## 1. Output Losses versus Initial Conditions

## Source details

**Canonical URL:** [1. Output Losses versus Initial Conditions](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09245.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09245.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09245.pdf.json)

---

### Scope and focus
- Examines "Output Losses versus Initial Conditions" as a unit within the broader work.
- Sample: 88 banking crises.
- Precrisis trend estimation: linear trend through actual output series during a 7-year precrisis period that ends three years before the crisis (extended to T-20 when necessary).
- Post-crisis window for medium-run assessment: seven years.
- Decompositions use a Cobb-Douglas production function with employment share α = 0.65.
- Factor and demand-side decompositions and OLS/Bayesian model averaging methods employed to assess correlates of output/growth losses.

### Main stylized facts (aggregate outcomes)
- On average, seven years after the crisis, output has declined relative to trend by close to 10 percent.
- The 90-percent confidence band indicates the average decline is statistically significant.
- Losses following currency crises are much smaller, around one third of the average loss associated with banking crises.
- Growth dynamics:
  - Annual growth falls substantially below precrisis trend during the first two years.
  - Growth is then statistically indistinguishable from precrisis trend thereafter.
  - Four-year average of growth ending in the seventh year after the crisis: mean difference with respect to precrisis trend growth rate of only -0.2 percentage points per year, with a standard error of 0.5 percentage points.
- Cross-case variation:
  - Mean change in output relative to trend following banking crises: -10 percent.
  - Middle 50 percent (interquartile range) of cases: -26 percent to +6 percent.
  - More than a quarter of cases ultimately exceeded the precrisis trend.
  - For growth deviations, middle 50 percent ranged from -2.8 percentage points to +1.7 percent.
- Historical benchmark: global output 27 percent below precrisis trend by 1936 during the Great Depression; global growth rate returned to precrisis trend by 1934.

### Decompositions: why aggregate output does not recover
- Theoretical channels considered:
  - Labor-force participation: ambiguous effects (discouraged-workers vs. second-earner entry).
  - Employment rate: potential persistent rise in structural unemployment; hysteresis from long unemployment spells.
  - Capital: slower accumulation due to tighter financing, lower asset prices, higher uncertainty.
  - Total-factor productivity (TFP): ambiguous (allocation/R&D losses vs. cleansing effects).
- Empirical factor decomposition (sample for factor decompositions: 27 observations):
  - Medium-run GDP per capita losses attributable to roughly equal losses in:
    - Employment rate
    - Capital-to-labor ratio
    - Total-factor productivity
  - TFP: significant initial decline; moves closer to precrisis trend toward end of seven-year horizon; medium-run TFP loss accounts for about one-third of total output loss but is not statistically significant at T+7 (though significant in short run).
  - Employment rate: initial loss persists into medium run.
  - Capital-to-labor ratio: losses worsen steadily over time.
- Demand-side decomposition findings:
  - Investment declines by around 30 percent relative to precrisis trend (large and significant).
  - Consumption declines by around 15 percent relative to precrisis trend (notable and significant).
  - Net exports improve overall relative to trend, partially offsetting losses.
- Interpretation:
  - Medium-term output shortfall primarily reflects enduring reductions in capital per worker, higher unemployment, and lower productivity that do not typically return to precrisis trends within seven years.

### What initial conditions are associated with larger medium-term output losses
- Empirical strategy: small-scale OLS, large-scale OLS, and Bayesian model averaging (BMA); typical controls include precrisis output deviation and first-year output change.
- Robust correlates (precrisis conditions):
  - Short-run crisis severity:
    - First-year output change is a strong predictor: a one-percentage point fall in output relative to trend in the first year is associated with a 1.1–1.8 percentage point gap between output and precrisis trend by T+7 (Table 1, Row 20).
  - Precrisis output position:
    - A depressed level of output relative to trend before the crisis associated with significantly larger medium-run output loss (Table 1, Row 19).
  - Precrisis investment share:
    - High precrisis investment-to-GDP ratio (average during last three precrisis years) strongly associated with larger output losses (Table 1, Row 1).
    - Investment gap (deviation from historical average) not statistically significant (Table 1, Row 2).
  - Macroeconomic imbalances:
    - Small-scale regressions: larger current account deficits, rising inflation, and deteriorating fiscal balance before the crisis associated with larger output losses (Table 1, Rows 3, 6, 8).
    - BMA: strong evidence mainly for rising inflation before the crisis.
  - Income level:
    - Post-crisis output losses not significantly correlated with level of income (Table 1, Row 11).
  - Financial deepening/openness:
    - Mixed evidence: some association of higher precrisis financial development with larger losses; financial openness associated with smaller losses in small-scale regressions (Table 1, Row 17) but weaker in broader specifications.
  - Twin crises:
    - Joint occurrence of currency and banking crisis (“twin crisis”) robustly associated with larger output losses (Table 1, Row 14).
- Interpreted mechanisms:
  - Large initial shocks produce persistent losses via bankruptcies, fire sales, costly capital rebuilding, impaired financial intermediation, and labor/product market rigidities.
  - High precrisis investment share correlation with larger losses may reflect unwinding of excessive or wasteful investment.

