## 5. Episodes of Complete and Partial Recovery

## Source details

**Canonical URL:** [5. Episodes of Complete and Partial Recovery](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05147.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05147.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05147.pdf.json)

---

### I. Introduction and research questions
- Empirical gap: few studies exploit time variation in growth to assess whether countries recover from crises, shocks, and downturns.
- Illustrative cases:
  - Nigeria: rapid recovery after 1965–68 recession, passing through pre-recession peak.
  - Sweden: banking crisis in early 1990s left only a tiny fraction of output loss recuperated; output decline persisted relative to OECD average.
  - Swaziland: surge after a shallow 1987 recession.
- Core questions:
  - Do crises derail growth permanently or temporarily?
  - Do output losses accumulate across shocks to generate absolute income divergence across countries?
  - Are welfare costs of volatility larger when output losses are not recovered (permanent)?
  - Does ignoring time-varying shocks bias inference about determinants of growth from cross-section regressions?
- Empirical program:
  - Test long-term impacts of negative shocks on income levels.
  - Test whether negative shocks contribute to unconditional divergence.
  - Investigate sources of recessions.
  - Test whether speed of recovery depends on type of shock.

### II. Theoretical frameworks relevant to recoveries
- Possible post-crisis behaviors:
  - Beneficial reform/"creative destruction": recoveries above original trend (temporary shock with compensating higher growth).
  - "Cleansing" recessions: may raise long-term productivity (mixed evidence).
- Statistical/econometric views:
  - Hamilton (1989) Markov-switching stochastic trend: regime switches imply persistent output loss; post-recession output grows on a parallel path below original trend (permanent loss).
  - Friedman (1993) "pluck" model: output springs back to original trend during fast recovery (temporary loss).
- Growth-model predictions:
  - Exogenous growth (diminishing returns per effective worker): negative capital shock can raise marginal product and induce investment spurt → reversion to trend (temporary loss; serial correlation in growth).
  - Endogenous growth (constant returns per effective worker): shocks to capital or structural parameters can have lasting effects on output level or growth rate (permanent effects).
- Productivity shocks: persistence of productivity shocks implies persistence of output shocks in both model classes.
- Empirical objective: compare statistical properties of recoveries to deterministic (trend reversion) vs. stochastic (permanent change) models.

### III. Data sources, sample coverage, and key constructions
- Primary dataset:
  - World Bank’s World Development Indicators (WDI): unbalanced panel of annual observations spanning 192 countries from 1960 to 2001.
- Secondary dataset for levels and per capita analysis:
  - Penn World Tables (Heston, Summers, and Aten, 2002): unbalanced panel of 154 countries from 1960 to 2000.
- Crisis and shock indicators:
  - Banking and currency crisis dates: Kaminsky and Reinhart (1999) where available; Caprio and Klingebiel (2003) for larger banking-crisis set.
  - Exchange Market Pressure Index (EMPI): percentage depreciation in exchange rate plus percentage loss in foreign exchange reserves; crisis dummy = 1 if EMPI is in the upper quartile across the panel.
  - Civil war: Sarkees (2000) Correlates of War Intra-State War Data, 1816–1997 (v3.0); dummy = 1 for country-years of internal conflict.
  - Trade liberalization: Wacziarg and Welch (2003); dummy = 0 pre-liberalization years, 1 in year of liberalization and subsequent years.
  - Financial liberalization: IMF Research Department Financial Reform database (Omori, 2005); overall index and capital account liberalization component used.
  - Political regime change: Polity International; durability measured as years since most recent regime change; dummy = 1 when durability becomes zero. Polity score scale: +10 (strongly democratic) to -10 (strongly autocratic).
- Terms of trade (TOT): IFS unit price index (exports/imports) and COMTRADE world commodity prices (Cashin, Cespedes, and Sahay (2002)); IFS used where available.
- Methodological notes:
  - EMPI not normalized by country standard deviation in main construction to preserve cross-country comparability.
  - Paper exploits within-country variation with annual frequency; acknowledges debates over optimal aggregation frequency and trend specification.

