by the amount of the purchase price.

### Magnitude of fiscal costs and recoveries (descriptive observations)
- Crisis-wave and sample observations:
  - The wave of five banking crises during 2000–03 was the costliest in terms of direct and overall fiscal costs.
  - The 2007–11 wave includes 25 systemic and borderline systemic banking crises predominantly in advanced economies; direct fiscal costs of around 5 percent of GDP were low compared to previous crises, but the increase in public debt of around 20 percent of GDP was particularly large.
- Recovery rates:
  - Median recovery rate for the sample of 38 countries is only about 7 percent of gross fiscal costs.
  - The maximum recovery rate of 94 percent was recorded in Sweden.
  - No recoveries were achieved in about one third of the cases.
  - Recovery rates are defined as recovery proceeds during the period T to T+5, where T is the first year of the crisis, divided by the gross direct fiscal cost.

### Key risk factors associated with higher direct fiscal costs (descriptive statistics)
- Precrisis financial-sector characteristics:
  - Larger banking sectors (banking sector assets-to-GDP ratios) are associated with higher direct fiscal costs.
  - Higher precrisis leverage of the banking sector (loan-to-deposit ratio) and higher private sector leverage (private sector credit-to-GDP ratio) are associated with higher direct fiscal costs.
- Cross-border interconnectedness:
  - Higher financing from non-resident bank loans as a share of GDP correlates with higher direct fiscal costs.
  - Higher ratio of offshore deposits to domestic deposits correlates with higher direct fiscal costs.
- Institutional settings:
  - Higher GDP per capita (proxy for institutional development/administrative capacity) tends to associate with lower direct fiscal costs.
  - Better quality of banking supervision (banking supervision index) is negatively correlated with fiscal costs.
  - Broader deposit insurance coverage (ratio of insured deposits over per capita GDP) is associated with lower direct fiscal costs in the sample, though interpretation requires caution because countries with broader insurance may also be more likely to experience crises.

### Policy responses and correlations with fiscal costs
- Containment policies (observed correlations):
  - Countries that provided guarantees on banks’ liabilities following crisis onset or allowed regulatory forbearance have, on average, incurred higher direct fiscal costs.
  - Issuing new guarantees in the wake of a crisis is found to be costlier for governments than having pre-existing deposit insurance schemes.
  - Allowing for forbearance and incentives for banks to “gamble for resurrection” appears associated with larger costs.
- Resolution policies (descriptive for overall costs):
  - Bank guarantees (initially zero-cost instruments) are associated with higher overall fiscal costs.
  - Asset purchases and bank recapitalizations—policies with potentially high upfront costs—are not necessarily associated with higher overall costs (increase in public debt), suggesting a trade-off where costly short-term interventions may reduce long-term fiscal deterioration.

### Econometric findings (multivariate regressions and robustness)
- Direct fiscal costs regressions (controlling for macro conditions):
  - Precrisis banking-sector size (bank assets-to-GDP) and non-financial sector leverage (private credit-to-GDP) are significant and positively correlated with direct fiscal costs.
  - International interconnectedness indicators (loans from non-resident banks, offshore deposits ratio) remain positively correlated with direct fiscal costs when controlling for banking-sector characteristics.
  - Banking supervision quality and deposit insurance coverage are negatively correlated with direct fiscal costs.
- Policy variables in regressions:
  - Bank guarantees are associated with higher direct fiscal costs over the crisis period despite no initial cost; forbearance shows only weak relations within the broader sample.
  - Introducing policy responses does not alter the significance of precrisis banking vulnerabilities.
- Recent crises (2007–11) interaction tests:
  - Interacting a recent-crisis dummy with explanatory variables yields mostly insignificant interaction terms—risk indicators for past crises are largely relevant for recent crises.
  - Exceptions: quality of banking supervision and international interconnectedness matter less in the recent wave, likely because those crises occurred in economies with similar institutional and financial development.
- Robustness checks:
  - Accounting for crisis shock size via peak NPL ratio: results broadly similar; size of banking sector and forbearance become less significant in some specifications.
  - Instrumenting policy variables (GMM) with creditor-rights measure and timing dummies: confirms OLS results, suggesting reverse causality is likely not a major problem.