### Post-crisis policies and conditions associated with smaller medium-term output losses
- Measurement: policy responses measured over crisis year and following three years; structural reform measured as change from T to T+7.
- Short-run macroeconomic policies:
  - Fiscal stimulus:
    - Strong short-run fiscal response (growth in real government consumption) significantly associated with smaller medium-run output losses (Table 2, Row 1).
    - Quantified implication: raising government consumption by one percent of GDP associated with a reduction in medium-run output loss by about 1.5 percentage points.
  - Monetary policy:
    - Evidence mixed. Decline in real lending rates associated with smaller output losses in some specifications (Table 2, Row 2).
  - Real exchange rate:
    - Mixed evidence that depreciations are associated with smaller output losses (Table 2, Row 3).
- Structural reforms:
  - Mixed evidence overall.
  - Capital-account liberalization change correlated with smaller output losses in small-scale regressions (Table 2, Row 4); significance weaker in larger specifications.
  - Domestic financial liberalization change positively associated with smaller losses in some small-scale regressions (Table 2, Row 5).
  - Trade liberalization not significantly related (Table 2, Row 6).
  - Improvements in government efficiency show some positive evidence in some specifications (Table 2, Row 7).
  - Caveats: measurement issues, timing/sequencing, and long gestation periods for reforms may attenuate observed effects.
- External environment:
  - Adverse external demand shocks during post-crisis period significantly related to larger domestic output losses (Table 2, Row 9).
  - Some evidence that higher global short-term interest rates associated with larger output losses (Table 2, Row 8).
- Interpretation:
  - Expansionary short-run fiscal and monetary policies are associated with smaller medium-term output losses, consistent with counter-cyclical policies dampening path-dependence.
  - Structural reform effects are mixed and harder to detect.
  - Favorable global conditions help mitigate medium-term losses.

### Factors associated with medium-term growth losses
- Definition: medium-run growth loss defined as average growth over the four years starting in T+4 relative to precrisis trend (five-year average growth ending in T+8 less precrisis trend used in regressions).
- Predictability:
  - Explaining postcrisis medium-run growth relative to precrisis trend is harder than explaining postcrisis level of output relative to trend; R-squared values lower for growth regressions.
  - First-year output change and precrisis output deviation are weaker predictors of growth loss than of output level at T+7.
  - Precrisis investment share remains a strong predictor of growth losses (Table 3, Row 1).
- Post-crisis policies and conditions:
  - Short-run fiscal and monetary stimulus correlated with smaller subsequent growth losses (Table 4, Rows 1–2).
  - Real exchange rate depreciation associated with smaller growth losses in some specifications (Table 4, Row 3).
  - Mixed evidence on structural reforms and weaker evidence on external conditions relative to output-level regressions.
- Overall interpretation:
  - Output tends to be trend-stationary after banking crises (growth returns to precrisis rate on average), making medium-run growth outcomes less predictable from initial conditions.

### Robustness and alternative measures
- Four alternatives to baseline precrisis trend considered (differences in precrisis window and use of IMF real-time forecasts).
- Loss estimates using Alternatives 1–3 highly correlated with baseline; 90-percent confidence bands overlap substantially with baseline.
- Alternative 4 (IMF desk forecasts) yields greater average output and growth losses than baseline because forecasts were, on average, more optimistic than baseline precrisis trend.
- Regressions omitting the short-run crisis severity variable:
  - Coefficients broadly similar to baseline, but R-squared declines substantially in many cases (in some cases by more than one-half).
  - Selected coefficients reported (Table A2 examples):
    - (1) investment/GDP: -1.242***  [-4.195]
    - (2) investment/GDP gap: 0.687  [1.597]
    - (3) current account/GDP: 0.677*  [1.848]
    - (4) current account/GDP gap: 1.245*  [1.814]
    - (5) inflation: 0.145*  [1.781]
    - (6) inflation gap: -0.246***  [-3.004]
    - (14) currency crisis: -0.201***  [-2.729]
    - (19) pre-crisis output: 1.695***  [4.127]
  - Number of observations and R-squared vary across specifications; examples include observations 88, 85, 80, and R-squared 0.214, 0.294, 0.282 among others.

### Conclusions and policy implications
- Aggregate findings:
  - Using a sample of 88 banking crises over the past four decades, output on average remains below precrisis trend seven years after crisis onset; trend growth rate itself appears largely unaffected and tends to return to precrisis rate.
  - Depressed path of output results from reductions of roughly equal proportions in employment rate, capital-to-labor ratio, and total-factor productivity.
  - Large variation across countries: about one quarter of cases outperform precrisis expectations.
- Robust correlates of larger medium-run output losses:
  - Larger first-year output decline.
  - Joint banking-and-currency crises.
  - High precrisis investment share.
- Mitigation evidence:
  - Short-run fiscal and monetary stimulus associated with smaller medium-run deviations of output and growth.
  - Mixed evidence on the role of structural reforms implemented post-crisis.
  - External conditions and global policy environment matter for medium-term outcomes.
- Policy implication:
  - Policies implemented in the aftermath of a crisis—particularly counter-cyclical fiscal and monetary actions—and favorable global conditions can help mitigate medium-term output losses, although causal interpretation is limited by contemporaneity and endogeneity concerns.