### IV. Descriptive statistics and stylized facts
- Dataset coverage:
  - WDI: 192 countries, 1960 to 2001.
  - Penn World Tables: 154 countries, 1960 to 2000.
- Growth around recessions:
  - Average growth for expansion years in the World Bank dataset: 5.7 percent.
  - Median growth rates in expansion years are lower than the mean due to positive skewness.
  - Median growth rates in the year and three years immediately following a trough are about ½ percentage point lower than in a typical expansion year.
  - Penn World Tables: expansion years surrounding a recession average slightly higher growth; median growth rates are the same or lower.
  - For a common country and sample period, the Penn World Tables data can show lower per capita growth due to population growth effects.
- Dataset-level divergence in measured long-run growth:
  - Reported 3.7 percent per year in the World Bank dataset falls to 2.1 percent in Penn World Tables (attributed to population growth).
- Recession frequency, duration, and cumulative losses:
  - Average cumulative output loss in a recession (full sample, both datasets): 7½ percent.
  - Recessions last 1.6–1.8 years on average across the full sample.
  - High-income countries: recessions are shallower and shorter — about 3–4 percent cumulative loss.
- Recessions coincide with shocks:
  - Financial crises coincide with more than one-third of years of negative growth.
  - Half of all recession years coincide with crises, regime change, civil war, or some combination.

### V. Econometric evidence on recoveries and persistence
- Timeline analysis:
  - Timelines align peak years (t = 0) and set output = 100 in peak year; pre-peak constructed using average growth of expansion years prior to peak.
  - In WDI, only a few percentage points lost during recession are recuperated for episodes associated with civil wars and banking crises; gap widens for other samples.
  - Recessions lead to permanent output losses for all samples at least through end of five years after a trough.
  - Few complete Friedman-style recoveries identified; many episodes resemble Hamilton-style persistent losses (example noted: Mexico 1995 as exception).
- Panel regressions (restricted to expansion years, fixed effects):
  - World Bank (1962-01): Trough (-1) coefficient = -0.59 *** (t-stat -4.1).
  - Penn World Tables (1962-00): Trough (-1) coefficient = -0.09 (t-stat -0.8).
  - Alternative Trough windows often negative and significant (examples in World Bank: Trough (-1,-2) = -0.43 ***; Trough (-1,-2,-3) = -0.34 ***).
  - Interpretation: coefficient of zero would be consistent with Friedman reversion-to-trend; observed negative and significant coefficients support Hamilton-style persistence (no fast reversion).
- Heterogeneity of post-trough slopes:
  - One year after trough, 107 out of 172 countries (World Bank sample with sufficient data) have negative slope coefficients (probability under fair coin toss < 1 percent).
  - Distribution of slopes centered at –0.4; null that mean = 0 rejected at 1 percent.
- Income-group and regional heterogeneity (selected estimates from Table 7 and Table 8):
  - All Countries (1962-2001): coefficient = -0.59 *** (No. Countries 189, No. Obs. 4756; t-stat -4.1).
  - High Income (1962-2001): coefficient = -0.87 *** (No. Countries 41, No. Obs. 1265; t-stat -3.7).
  - Low Income (1962-2001): coefficient = 0.58 * (No. Countries 58, No. Obs. 1372; t-stat 1.9) — some evidence of partial rebound in low-income countries, but insufficient to return to pre-crisis trend.
  - Industrial Countries (1962-2001): coefficient = -0.75 *** (No. Countries 26, No. Obs. 912; t-stat -2.9).
  - Africa (1962-2001): coefficient = 0.40 (No. Countries 51, No. Obs. 1271; t-stat 1.3); rebound magnitudes small and sensitive to outliers.
  - Transition economies: largest negative coefficients (1962-2001: -1.72 ***; 1990-2001: -2.71 ***).
- Tests of amplitude and duration (Table 9):
  - World Bank sample (469 pairs): Duration of prior recession coefficient on duration of expansion = -3.27 *** (t-stat -3.6).
  - Penn World Tables (747 pairs): Duration of prior recession coefficient on duration of expansion = -1.98 *** (t-stat -5.7).
  - Conclusion: expansions are weaker or shorter when preceding recessions are longer; no evidence that deep/prolonged recessions lead to rebound/growth takeoff.
- Tests of booms prior to recessions (Table 10):
  - Dummy (P3) for peak year and previous two years: coefficient = -0.27 *** (t-stat -2.7) in World Bank — years immediately prior to recessions tend to experience significantly lower growth, contradicting the strong-boom-trigger hypothesis.