### Overall fiscal costs (debt changes) — findings and drivers
- Definition used: overall fiscal costs = change in debt-to-GDP ratio between year before crisis onset and four years after (alternative horizons tested).
- Descriptive results:
  - Size and leverage of banking and nonbanking sectors are positively associated with higher overall fiscal costs.
  - Cross-border interconnectedness is associated with costlier crises for overall costs in descriptive stats.
  - Contrary to direct costs, overall fiscal costs are higher in countries with broader deposit insurance coverage, possibly reflecting moral hazard in the long run.
  - Guarantee provision and regulatory forbearance correlate with higher overall costs.
  - Asset purchases and recapitalizations with high initial costs are not necessarily correlated with larger increases in public debt over time.
- Econometric results for overall costs:
  - Larger and more leveraged banking sectors correlate with higher overall fiscal costs.
  - Higher loan-to-deposit ratios are associated with higher overall fiscal costs; however, when controlling for supervision quality, funding-mix indicators become less relevant.
  - Bank guarantees that carry initial zero costs are consistently associated with higher overall costs; asset purchases and recapitalizations are not correlated with increased public debt.
- Robustness:
  - Changing the time span for measuring overall costs (four, five, six years) yields substantially the same results; coefficients tend to be smaller for shorter horizons.
  - Accounting for peak NPLs and depositor losses does not materially change main results.
  - GMM instrumentation yields less significance for precrisis variables but more for policy response variables.

### Policy implications and conclusions
- Identified risk indicators (precrisis macro, financial, institutional conditions) can help explain fiscal costs of banking crises.
- Direct fiscal costs are higher where banking sectors are larger, more leveraged, and more reliant on external funding.
- Better institutions—especially quality banking supervision—and broader deposit insurance coverage are associated with lower direct fiscal costs.
- There is evidence of a trade-off: some early, costly interventions (e.g., asset purchases, recapitalizations) can limit longer-term increases in public debt, whereas low-upfront-cost measures like guarantees may raise both direct and overall costs.
- Policy recommendations implied by findings:
  - Governments should identify and monitor specific banking-sector risk indicators and assess potential fiscal and debt-sustainability impacts.
  - Strengthening regulatory and supervisory frameworks can mitigate fiscal risks.
  - Acting early in crises may be beneficial; costly short-term interventions can be cost-effective in limiting longer-term fiscal deterioration.
- Areas for further research highlighted:
  - Develop a more refined measure of overall fiscal cost and isolate the debt change portion specifically due to banking crises.
  - Study determinants of recovery rates and crisis duration.
  - Investigate whether banks’ ownership structure (foreign or sovereign shareholders) affects fiscal costs.

*Source: The state of public finances: outlook and medium-term policies after the 2008 crisis, 2009, IMF, companion paper.*