*Source: _wp09245 - 1. Output Losses versus Initial Conditions (section and referenced figures/tables as provided in the source PDF).*

### 1. Output Losses versus Initial Conditions .............................................................................

### 1. Output Losses versus Initial Conditions

### Scope and focus
- Examines "Output Losses versus Initial Conditions" as a unit within the broader work.
- Positioned in the document such that the section begins on page 17.

### Related sections in the same chapter (contextual linkage)
- 2. Output Losses versus Post-Crisis Conditions and Policies — page 23
- 3. Growth Losses versus Initial Conditions — page 26
- 4. Growth Losses versus Post-Crisis Conditions and Policies — page 27

### Figures referenced that directly relate to this content unit
- Figure 7: Output Evolution and Precrisis Investment Share — listed in the Figures section
- Figure 8: Output Evolution and Precrisis Income Level — listed in the Figures section

### Structural role within the chapter
- Part of a sequence analyzing output and growth losses across dimensions of initial conditions and post-crisis policies.
- Preceded by methodological and illustrative material, including:
  - Figure 2: Output Loss Methodology Example (Republic of Korea 1997)
  - Figure 5: Output Decomposition
  - Figure 6: Demand-Side Decomposition
- Supported by empirical comparisons across crisis types (see Figures 3 and 4).

*Source: _wp09245 - 1. Output Losses versus Initial Conditions (page and figure listings as provided in the source PDF).*

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

### _wp09245 - References..............................................................................................................

### I. INTRODUCTION
- Context: global economy beginning to recover from the most severe financial crisis since the Great Depression and the deepest recession since World War II; financial systems remain impaired and domestic and external imbalances persist.
- Paper scope:
  - Examines medium-term macroeconomic performance following past banking crises in advanced, emerging, and developing economies over the past 40 years.
  - Uses a sample of 88 banking crises.
  - Focuses on both “output loss” (post-crisis level of output vs. precrisis trend) and “growth loss” (post-crisis growth rate vs. precrisis trend growth rate).
  - Precrisis trend estimated over a 7-year precrisis period that ends three years before the crisis (extended to T-20 when necessary).
  - Decomposes medium-term dynamics into factor components (capital, employment, labor-force participation, total-factor productivity) and demand-side factors (consumption, investment, exports, imports).
  - Assesses correlation of output/growth losses with initial conditions and policy responses.

- Key methodological notes:
  - Post-crisis window: seven years.
  - Precrisis trend estimation: linear trend through actual output series during 7-year precrisis period ending three years before crisis; extended back when trend negative.
  - Cobb-Douglas production function used for factor decompositions with employment share α = 0.65.

### II. DOES OUTPUT RECOVER OVER THE MEDIUM RUN?
- Main stylized facts:
  - On average, output does not rebound to its precrisis trend over the medium run.
  - Seven years after the crisis, output has declined relative to trend by close to 10 percent on average.
  - The 90-percent confidence band indicates the average decline is statistically significant.
  - Losses following currency crises are much smaller, around one third of the average loss associated with banking crises.
  - Growth rates: annual growth falls substantially below precrisis trend during the first two years, then is statistically indistinguishable from precrisis trend thereafter.
    - Four-year average of growth ending in the seventh year after the crisis: mean difference with respect to precrisis trend growth rate of only -0.2 percentage points per year, with a standard error of 0.5 percentage points.
  - Variation across cases:
    - Mean change in output relative to trend following banking crises: -10 percent.
    - Middle 50 percent (interquartile range) of cases: -26 percent to +6 percent.
    - More than a quarter of cases ultimately exceeded the precrisis trend.
    - For growth deviations, middle 50 percent ranged from -2.8 percentage points to +1.7 percent.
  - Historical benchmark (Great Depression): global output 27 percent below precrisis trend by 1936; global growth rate returned to precrisis trend by 1934.