### VI. Persistence, unit roots, Monte Carlo evidence, and serial correlation
- Panel unit root tests (Table 11):
  - Hadri panel test rejects null of no unit root in favor of unit root for ln(GDP per capita) in PWT (Hadri Z-statistic example: 53.5 with Prob. 0.01 reported).
  - Pesaran CIPS test statistics (World Bank: –1.09, –1.33, –1.86 across deterministic specifications) do not reject null of unit root given cited critical values; paper concludes null of a unit root cannot be rejected.
- Serial correlation (Table 12):
  - Growth Rate (-1) coefficient (FE, World Bank) = 0.29 *** (t-stat 12.0).
  - D(Growth Rate (-1)) coefficient (FE) = 0.37 *** (t-stat 7.7).
  - Serial correlation in growth is positive and increases across samples.
- Monte Carlo experiments:
  - DGP parameters: μ = a = 0.03, φ = 0.2, σ = 0.07 to match first two moments.
  - Stochastic trend DGP yields significantly negative coefficient on dummy for expansion year after trough; deterministic trend DGP yields significantly positive coefficient.
  - Conclusion: stochastic trend DGP consistent with data properties; deterministic trend DGP is not.

### VII. Shocks, crises, political change, and effects on growth and recovery
- Impact of shocks on growth (selected coefficients from Tables 15–16):
  - Currency crisis coefficients on growth: large and negative across specifications (examples range approximately -0.8 *** to -2.3 *** depending on lag and sample).
  - Currency crisis(-1) examples: -1.0 *** to -2.1 *** in various specifications.
  - Banking crisis coefficients: large and negative (examples approximately -1.5 *** to -2.5 ***).
  - Civil War coefficient: -2.7 *** (World Bank) and -1.4 *** (Penn World Tables).
  - Change in government coefficient: -1.5 *** (World Bank) and -1.3 *** (Penn World Tables).
  - Trade liberalization (Tradelib): 0.4 *** (World Bank) and -0.1 (Penn World Tables).
  - Change in terms of trade: small quantitative effect (example 0.001 ** in one specification).
- Interaction effects on recoveries (Tables 17–18):
  - Recoveries weaker when contraction associated with financial crisis:
    - Example (Penn World Tables): Trough(-1)*Bankcrisis(-1) = -0.96 *** (t-stat -3.3).
    - Trough(-1)*Currency crisis(-1) = -0.58 ** (t-stat -2.1).
  - Recoveries weaker in countries with more liberalized capital accounts/financial markets:
    - Trough(-1)*Capacctlib(-2) = -0.46 ** (Penn World Tables).
    - Trough(-1)*Finlib(-2) = -0.09 ** (World Bank).
  - Aid inflows boost recovery:
    - Trough(-1)*Aid2GNI(-1) = 0.06 *** (Penn World Tables) and 0.08 *** (World Bank).
  - Political transitions:
    - Trough(-1)*Autocratic change in government(-1) = -2.55 *** (Penn World Tables) and -1.04 * (World Bank) — large negative effects on recovery.
    - Trough(-1)*Democratic change in government(-1) not significant in pooled samples; in Africa democratic change associated with positive and significant recovery effects in Penn World Tables.
  - Civil war interactions:
    - Civil wars lead to weak recoveries generally.
    - Civil wars in Africa that coincide with autocratic transitions produce especially weak recoveries (example interaction Trough(-1)*Civil war(-1)*Africa*Autocratic change = -5.32 *** in one specification).