### APPENDIX TABLES

### APPENDIX TABLES

### Basic Model: Conditional Correlations of Direct Fiscal Costs (Tables A1a, A1b)
- Dependent variable: log of direct fiscal costs in percent of GDP. Estimated coefficients reflect a log-linear relation.
- Table A1a (columns (1)–(5), N = 53/53/53/36/37):
  - Income per capita (t-1): -0.00300; -0.00291; 0.000124; -0.000202; -0.00138 (standard errors: (0.00179), (0.00181), (0.00157), (0.00205), (0.00209)).
  - Public debt/GDP (t-1): -0.00369; -0.00495; -0.00412; -0.00713; -0.00441 (standard errors: (0.00529), (0.00527), (0.00367), (0.00776), (0.00374)).
  - Current account/GDP (t-1): -0.0287; -0.0325; -0.0475*; -0.0516*; -0.0448 (standard errors: (0.0263), (0.0263), (0.0283), (0.0295), (0.0298)).
  - Banking supervision index: -0.443***; -0.428***; -0.403***; -0.545**; -0.615*** (standard errors: (0.125), (0.120), (0.132), (0.202), (0.177)).
  - Non-financial sector leverage (private credit/GDP, t-1): 0.00871** (0.00376) in column where included.
  - Size of financial sector (assets/GDP, t-1): 0.00771** (0.00320) in column where included.
  - Financial sector leverage (loans/deposits, t-1): -0.00139 (0.00133) where included.
  - Interconnectedness (offshore deposits/domestic deposits, t-1): 0.0230** (0.0110).
  - Interconnectedness (non-resident loans/GDP, t-1): 0.00704** (0.00338).
  - Constants: 2.504***; 2.515***; 2.857***; 2.911***; 3.068*** (standard errors reported).
  - R-sq: 0.309; 0.304; 0.270; 0.396; 0.403. F reported and p-values: p = 0.000417; 0.000421; 0.00692; 0.000202; 0.000205.
- Table A1b (columns (1)–(4), N = 46/46/34/35):
  - Income per capita (t-1): -0.00343; -0.00307; -0.00190; -0.00214 (se: (0.00229), (0.00224), (0.00217), (0.00241)).
  - Public debt/GDP (t-1): -0.00109; -0.00196; -0.00574; -0.00591 (se: (0.00517), (0.00505), (0.00691), (0.00349)).
  - Current account/GDP (t-1): -0.0437; -0.0474*; -0.0674**; -0.0558* (se: (0.0276), (0.0274), (0.0255), (0.0281)).
  - Banking supervision index: -0.320**; -0.315*; -0.410**; -0.520*** (se: (0.157), (0.160), (0.199), (0.179)).
  - Deposit insurance coverage (covered deposits/GDP per capita, t-1): -0.00186**; -0.00191**; -0.00348***; -0.00279** (se: (0.000899), (0.000906), (0.00125), (0.00132)).
  - Non-financial sector leverage: 0.00771* (0.00452) where included.
  - Size of financial sector: 0.00649 (0.00430) where included.
  - Interconnectedness (offshore deposits/domestic deposits, t-1): 0.0295*** (0.00971).
  - Interconnectedness (non-resident loans/GDP, t-1): 0.00666* (0.00348).
  - Constants and model fit: R-sq 0.373; 0.361; 0.554; 0.519. p = 0.00138; 0.00155; 0.00011; 0.00036.

### Extended Model: Conditional Correlations of Direct Fiscal Costs (Table A2)
- Dependent variable: log of direct fiscal costs in percent of GDP.
- Ten specifications (columns (1)–(10)), various sample sizes (N = 53/53/53/36/37/35/35/35/18/19).
- Key coefficients across specifications:
  - Income per capita (t-1): values include -0.00320*; -0.00312*; -0.000626; -0.00139; -0.00213; -0.00304; -0.00286; -0.000475; 0.00131; -0.000301 (standard errors reported).
  - Public debt/GDP (t-1): values include -0.00363; -0.00467; -0.00478; -0.00636; -0.00504; -0.00260; -0.00383; -0.00112; -0.0126*; -0.00389.
  - Current account/GDP (t-1): values include -0.0279; -0.0311; -0.0432; -0.0499*; -0.0413; 0.0218; 0.00422; -0.0178; 0.0300; -0.0276.
  - Banking supervision index: -0.461***; -0.448***; -0.479***; -0.582***; -0.665***; -0.145; -0.121; -0.159; -0.472; -0.396.
  - Non-financial sector leverage: 0.00726*; 0.00853** in columns where included.
  - Size of financial sector: 0.00636*; 0.00664** where included.
  - Financial sector leverage: -0.00136; 0.000204 where included.
  - Interconnectedness (offshore deposits/domestic deposits): 0.0288***; 0.00270.
  - Interconnectedness (non-resident loans/GDP): 0.00753**; 0.00542.
  - Bank guarantee (dummy): 0.380; 0.388; 0.625**; 0.743**; 0.615* in specified columns.
  - Forbearance (dummy): 0.568; 0.581; 0.530; 0.205; 0.116 where included.
  - Constants and fit: R-sq range 0.169 to 0.460; p-values vary across columns (detailed p reported per column).