### III. DECOMPOSITIONS: WHY DOES AGGREGATE OUTPUT NOT RECOVER?
- Theoretical channels considered:
  - Labor-force participation: ambiguous (discouraged workers vs. second-earner entry).
  - Employment rate: potential persistent rise in structural unemployment; hysteresis from long unemployment spells.
  - Capital: slower capital accumulation due to tighter financing, lower asset prices, higher uncertainty.
  - Total-factor productivity (TFP): ambiguous; negative effects from impaired allocation and reduced R&D vs. positive “cleansing” effects.
- Empirical decompositions (sample for factor decompositions: 27 observations):
  - Cobb-Douglas production function with α = 0.65; logs used for exact add-up of factor contributions.
  - Factor-component findings (percent of precrisis trend, mean differences reported with 90-percent confidence intervals in figures):
    - Medium-run losses in GDP per capita can be attributed to roughly equal losses in:
      - Employment rate
      - Capital-to-labor ratio
      - Total-factor productivity
    - TFP: significant initial decline; moves closer to precrisis trend toward end of seven-year horizon; medium-run TFP loss accounts for about one-third of total output loss but is not statistically significant at T+7 (though significant in short run).
    - Employment rate: initial loss persists into medium run.
    - Capital-to-labor ratio: losses worsen steadily over time.
  - Demand-side decomposition:
    - Investment declines by around 30 percent relative to precrisis trend (large and significant).
    - Consumption declines by around 15 percent relative to precrisis trend (notable and significant).
    - Net exports improve overall relative to trend, partially offsetting losses.
- Interpretation:
  - Medium-term output shortfall primarily reflects enduring reductions in capital per worker, higher unemployment, and lower productivity that do not typically return to precrisis trends within seven years.

### IV. WHAT FACTORS ARE ASSOCIATED WITH MEDIUM-TERM OUTPUT LOSSES?
- Empirical strategy:
  - Small-scale OLS regressions including one or two variables plus controls.
  - Large-scale OLS including all factors simultaneously.
  - Bayesian model averaging (BMA) to assess robustness and inclusion probabilities (variables deemed “effective” if inclusion probability > 50 percent).
  - All regressions typically control for: precrisis output deviation and first-year output change.
- Key predictors (precrisis conditions):
  - Severity of short-run crisis:
    - First-year output change is a strong predictor: a one-percentage point fall in output relative to trend in the first year is associated with a 1.1–1.8 percentage point gap between output and precrisis trend by T+7 (Table 1, Row 20).
  - Precrisis output position:
    - A depressed level of output relative to trend before the crisis associated with significantly larger medium-run output loss (Table 1, Row 19).
  - Precrisis investment share:
    - High precrisis investment-to-GDP ratio (average during last three precrisis years) strongly associated with larger output losses (Table 1, Row 1).
    - Investment gap (deviation from historical average) not statistically significant (Table 1, Row 2).
    - Correlation robust to controlling for current account balance.
  - Macroeconomic imbalances (mixed evidence):
    - Small-scale regressions suggest larger current account deficits, rising inflation, and deteriorating fiscal balance before the crisis associated with larger output losses (Table 1, Rows 3, 6, 8).
    - BMA suggests strong evidence mainly for rising inflation before the crisis.
  - Income level:
    - Post-crisis output losses not significantly correlated with level of income (Table 1, Row 11); output evolution similar across income categories.
  - Financial deepening/openness:
    - Mixed evidence: some association of higher precrisis financial development with larger losses; financial openness associated with smaller losses in small-scale regressions (Table 1, Row 17), weaker in broader specifications.
  - Twin crises:
    - Joint occurrence of currency and banking crisis (“twin crisis”) robustly associated with larger output losses (Table 1, Row 14).
- Interpreted mechanisms:
  - Large initial shocks lead to persistent losses due to bankruptcies, fire sales, costly capital rebuilding, impaired financial intermediation, and labor/product market rigidities.
  - The strong correlation between high precrisis investment share and post-crisis losses may reflect unwinding of excessive or wasteful investment; further investigation required.

### V. AFTER THE CRISIS: WHAT IS ASSOCIATED WITH SMALLER OUTPUT LOSSES?
- Focus: post-crisis policies and external conditions (measured over crisis year and following three years for macro policy responses; structural reform measured as change from T to T+7).
- Short-run macroeconomic policies:
  - Fiscal stimulus:
    - Strong short-run fiscal response (growth in real government consumption) significantly associated with smaller medium-run output losses (Table 2, Row 1).
    - Quantified implication: raising government consumption by one percent of GDP associated with a reduction in medium-run output loss by about 1.5 percentage points.
  - Monetary policy:
    - Evidence mixed. Decline in real lending rates associated with smaller output losses in some specifications (Table 2, Row 2).
  - Real exchange rate:
    - Mixed evidence that depreciations are associated with smaller output losses (Table 2, Row 3).
- Structural reforms:
  - Mixed evidence overall.
  - Capital-account liberalization change correlated with smaller output losses in small-scale regressions (Table 2, Row 4); significance weaker in larger specifications.
  - Domestic financial liberalization change positively associated with smaller losses in some small-scale regressions (Table 2, Row 5).
  - Trade liberalization not significantly related (Table 2, Row 6).
  - Improvements in government efficiency show some positive evidence in some specifications (Table 2, Row 7).
  - Caveats: measurement issues, timing/sequencing, and long gestation periods for reforms may attenuate observed effects.
- External environment:
  - Adverse external demand shocks during post-crisis period significantly related to larger domestic output losses (Table 2, Row 9).
  - Some evidence that higher global short-term interest rates associated with larger output losses (Table 2, Row 8).
- Interpretation:
  - Expansionary short-run fiscal and monetary policies are associated with smaller medium-term output losses, consistent with counter-cyclical policies dampening path-dependence.
  - Structural reform effects are mixed and harder to detect given measurement and timing issues.
  - Favorable global conditions help mitigate medium-term losses.