### VIII. Interpretation, policy-relevant implications, and implications for convergence
- Empirical properties:
  - (1) Negative shocks to level of output typically do not dissipate — persistent level losses common.
  - (2) Growth is positively serially correlated.
- Implications for theory:
  - Data inconsistent with temporary demand-driven, trend-stationary business cycles (which predict negative serial correlation).
  - Data inconsistent with simple neoclassical prediction that capital reductions should trigger rapid catch-up growth, except for some low-income/African partial rebounds.
  - Stochastic trend/Hamilton-style persistence better matches observed recovery patterns than Friedman-style fast reversion.
- Policy-relevant findings:
  - Limiting susceptibility to financial and political crises could materially affect long-run income levels by reducing persistent output losses.
  - Aid inflows can materially improve recovery (positive and significant Trough(-1)*Aid2GNI(-1) coefficients).
  - Democratic political change tends to be associated with stronger recoveries relative to autocratic shifts.
  - Trade liberalization tends to raise long-run growth but may weaken recovery when combined with liberalized capital accounts (possible mechanism: financing constraints for imported intermediate inputs).
- Convergence implications:
  - Conditional on expansion phases, there is statistically significant evidence of convergence.
  - Absolute divergence arises because poor countries suffer more frequent and deeper recessions.
  - Poor countries do not appear to be stuck in a savings trap, but many are subject to a crisis trap that hinders long-run catch-up.

*IMF Working Paper chapter "5. Episodes of Complete and Partial Recovery" (content unit).*

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

### _wp05147 - References

### Tables
- 1. Average Growth Rates ........................................................................................................25
- 2. Characteristics of Recessions.............................................................................................. 25
- 3. Strength of Recoveries ........................................................................................................ 26
- 4. Trend Growth and Volatility............................................................................................... 27
- 5. Heterogeneous Slopes ......................................................................................................... 27
- 6. Individual Slopes ................................................................................................................ 28
- 7. Income Groups.................................................................................................................... 29
- 8. Regional Tests..................................................................................................................... 29
- 9. Tests of Amplitude, Duration, and Steepness of Complete Expansions............................. 30
- 10. Tests of Strong Boom Prior to Recessions ....................................................................... 30
- 11. Panel Unit Root Tests ....................................................................................................... 31
- 12. Serial Correlation of Growth Rates................................................................................... 31
- 13. Monte Carlo Results ......................................................................................................... 31
- 14. Phase Conditional Convergence ....................................................................................... 32
- 15. Crises and Growth............................................................................................................. 32
- 16. Other Shocks and Growth ................................................................................................. 32
- 17. Recovery Conditional on Financial Crises and Liberalization ......................................... 33
- 18. Recovery Conditional on Political Crises and Africa ....................................................... 34

### Figures
- 1. Timeline of Recession: World Bank Data .......................................................................... 35
- 2. Timeline of Recession: Penn World Tables Data ............................................................... 36
- 3. Episodes of No Recovery in Selected Crisis and Asian Countries ..................................... 37
- 4. Episodes of No Recovery in Developed and Other Countries............................................ 38

*Source: _wp05147 - References*

### 5. Episodes of Complete and Partial Recovery....................................................................... 39