### Extended Model: Direct Fiscal Costs During Recent Crisis (Table A3)
- Focus: recent crisis subsample (columns (1)–(4), N = 53/53/36/37).
- Income per capita (t-1): -0.00369**; -0.00410**; -0.000869; -0.00109.
- Public debt/GDP (t-1): -0.00469; -0.00662; -0.00631; -0.00417.
- Current account/GDP (t-1): -0.0256; -0.0299; -0.0496; -0.0439.
- Banking supervision index: -0.487***; -0.505***; -0.560**; -0.588***.
- Interaction: Banking supervision index (recent crisis in AMs): 0.368; 0.578; -0.0414; -0.0566 (noted per column).
- Non-financial sector leverage: 0.0115*** where included.
- Size of financial sector: 0.0104*** where included.
- Interconnectedness measures and interactions with recent crisis in AMs reported (coefficients and standard errors).
- Model fit: R-sq 0.321; 0.325; 0.406; 0.404. p-values reported.

### Extended Model for Direct Fiscal Costs: Accounting for NPLs (Table A4)
- Dependent variable as before; includes Peak NPLs (% of total loans).
- Columns (1)–(4), N = 53/53/36/37.
- Peak NPLs: 0.0238**; 0.0233**; 0.0296*; 0.0236* (standard errors: (0.0109), (0.0111), (0.0150), (0.0141)).
- Banking supervision index: -0.345**; -0.332**; -0.340; -0.449**.
- Current account/GDP (t-1): -0.0298; -0.0326; -0.0475*; -0.0418.
- Non-financial sector leverage: 0.00580* where included.
- Interconnectedness: 0.0272*** (offshore deposits/domestic deposits); 0.00724*** (non-resident loans/GDP).
- Bank guarantee (dummy): 0.370; 0.390; 0.737**; 0.645*.
- Constants and model fit: R-sq 0.379; 0.372; 0.517; 0.491. p-values reported.

### Extended Model: Generalized Methods of Moments and Reverse Causality (Table A5)
- GMM specifications (columns (1)–(4), N = 54/54/36/37).
- Income per capita (t-1): -0.00496***; -0.00432***; -0.000876; -0.000720.
- Public debt/GDP (t-1): 0.00141; 0.000672; -0.00901*; -0.00405*.
- Current account/GDP (t-1): -0.0270; -0.0318; -0.0343*; -0.0358.
- Banking supervision index: -0.434***; -0.436***; -0.747***; -0.693***.
- Non-financial sector leverage: 0.00964*** where included.
- Size of financial sector: 0.00791*** where included.
- Interconnectedness: 0.0304*** (offshore deposits/domestic deposits); 0.00605** (non-resident loans/GDP).
- Bank guarantee (dummy): 0.812**; 0.883***; 1.189***; 0.707***.
- Tests reported: difference-in-Sargan test and Hansen test values per column.