### VI. WHAT FACTORS ARE ASSOCIATED WITH MEDIUM-TERM GROWTH LOSSES?
- Focus: medium-run growth loss defined as average growth over the four years starting in T+4 relative to precrisis trend (five-year average growth ending in T+8 less precrisis trend used in regressions).
- Predictability:
  - Explaining postcrisis medium-run growth relative to precrisis trend is harder than explaining postcrisis level of output relative to trend; R-squared values lower for growth regressions.
  - First-year output change and precrisis output deviation are weaker predictors of growth loss than of output level at T+7.
  - Precrisis investment share remains a strong predictor of growth losses (Table 3, Row 1).
- Post-crisis policies and conditions:
  - Short-run fiscal and monetary stimulus correlated with smaller subsequent growth losses (Table 4, Rows 1–2).
  - Real exchange rate depreciation associated with smaller growth losses in some specifications (Table 4, Row 3).
  - Mixed evidence on structural reforms and weaker evidence on external conditions relative to output-level regressions.
- Overall interpretation:
  - Output tends to be trend-stationary after banking crises (growth returns to precrisis rate on average), making medium-run growth outcomes less predictable from initial conditions.

### VII. CONCLUSIONS
- Aggregate findings:
  - Using a sample of 88 banking crises over the past four decades, output on average remains below precrisis trend seven years after crisis onset; trend growth rate itself appears largely unaffected and tends to return to precrisis rate.
  - Depressed path of output results from reductions of roughly equal proportions in employment rate, capital-to-labor ratio, and total-factor productivity.
  - Large variation across countries: about one quarter of cases outperform precrisis expectations.
- Predictors and mitigation:
  - Robust correlates of larger medium-run output losses: larger first-year output decline, joint banking-and-currency crises, and high precrisis investment share.
  - Short-run fiscal and monetary stimulus associated with smaller medium-run deviations of output and growth; mixed evidence on the role of structural reforms implemented post-crisis.
  - External conditions and global policy environment matter for medium-term outcomes.
- Policy implication:
  - Evidence is consistent with the notion that policies implemented in the aftermath of a crisis—particularly counter-cyclical fiscal and monetary actions—and favorable global conditions can help mitigate medium-term output losses, although causal interpretation is limited by contemporaneity and endogeneity concerns.