### 5. Episodes of Complete and Partial Recovery

### I. INTRODUCTION
- Empirical growth literature has emphasized cross-sectional growth regressions; few studies exploit variation of growth over time and whether countries recover from crises, shocks, and downturns.
- Three illustrative country cases:
  - Nigeria: following a steep recession from 1965–68, output recovered rapidly to its former trend line (passing through the pre-recession peak).
  - Sweden: only a tiny fraction of the output loss from Sweden’s banking crisis in the early 1990s was recuperated; output decline persisted relative to OECD average.
  - Swaziland: growth surged after a shallow recession in 1987.
- Key questions motivated:
  - Do crises derail growth permanently or temporarily?
  - Do output losses accumulate across shocks to generate absolute income divergence across countries?
  - Are welfare costs of volatility larger when output losses are not recovered (permanent)?
  - Does ignoring time-varying shocks bias inference about determinants of growth from cross-section regressions?
- Implications discussed:
  - If shocks have permanent effects, panel data with higher frequency become more informative than long-span cross-section averages.
  - If output follows a stochastic trend, every shock changes the conditional expectation of future income one for one; if output reverts to trend, shocks are transitory.
- The paper’s empirical program:
  - Test whether negative shocks have long-term impacts on income levels (Section IV).
  - Test whether negative shocks contribute to unconditional divergence across countries (Section V).
  - Investigate sources of recessions (Section VI).
  - Test whether speed of recovery depends on the type of shock (Section VII).

### II. THEORIES OF CRISES AND GROWTH
- Theoretical possibilities for post-crisis behavior:
  - Beneficial reform or "creative destruction" could generate recoveries above original trend (temporary shock with compensating higher growth).
  - Recessions that cleanse inefficient firms may raise long-term productivity (evidence mixed: Gali and Hammour (1993) vs. Caballero and Hammour (2005)).
- Statistical/econometric views:
  - Hamilton (1989) Markov-switching stochastic trend: regime switches in growth rates imply persistent output loss; post-recession output grows on a parallel path below original trend (permanent loss).
  - Friedman (1993) "pluck" model: output springs back to original trend during a fast recovery phase (temporary loss).
- Predictions from growth models:
  - Exogenous growth models (diminishing returns to capital per effective worker):
    - Steady-state condition: s f(k*) = (n + d + x) k*
    - A negative shock to capital below steady state raises marginal product, inducing an investment spurt and reversion to trend (temporary loss and serial correlation in growth rates).
    - Shocks to s, n, d, or x change the steady state and produce transitional dynamics; recovery is gradual toward the new steady state.
  - Endogenous growth models (constant returns to capital per effective worker):
    - Long-term growth rate set by structural parameters and policy; shocks to capital or structural parameters can have lasting effects on output level or growth rate (permanent effects).
- Productivity shocks:
  - Persistence of productivity shocks will be mirrored in persistence of output shocks since productivity enters linearly in production functions of both model classes.
- Empirical testing objective:
  - Test statistical properties of recoveries and compare to predictions of deterministic (trend reversion) vs. stochastic (permanent change) models of output.
  - Assess whether volatility affects convergence of income levels across countries.