### Basic Model: Conditional Correlations of Overall Fiscal Costs (Table A6)
- Dependent variable: change in public debt ratios over [T-1; T+4], where T is the starting year of the banking crisis.
- Ten specifications (columns (1)–(10)), observations vary (e.g., 53, 53, 53, 38, 40, 44, 44, 43, 43, 35).
- Key coefficients:
  - Public Debt/GDP (t-1): -0.113; -0.119; -0.216*; 0.083; -0.199; -0.051; -0.058; -0.246*; 0.088; -0.222 (standard errors in brackets).
  - Change in Primary Balance (t, t+3): -0.905***; -0.972***; -1.046***; -1.073***; -1.011***; -1.190***; -1.252***; -1.120***; -1.301***; -0.992*** ([0.217], [0.205], [0.198], [0.180], [0.199], [0.233], [0.214], [0.255], [0.227], [0.324]).
  - Non-Financial Sector Leverage (Private Credit/GDP, t-1): 0.113**; 0.148** where included.
  - Size of Financial Sector (Assets/GDP, t-1): 0.095*; 0.136** where included.
  - Financial Sector Leverage (Loans/Deposits, t-1): 0.049***; 0.005 where included.
  - Banking Supervision Index: -6.596*; -6.256*; 0.171; -4.419; -1.158 in specified columns ([3.530], [3.563], ...).
  - Deposit Insurance Coverage: coefficients reported (e.g., 0.031) with brackets.
  - Constants reported (e.g., 13.813**; 15.225***; 19.121***; 14.661**; 27.165***; ...).
  - R Squared: range from 0.432 to 0.578 across specifications. F Statistics and Prob > F reported (all Prob > F = 0.000).

### Extended Model: Conditional Correlations of Overall Fiscal Costs (Table A7)
- Dependent variable: change in public debt ratios over [T-1; T+4].
- Eleven specifications (columns (1)–(11)), observations vary (e.g., 53, 53, 53, 53, 53, 53, 53, 53, 53, 38, 40).
- Key coefficients:
  - Public Debt/GDP (t-1): values include -0.124; -0.110; -0.109; -0.124; -0.116; -0.113; -0.258**; -0.218; -0.192; 0.052; -0.232.
  - Change in Primary Balance (t, t+3): -0.879***; -0.899***; -0.901***; -0.930***; -0.965***; -0.967***; -0.904***; -1.033***; -1.010***; -0.952***; -0.889***.
  - Non-Financial Sector Leverage: 0.072; 0.111*; 0.106* where included.
  - Size of Financial Sector: 0.044; 0.094; 0.086 where included.
  - Financial Sector Leverage: 0.037**; 0.052***; 0.045*** where included.
  - Bank Guarantee (Dummy): 9.223; 11.224*; 12.532**; 11.993*; 13.260* in specified columns.
  - Asset Purchases (Dummy) and Bank Recapitalization (Dummy) coefficients reported where included.
  - Constants and fit: R Squared range 0.463 to 0.524; Prob > F = 0.000 across specifications.

### Extended Model for Overall Fiscal Costs: Accounting for NPLs and Losses Imposed on Depositors (Table A8)
- Dependent variable: change in public debt ratios over [T-1; T+4]. NPLs = non-performing loans.
- Large set of specifications (columns (1)–(25)), observations vary (e.g., 52, 52, 52, 38, 40, ...).
- Key variables and representative coefficients (standard errors in brackets):
  - Public debt/GDP (change between t-7 and t-1): values include -0.103; -0.117; -0.213; 0.124; -0.173; ...; -0.263**; -0.214; ...; -0.110 (bracketed standard errors per column).
  - Change in primary balance (t, t+3): values include -1.036***; -1.098***; -1.138***; -1.222***; -1.150***; -1.240***; -1.304***; -1.205***; -1.310***; -1.034***; -1.030***; -1.034***; -1.042***; -1.068***; -1.096***; -1.102***; -1.036***; -1.130***; -1.139***; -1.107***; -1.221***; -1.197***; -1.037***; -1.162***; -1.121***.
  - Peak NPL (% total loans): coefficients reported across columns (e.g., 0.350; 0.366; 0.272; ...; 0.123) with bracketed standard errors.
  - Loss imposed on depositors (dummy): coefficients vary widely (e.g., -1.789; -0.151; -3.218; 17.919; 14.782; ...; 22.703**), with large standard errors ([11.882], [11.850], ...).
  - Changes in structural banking variables (t-7 to t-1):
    - Non-financial sector leverage: 0.113***; 0.164**; 0.067; 0.113***; 0.107** (brackets).
    - Size of financial sector: 0.107**; 0.152**; 0.055; 0.106**; 0.099**.
    - Financial sector leverage: 0.041***; 0.024; 0.022; 0.043**; 0.035**.
    - Interconnectedness variables and banking supervision index reported.
  - Policy dummies:
    - Bank guarantee (dummy): 10.851; 12.221*; 14.659**; 16.307**; 16.914** in specified columns.
    - Asset purchases (dummy): coefficients reported (e.g., 0.492; 0.469; 1.110; 0.340; -3.813).
    - Bank recapitalization (dummy): 4.458; 5.393; 12.317; 14.039; 23.176*.
  - Model fit: Observations vary widely; R-squared range 0.484 to 0.637; Adjusted R-squared and F-statistics reported per specification. Robust standard errors in brackets. Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