*Source: IMF working paper content unit _wp09245 - References.*

### References

### References

### Bibliographic Sources Cited
- Abiad, Abdul, Daniel Leigh, and Ashoka Mody, 2009, “Financial Integration, Capital Mobility, and Income Convergence,” Economic Policy, Vol. 24, No. 58, pp. 241–305.
- Aghion, Philippe, and Gilles Saint-Paul, 1998, “On the Virtue of Bad Times: An Analysis of the Interaction Between Economic Fluctuations and Productivity Growth,” Macroeconomic Dynamics, Vol. 2, No. 3, pp. 322–44.
- Angkinand, Apanard P., 2008, “Output Loss and Recovery from Banking and Currency Crises: Estimation Issues,” available at http://ssrn.com/abstract=1320730.
- Balakrishnan, Ravi, and Claudio Michelacci, 2001, “Unemployment Dynamics Across OECD Countries,” European Economic Review, Vol. 45, No. 1, pp. 135–65.
- Bassanini, Andrea, and Romain Duval, 2006, “The Determinants of Unemployment across OECD Countries: Reassessing the Role of Policies and Institutions,” OECD Economic Studies, No. 42, pp. 7–86.
- Bernanke, Ben S., 1983, “Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression,” American Economic Review, Vol. 73 (June), pp. 257–76.
- Bernanke, Ben S., and Alan S. Blinder, 1988, “Credit, Money, and Aggregate Demand,” American Economic Review, Vol. 78 (May), pp. 435–39.
- Bernanke, Ben S., and Mark Gertler, 1989, “Agency Costs, Net Worth, and Business Fluctuations,” American Economic Review, Vol. 79 (March), pp. 14–31.
- Bernanke, Ben S., 1995, “Inside the Black Box: The Credit Channel of Monetary Policy Transmission,” Journal of Economic Perspectives, Vol. 9 (Autumn), pp. 27–48.
- Blanchard, Olivier, and Justin Wolfers, 2000, “The Role of Shocks and Institutions in the Rise of European Unemployment: The Aggregate Evidence,” The Economic Journal, Vol. 110 (March), pp. 1–33.
- Blanchard, Olivier, and Pedro Portugal, 2001, “What Hides Behind an Unemployment Rate: Comparing Portuguese and U.S. Labor Markets,” American Economic Review, Vol. 91, No. 1, pp. 187–207.
- Bordo, Michael, 2006, “Sudden Stops, Financial Crises, and Original Sin in Emerging Countries: Déjà vu?” NBER Working Paper No. 12393.
- Boyd, John H., Sungkyu Kwak, and Bruce Smith, 2005, “The Real Output Losses Associated with Modern Banking Crises,” Journal of Money, Credit, and Banking, Vol. 37, No. 6, pp. 977–99.
- Caballero, Ricardo, and Mohammed Hammour, 1994, “The Cleansing Effect of Recessions,” American Economic Review, Vol. 84, No. 5, pp. 1350–68.
- Calvo, Guillermo A., Alejandro Izquierdo, and Luis-Fernando Mejía, 2008, “Systemic Sudden Stops: The Relevance of Balance-Sheet Effects and Financial Integration,” NBER Working Paper No. 14026.
- Cambell, John Y., and N. Gregory Mankiw, 1987, “Are Output Fluctuations Transitory?” The Quarterly Journal of Economics, November 1987.
- Cerra, Valerie, and Sweta Saxena, 2008, “Growth Dynamics: The Myth of Economic Recovery,” American Economic Review, Vol. 98, No. 1, pp. 439–57.
- Coulibaly, Brahima, and Jonathan Millar, 2008, “The Asian Financial Crisis, Uphill Flow of Capital, and Global Imbalances: Evidence from a Micro Study,” FRB International Finance Discussion Paper No. 942.
- Furceri, Davide, and Annabelle Mourougane, 2009, “The Effect of Financial Crises on Potential Output: New Empirical Evidence from OECD countries,” OECD Economics Department Working Paper No. 699.
- Giuliano, Paola, Prachi Mishra, and Antonio Spilimbergo, 2009, “Democracy and Reforms,” IZA Discussion Paper No. 4032.
- Guellec, Dominique, and Bruno van Pottelsberghe de la Potterie, 2008, “R&D and Productivity Growth: Panel Data Analysis for 16 OECD Countries,” Oxford Bulletin of Economics and Statistics, pp. 103–25.
- Gupta, Poonam, Deepak Mishra, and Ratna Sahay, 2007, “Behavior of Output during Currency Crises,” Journal of International Economics, Vol. 72, No. 2, pp. 428–50.
- Haugh, David, Patrice Ollivaud, and David Turner, 2009, “The Macroeconomic Consequences of Banking Crises in OECD Countries,” OECD Economics Department Working Paper No. 683.
- Hoeting, Jennifer A., David Madigan, Adrian E. Raftery, and Chris T. Volinsky, 1999, “Bayesian Model Averaging: A Tutorial,” Statistical Science, Vol. 14, No. 4, pp. 382–401.
- Hutchison, Michael, and Ilan Noy, 2002, “How Bad Are Twins? Output Costs of Currency and Banking Crises,” Journal of Money, Credit, and Banking, Vol. 37, No. 4, pp. 725–52.
- International Monetary Fund (IMF), 2009a, “---------------------------------------------------------- --------------” (Washington: International Monetary Fund).
- International Monetary Fund (IMF), 2009b, “From Recession to Recovery: How Soon and How Strong?” (Washington: International Monetary Fund).
- International Monetary Fund (IMF), 2008, “Structural Reforms and Economic Performance in Advanced and Developing Countries,” (Washington: International Monetary Fund). Available at http://www.imf.org/external/np/res/docs/2008/0608.htm
- International Monetary Fund (IMF), 2005, “Uruguay: Ex Post Assessment of Longer-Term Program Engagement,” IMF Country Report No. 05/202 (Washington: International Monetary Fund).
- International Monetary Fund (IMF), 2006, “Uganda: Ex Post Assessment of Longer-Term Program Engagement” (Washington: International Monetary Fund).
- Kiyotaki, Nobuhiro, and John Moore, 1997, “Credit Cycles,” Journal of Political Economy, Vol. 105 (April), pp. 211–48.
- Kose, M. Ayhan, Guy Meredith, and Christopher Towe, 2004, “How Has NAFTA Affected the Mexican Economy? Review and Evidence,” IMF Working Paper 04/59.
- Laeven, Luc, and Fabian Valencia, 2008, “Systemic Banking Crises: A New Database,” IMF Working Paper 08/224.
- Masanjala, Winford H., and Chris Papageorgiou, 2008, “Rough and Lonely Road to Prosperity: A Reexamination of the Sources of Growth in Africa Using Bayesian Model Averaging,” Journal of Applied Econometrics, Vol. 23, No. 5, pp. 671–82.
- Nickell, Stephen, Luca Nunziata, and Wolfgang Ochel, 2005, “Unemployment in the OECD since the 1960s: What Do We Know?” The Economic Journal, Vol. 115, No. 500, pp. 1–27.
- Pisani-Ferry, Jean, and Bruno van Pottelsberghe, 2009, “Handle with Care! Post-Crisis Growth in the EU,” Bruegel Policy Brief No. 2009/02.
- Pou, Pedro, 2000, “Argentina’s Structural Reforms of the 1990s,” Finance and Development, Vol. 37, No. 1, pp. 13–15.
- Reinhart, Carmen, and Kenneth Rogoff, 2009a, “Banking Crises: An Equal Opportunity Menace,” NBER Working Paper No. 14587.
- Reinhart, Carmen, and Kenneth Rogoff, 2009b, “The Aftermath of Financial Crises,” NBER Working Paper No. 14656.
- Schumpeter, Joseph, 1942, Capitalism, Socialism and Democracy (New York: Harper, 1975; originally published in 1942).