### III. DATA SOURCES AND DESCRIPTIVE STATISTICS
- Primary data:
  - World Bank’s World Development Indicators (WDI): GDP growth rates; unbalanced panel of annual observations spanning 192 countries from 1960 to 2001.
- Secondary data for levels and per capita analysis:
  - Penn World Tables (Heston, Summers, and Aten, 2002): comparable levels of GDP per capita; unbalanced panel of 154 countries from 1960 to 2000.
  - Note: WDI comparable levels of GDP per capita available only from 1980.
- Definitions and construction of explanatory variables and shock indicators:
  - Crisis indicators:
    - Banking and currency crisis dates from Kaminsky and Reinhart (1999) used where available (23 countries in that study).
    - Exchange Market Pressure Index (EMPI) constructed as percentage depreciation in exchange rate plus percentage loss in foreign exchange reserves; crisis dummy = 1 if EMPI is in the upper quartile of all observations across the panel.
    - Banking crisis dates on a larger set from Caprio and Klingebiel (2003).
  - Civil war:
    - Data from Sarkees (2000) Correlates of War Intra-State War Data, 1816–1997 (v3.0); dummy = 1 for country-years of internal conflict, 0 otherwise.
  - Trade liberalization:
    - Dates from Wacziarg and Welch (2003); dummy = 0 pre-liberalization years, 1 in year of liberalization and subsequent years.
  - Financial liberalization:
    - Measures from the Financial Reform database compiled by the IMF's Research Department (Omori, 2005); both overall index and capital account liberalization component used. Overall index includes directed credit/reserve requirements, interest rate controls, entry barriers/pro-competition measures, banking supervision, privatization, and security markets; countries can backtrack.
  - Terms of trade (TOT):
    - Sources: IMF’s International Financial Statistics (IFS) unit price of exports and imports (index = unit price of exports / unit price of imports), and world commodity prices from COMTRADE weighted by top three exports for 60 countries (Cashin, Cespedes, and Sahay (2002)) for constructing TOT index; where both available, IFS unit prices used.
  - Political regime change:
    - Data from Polity International; regime durability measures years since most recent regime change (defined by a three point change in the polity score over three years or less or the end of transition). Dummy = 1 when durability variable becomes zero.
    - Polity score scale ranges from +10 (strongly democratic) to -10 (strongly autocratic).
- Methodological notes:
  - The crisis literature often normalizes reserves and exchange rate movements by their standard deviations, but that makes EMPI magnitudes comparable only within countries; interest rates dropped due to data scarcity.
  - The paper will exploit within-country variation to study recovery patterns, volatility, and convergence, acknowledging debates on optimal frequency (annual vs. 5- or 10-year averages) and deterministic vs. stochastic trend specifications.
- Descriptive statistics:
  - The WDI sample: 192 countries, 1960 to 2001.
  - The Penn World Tables sample: 154 countries, 1960 to 2000.

*Italic source attribution: IMF Working Paper chapter "5. Episodes of Complete and Partial Recovery" (content unit)._

### 3.7 percent per year in the World Bank dataset, and falls to 2.1 percent in Penn World Tables

### _wp05147 - 3.7 percent per year in the World Bank dataset, and falls to 2.1 percent in Penn World Tables

### Growth rates around recessions
- Average growth for expansion years in the World Bank dataset is 5.7 percent.
- Median growth rates in expansions are lower-than-average rates due to some positive skewness.
- Median growth rates in the year and three years immediately following a trough are about ½ percentage point lower than in a typical expansion year.
- In the Penn World Tables data:
  - Expansion years surrounding a recession average slightly higher growth.
  - Median growth rates are the same or lower.
- For a common country and sample period, the Penn World Tables data could include more episodes of “recession” to the extent that output growth in a particular year, although positive, was insufficient to outpace population growth.

### Dataset-level and sample considerations
- The reported 3.7 percent per year in the World Bank dataset falls to 2.1 percent in Penn World Tables due to population growth.
- In Table 1, the three-year average growth rate immediately before and after a recession is higher than the average growth rate because, by definition, a peak year and the year after a trough are expansion years.

### Recession frequency, duration, and cumulative losses
- For the sample of World Bank and Penn World Tables growth rates, the authors calculated:
  - The average cumulative loss of recessions (defined as years of negative growth).
  - The average length of recessions.
- The cumulative output loss in a recession averages 7½ percent for the full sample of countries in both datasets.
- Recessions last 1.6–1.8 years on average across the full sample.
- Recessions are much shallower and shorter for high-income countries—about 3–4 percent cumulative loss.

### Additional sample breakdowns reported
- Recession statistics are further broken into:
  - Income groups and regions.
  - Recessions corresponding with crises, wars, new governments.
  - Countries that have liberalized their trade or financial systems.

*Source: _wp05147 - 3.7 percent per year in the World Bank dataset, and falls to 2.1 percent in Penn World Tables*

### 1.4 years—than for any other group. Civil wars, changes in government regime, and banking

### _wp05147 - 1.4 years—than for any other group. Civil wars, changes in government regime, and banking

### Key findings on recessions and recoveries
- Recessions typically produce permanent losses in the level of output: output in the recovery does not recoup the level associated with a linear extrapolation of the original trend at least through the end of five years after a trough.
- Following a recession, growth rebounds at a rate significantly below that of an average expansion year.
- Financial crises, civil wars, and changes in government regime correspond with the deepest and most prolonged recessions.
- Financial crises coincide with more than one-third of years of negative growth. Half of all recession years coincide with crises, regime change, civil war, or some combination of these variables.
- Trade liberalization has a significant positive effect on long-run growth, but trade liberalization can weaken recovery from recession when combined with liberalized capital account regimes.