### Extended Model for Overall Fiscal Costs: Generalized Method of Moments and Reverse Causality (Table A9)
- Dependent variable: change in public debt ratios over [T-1; T+4]. GMM specifications.
- Fifteen specifications (columns (1)–(15)), observations vary (e.g., 51, 51, 51, 51, 51, 51, 51, 51, 51, 36, 36, 36, 38, 38, 38).
- Key coefficients:
  - Public debt/GDP (t-1): -0.122; -0.082; -0.077; -0.099; -0.071; -0.050; -0.291***; -0.197**; -0.201***; 0.097; 0.179**; 0.183**; -0.107; -0.112; -0.127 (bracketed standard errors).
  - Change in primary balance (t, t+3): -1.043***; -1.004***; -1.048***; -1.105***; -1.035***; -1.130***; -0.993***; -1.115***; -1.132***; -1.017***; -1.138***; -1.069***; -1.181***; -1.148***; -1.145*** (brackets).
  - Other structural and policy variables: coefficients and standard errors reported; Asset purchases (dummy) and Bank recapitalization (dummy) show significance in some columns (e.g., Asset purchases 10.506** in column (2); Bank recapitalization 9.290**; 9.246*; 13.908*** in columns shown).
  - Model fit: R-squared range 0.433 to 0.545; Wald chi2 and Prob > chi2 reported (Prob > chi2 = 0.000 across specifications). Robust standard errors in brackets.

### Data and Data Sources (Table A10)
- Direct fiscal costs: Direct fiscal costs as a share of GDP — Laeven and Valencia (2013).
- Income per capita: GDP per capita — WEO.
- Public debt and overall fiscal costs: Public debt as a share of GDP — Abbas and others (2010) spliced with general government gross debt/GDP from Mauro and others (2013).
- Current account balance: Current account balance as a share of GDP — WEO.
- Banking sector supervision: Banking sector supervision index — Abiad and others (2008).
- Deposit insurance coverage: Deposit insurance coverage as a share of GDP per capita — Laeven and Valencia (2013).
- Peak NPLs: Peak non-performing loans (NPL) as a share of total loans — Laeven and Valencia (2013).
- Financial sector leverage: Private credit by deposit money banks as a share of demand, time and saving deposits in deposit money banks — World Bank (2012).
- Size of financial sector: Claims on domestic real nonfinancial sector by deposit money banks as a share of GDP — World Bank (2012).
- Non-financial sector leverage: Private credit by deposit money banks as a share of GDP — World Bank (2012).
- Interconnectedness 1: Offshore bank loans (amount outstanding) as a share of GDP — World Bank (2012).
- Interconnectedness 2: Offshore bank deposits relative as a share of domestic deposits — World Bank (2012).
- Policy response variables: Bank guarantee and forbearance dummies, asset purchases, bank recapitalization, and losses imposed on depositors — Laeven and Valencia (2013).