### Appendix — Data Sources (summary)
- Main databases: World Economic Outlook (WEO) and International Financial Statistics (IFS) of the IMF, and World Bank’s World Development Indicators (WDI).
- Real GDP and demand components: WDI spliced with WEO for observations after 2007.
- Current account balance, GDP deflator, fiscal balance: WEO.
- Exchange rate series: IFS.
- Domestic real interest rate: nominal lending rate (IFS) minus GDP deflator inflation.
- Capital stock: Bosworth and Collins (2003); for missing observations, constructed using perpetual inventory method with depreciation rate of 5 percent and real investment data.
- Employment and labor-force: WEO.
- Financial development: ratio of bank credit to GDP (Abiad, Dell’Ariccia, and Li (2009)); bank credit to private non-financial sector from IFS; breaks identified using IFS Country Notes and growth-spliced.
- Financial openness: sum of foreign assets and foreign liabilities divided by GDP, using External Wealth of Nations Mark II (Lane and Milesi-Ferretti (2006)).
- Trade openness: sum of exports and imports divided by GDP.
- Partner-country growth for external demand shocks: WEO.
- Three-month U.S. Treasury bill rate: Datastream.
- Structural reform indicators (trade liberalization, capital-account liberalization, financial liberalization, government efficiency): IMF; described by Giuliano et al. (2009) and IMF (2008).
- Global GDP: aggregation of individual country series in 2008 PPP dollars using IMF WEO, Total Economy Database of the Conference Board, and Historical Statistics Database of Angus Maddison; changes in sample composition smoothed by pasting together aggregate growth rates before and after each change.

### Table A1 — List of Banking Crisis Episodes (selected entries)
Note: Table reports first year of banking crisis (T), and medium-run output and growth performance based on baseline measure of precrisis trend.
- Algeria 1990  - Output Level in T+7: -20.4  - Average Growth During T+4:T+7 (percent of precrisis trend)(deviation from precrisis trend): -0.5
- Argentina 1980  - -12.6  - -0.6
- Argentina 1989  - 15.4  - 2.7
- Argentina 1995  - 7.8  - -5.5
- Argentina 2001  - -7.7  - 4.5
- Bangladesh 1987  - 16.9  - 2.1
- Benin 1988  - -17.7  - -0.4
- Bolivia 1986  - -18.7  - 1.0
- Bolivia 1994  - 6.6  - -0.3
- Brazil 1990  - -13.6  - 0.4
- Brazil 1994  - -3.0  - -0.6
- Burkina Faso 1990  - -0.4  - 1.8
- Burundi 1994  - -47.2  - -2.5
- Cameroon 1987  - -85.9  - -9.8
- Cameroon 1995  - -29.3  - 1.1
- Central African Rep. 1976  - -17.7  - -3.9
- Central African Rep. 1995  - 12.1  - 1.4
- Chad 1983  - 25.8  - 2.3
- Chad 1992  - -43.7  - -2.8
- Chile 1976  - -14.5  - -2.7
- Chile 1981  - 21.0  - 4.6
- Colombia 1982  - -15.2  - -0.3
- Colombia 1998  - -16.3  - 0.1
- Congo, Dem. Rep. of 1983  - -18.9  - -3.2
- Congo, Dem. Rep. of 1991  - -85.0  - -4.6
- Congo, Dem. Rep. of 1994  - -72.8  - -4.3
- Congo, Republic of 1992  - -58.8  - -3.9
- Costa Rica 1987  - -1.3  - 2.2
- Costa Rica 1994  - 8.8  - 1.1
- Côte d'Ivoire 1988  - -37.3  - -1.4
- Ecuador 1982  - -37.9  - -3.0
- Ecuador 1998  - 9.6  - 3.6
- Egypt 1980  - -8.9  - -2.8
- El Salvador 1989  - 37.2  - 5.3
- Finland 1991  - -14.8  - 1.6
- Ghana 1982  - -9.4  - 2.7
- Guinea 1993  - -1.7  - 1.0
- Guinea-Bissau 1995  - -46.8  - -2.3
- Haiti 1994  - -32.8  - -0.1
- India 1993  - -1.9  - -0.1
- Indonesia 1997  - -48.5  - -3.1
- Israel 1977  - -48.5  - -4.8
- Jamaica 1996  - -36.8  - -3.1
- Japan 1997  - -15.0  - -1.2
(Full table continues in original source covering 88 episodes; note also a second continued panel with entries such as Jordan 1989 -36.9 -0.1; Kenya 1985 -19.6 -2.5; Korea 1997 -22.3 -2.2; Malaysia 1997 -37.0 -4.3; Mexico 1981 -19.1 -4.0; Norway 1991 -8.2 0.3; United States 1988 4.9 0.5; Uruguay 1981 -26.4 1.2; Uruguay 2002 -3.7 4.0; Zimbabwe 1995 -23.6 -6.6.)