### Systemic Banking Crises and Recent Borderline Crisis Episodes (Table A11)
- Source: Laeven and Valencia (2013).
- Country-year listings (selected examples as presented):
  - Argentina: 1980, 1989, 1995, 2001
  - Austria: 2008
  - Belgium: 2008
  - Bolivia: 1994
  - Brazil: 1990, 1994
  - Bulgaria: 1996
  - Chile: 1981
  - Colombia: 1982, 1998
  - Côte d’Ivoire: 1988
  - Croatia: 1998
  - Czech Republic: 1996
  - Denmark: 2008
  - Dominican Rep: 2003
  - Ecuador: 1998
  - Estonia: 1992
  - Finland: 1991
  - France: 2008
  - Germany: 2008
  - Ghana: 1982
  - Greece: 2008
  - Hungary: 2008
  - Iceland: 2008
  - Indonesia: 1997
  - Ireland: 2008
  - Italy: 2008
  - Jamaica: 1996
  - Japan: 1997
  - Kazakhstan: 2008
  - Korea: 1997
  - Latvia: 1995, 2008
  - Lithuania: 1995
  - Luxembourg: 2008
  - Malaysia: 1997
  - Mexico: 1994
  - Mongolia: 2008
  - Netherlands: 2008
  - Nicaragua: 2000
  - Nigeria: 2009
  - Norway: 1991
  - Paraguay: 1995
  - Philippines: 1997
  - Portugal: 2008
  - Russia: 1998, 2008
  - Slovenia: 2008
  - Spain: 2008
  - Sri Lanka: 1989
  - Sweden: 1991, 2008
  - Switzerland: 2008
  - Thailand: 1997
  - Turkey: 2000
  - Ukraine: 1998, 2008
  - United Kingdom: 2007
  - United States: 2007
  - Uruguay: 2002
  - Venezuela: 1994
  - Vietnam: 1997
  - (Full list presented in table format in source.)

*Source: _wp15166 - APPENDIX TABLES*

### REFERENCES

### _wp15166 - REFERENCES

### Banking crises, systemic risk, and crisis resolution
- Allen, F. and D. Gale (2003), “Liquidity, Asset Prices and Systemic Risk,” Bank for International Settlements, Proceedings of the Third Joint Central Bank Research Conference.
- Boissay, F., Collard, F., and F. Smets (2013), “Booms and Systemic Banking Crises,” ECB Working Paper Series, 1514.
- Claessens, S., D. Klingebiel, and L. Laeven (2005), “Crisis Resolution, Policies, and Institutions: Empirical Evidence,” in Patrick Honohan and Luc Laeven, Systemic Financial Crises: Containment and Resolution (Cambridge University Press).
- Claessens, S., C. Pazarbasioglu, L. Laeven, M. Dobler, F. Valencia, O. Nedelescu, and K. Seal (2011), “Crisis Management and Resolution: Early Lessons from the Financial Crisis,” IMF Staff Discussion Note, 11/05.
- Frydl, E. (1999), “The Length and Cost of Banking Crises,” IMF Working Paper, 99/30.
- Frydl, E. and M. Quintyn (2000), “The Benefits and Costs of Intervening in Banking Crises,” IMF Working Paper, 00/147.
- Honohan, P. and D. Klingebiel (2003), “The Fiscal Cost Implications of an Accommodating Approach to Banking Crises,” Journal of Banking and Finance, 27: 1539–60.
- Hoggarth, G., R. Reis, and V. Saporta (2002), “Output Costs of Banking System Instability: Some Empirical Evidence,” Journal of Banking and Finance, 26: 825–55.
- Laeven, L. and F. Valencia (2008), “Systemic Banking Crises: A New Database,” IMF Working Paper, 08/224.
- Laeven, L. and F. Valencia (2010), “Resolution of Banking Crises: The Good, the Bad, and the Ugly,” IMF Working Paper, 10/44.
- Laeven, L. and F. Valencia (2012), “Systemic Banking Crises Database: An Update,” IMF Working Paper, 12/163.
- Laeven, L. and F. Valencia (2013), “Systemic Banking Crises Database,” IMF Economic Review, 61(2).
- Reinhart, C. M. and K. S. Rogoff (2008), “Banking Crises: An Equal Opportunity Menace,” NBER Working Paper, 14587.
- Reinhart, C. M. and K. S. Rogoff (2013), “Banking Crises: An Equal Opportunity Menace,” Journal of Banking & Finance, 37(11): 4557–73.
- Deutsche Bank (2013), “Financial Crises: Past and Present,” Global Economic Perspectives, Deutsche Bank Market Research.
- International Monetary Fund (1998), “Financial Crises: Characteristics and Indicators of Vulnerability,” World Economic Outlook.
- International Monetary Fund (2003), “Managing Systemic Banking Crises,” Occasional Paper 224.