### Robustness: Alternative Measures of Output Losses
- Baseline measure compared to four alternatives differing in precrisis trend construction:
  - Alternative 1: Precrisis window ending one year before crisis (estimation window ends one year before crisis rather than three years).
  - Alternative 2: Longer estimation window applied to the lowest and highest 10 percent of initial trend growth-rate estimates (longer window goes back 20 years before the crisis); if unavailable or worse, initial estimate kept.
  - Alternative 3: Longer estimation window applied to all crises (solely based on 20-year window ending three years before crisis).
  - Alternative 4: Precrisis trend based on real-time IMF country desk forecasts (precrisis trend growth rate = desk forecast for real GDP growth in year t=4 made in year t=-1; per-capita growth forecast subtracts population growth in year t-1). Note: availability of these forecasts limited to post-1989 period.
- Empirical findings:
  - Losses obtained using different approaches were highly correlated and confirm large and statistically significant output losses after banking crises.
  - The 90-percent confidence bands for each measure overlap with the 90-percent confidence band of the baseline measure.
  - For Alternatives 1, 2, and 3: overlap is substantial; mean output and growth losses statistically indistinguishable from baseline.
  - For Alternative 4: average output and growth losses are greater than baseline because IMF country desk forecasts were, on average, more optimistic than the baseline precrisis trend; hence underperformance relative to forecast is greater.

### Figures (descriptive)
- Figure A1: Robustness — Alternative Precrisis Trends: Output Evolution after Banking Crises (Percent of precrisis trend). Reports estimated mean difference from year t-1 (line) and 90-percent confidence interval for estimated mean (shading); first year of crisis at t = 0; dashes indicate 90-percent confidence interval for the baseline output-loss measure.
- Figure A2: Robustness — Alternative Precrisis Trends: Growth Evolution after Banking Crises (Deviation from precrisis trend). Reports mean difference relative to precrisis growth trend in percentage points; 90-percent confidence interval for estimated mean (shaded area); first year of crisis at t = 0; dashes indicate 90-percent confidence interval for the baseline measure.

### Estimation Results without Controlling for Short-Run Crisis Severity
- Context:
  - Short-run crisis severity variable (change in output relative to trend in crisis year) is a strong predictor of medium-run output losses (Tables 4.1 and 4.2).
  - Concern: short-run crisis severity may be correlated with other explanatory variables, complicating coefficient interpretation.
- Robustness exercise:
  - Reestimated regressions omitting the short-run crisis severity variable; results in Tables 4.3 and 4.4.
  - Overall: coefficients are similar to baseline, but R-squared declines substantially relative to baseline specifications, in some cases by more than one-half.
- Table A2 (selected notes and reported coefficients):
  - Dependent variable: output at T+7 in percent of precrisis trend.
  - Selected reported coefficients and inference markers (exact formatting preserved where given):
    - (1) investment/GDP: -1.242***  [-4.195]
    - (2) investment/GDP gap: 0.687  [1.597]
    - (3) current account/GDP: 0.677*  [1.848]
    - (4) current account/GDP gap: 1.245*  [1.814]
    - (5) inflation: 0.145*  [1.781]
    - (6) inflation gap: -0.246***  [-3.004]
    - (14) currency crisis: -0.201***  [-2.729]
    - (19) pre-crisis output: 1.695***  [4.127]
    - Constant term examples: -0.099***  [-3.974]; 0.176**  [2.497]
  - Number of observations examples: 88, 85, 80, 87, 81, 26, 88, 77, 88, 85, 24, 44, 4 (presented as concatenated numbers in source).
  - R-squared examples: 0.214, 0.294, 0.282, 0.185, 0.201, 0.195, 0.214, 0.135, 0.254, 0.218, 0.134, 0.676 (as presented).
  - Note: columns 1–12 report OLS with robust t-statistics in square brackets; ***, **, and * indicate significance at the 1, 5, and 10 percent level, respectively. Column 13 reports Bayesian model averaging with estimated probability of inclusion in parentheses. The term “gap” denotes deviation of variable from precrisis historical average (years T-10 to T-7 where T denotes crisis year) during the last three years preceding the crisis.

*Source: _wp09245 - References (PDF chapter/section).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2009/_wp09245.pdf_