### Fiscal costs, public debt, and macro-fiscal implications
- Abbas, S.M.A., N. Belhocine, A. El-Ganainy, and M. Horton (2011), “Historical Patterns and Dynamics of Public Debt—Evidence From a New Database,” IMF Economic Review, 59(4): 717-42.
- European Commission (2009), “The fiscal Costs of Financial Crises: Past Evidence and Implications for Today's Crisis,” Public Finances in EMU.
- International Monetary Fund (2009), “The State of Public Finances: Outlook and Medium-Term Policies after the 2008 Crisis,” Companion Paper.
- International Monetary Fund (2015), “From Banking to Sovereign Stress: Implications for Public Debt,” Policy Paper.
- Mauro, P., R. Romeu, A. Binder, and A. Zaman (2013), “A Modern History of Fiscal Prudence and Profligacy,” IMF Working Paper, 13/05.

### Financial reforms, regulation, and macroprudential policy
- Abiad, A., E. Detragiache, and T. Tressel (2008), “A New Database of Financial Reforms,” IMF Working Paper, 08/266.
- International Monetary Fund (2011), “Towards Operationalizing Macroprudential Policies: When to Act?,” September Global Financial Stability Report.
- Demirgüç-Kunt, A. and H. Huizinga (2004), “Market Discipline and Deposit Insurance,” Journal of Monetary Economics, 51, 375-99.
- Demirgüç-Kunt, A. and E. Detragiache (1998), “The Determinants of Banking Crises in Developing and Developed Countries,” IMF Staff Papers, 45: 81–109.
- Claessens, S., C. Pazarbasioglu, L. Laeven, M. Dobler, F. Valencia, O. Nedelescu, and K. Seal (2011), “Crisis Management and Resolution: Early Lessons from the Financial Crisis,” IMF Staff Discussion Note, 11/05.

### Credit, leverage, and cross-border banking linkages
- Djankov, S., C. McLiesh, and A. Shleifer (2007), “Private Credit in 129 Countries” Journal of Financial Economics, 84 (May): 299–329.
- Kalemli-Ozcan, S., B. Sorensen and S. Yesiltas (2012), “Leverage across Firms, Banks and Countries,” Journal of International Economics, 88(22): 284–298.
- Čihák M., R. Scuzzarella, and S. Munoz (2011), “The Bright and the Dark Side of Cross-Border Banking Linkages,” IMF Working Paper 11/186.
- Kaminsky, G. and C.M. Reinhart (1999), “The Twin Crises: The Causes of Banking and Balance-of-Payments Problems,” American Economic Review, 89(3): 473–500.
- Inter-American Development Bank (2005), “Unlocking Credit. The Quest for Deep and Stable Bank Lending,” Economic and Social Progress Report.

### Historical narratives and country studies
- Gourinchas, P. and M. Obstfeld (2012), “Stories of the Twentieth Century for the Twenty-First,” American Economic Journal: Macroeconomics, 4(1): 226–65.
- Lane, P. (2011), “The Irish Crisis,” IIIS Discussion Paper, 356.

### Datasets and data resources
- Laeven, L. and F. Valencia (2008), “Systemic Banking Crises: A New Database,” IMF Working Paper, 08/224.
- Laeven, L. and F. Valencia (2012), “Systemic Banking Crises Database: An Update,” IMF Working Paper, 12/163.
- World Bank (2012), “Financial Structure Dataset,” September: siteresources.worldbank.org/.../FinStructure_2012_September_Update2x.

*Referenced in: _wp15166 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15166.pdf_