### Timeline evidence and descriptive statistics
- Timelines are constructed aligning peak years (t = 0) and show output in a 12-year window around the peak; output is set to 100 in the peak year and pre-peak years are constructed using the average growth rate of expansion years prior to the peak.
- In the World Bank dataset, a few percentage points of output lost during the recession are recuperated in the recovery for episodes associated with civil wars and banking crises, but the gap widens for all other samples.
- Recessions lead to permanent output losses for all samples, at least through the end of five years after a trough.
- Individual-country heterogeneity: many country episodes resemble Hamilton-style (persistent loss) recoveries; very few complete Friedman recoveries are identified (Mexico 1995 is a noted exception).

### Econometric evidence on recovery strength
- Basic panel specification (restricted to expansion years) conditions on positive growth and exploits within-country time variation with fixed effects; basic estimation equation incorporates dummy(s) for post-trough years (Trough lags).
- Main regression evidence (Table 3):
  - World Bank (sample 1962-01): Trough (-1) coefficient = -0.59 *** (t-stat -4.1).
  - Penn World Tables (sample 1962-00): Trough (-1) coefficient = -0.09 (t-stat -0.8).
  - Alternative specifications using Trough(-1,-2), Trough(-1,-2,-3), Trough(-1,-2,-3,-4) show negative and often significant coefficients (examples: Trough (-1,-2) = -0.43 ***; Trough (-1,-2,-3) = -0.34 *** in World Bank).
- Interpretation: coefficient of zero would be consistent with Friedman reversion-to-trend; observed negative and significant coefficients support Hamilton-style persistence (no fast reversion).
- Cross-section evidence (Table 4): countries experiencing more frequent years of negative growth have sharply lower average growth rates. Example: a country experiencing contractions 20 percent of the time will have an average annual growth rate approximately 1 percentage point lower than a country experiencing contractions only 10 percent of the time.

### Heterogeneity across countries, income groups, and regions
- Heterogeneous slopes:
  - One year after the trough, 107 out of the 172 countries with sufficient data in the World Bank data have negative slope coefficients (probability of this under a fair coin toss is less than 1 percent).
  - Distribution of slopes in the full World Bank sample is centered at –0.4; null hypothesis that mean = 0 can be rejected at a 1 percent confidence level.
- Income groups (Table 7, regression of growth rate in expansion on dummy for one year after a trough):
  - All Countries (1962-2001): coefficient = -0.59 *** (No. Countries 189, No. Obs. 4756; t-stat -4.1).
  - High Income (1962-2001): coefficient = -0.87 *** (No. Countries 41, No. Obs. 1265; t-stat -3.7).
  - Low Income (1962-2001): coefficient = 0.58 * (No. Countries 58, No. Obs. 1372; t-stat 1.9) — only low-income countries show some evidence of partial rebound, though insufficient to return to pre-crisis trend.
- Regional tests (Table 8):
  - Industrial Countries (1962-2001): coefficient = -0.75 *** (No. Countries 26, No. Obs. 912; t-stat -2.9).
  - Africa (1962-2001): coefficient = 0.40 (No. Countries 51, No. Obs. 1271; t-stat 1.3); but rebound magnitudes are small (average output loss ~6½ percent, rebound <½ percent) and sensitive to outliers.
  - Transition economies show the largest negative coefficients (1962-2001: -1.72 ***; 1990-2001: -2.71 ***).

### Tests of rebound, booms, amplitude and duration
- Tests of amplitude/duration of complete recession-expansion pairs (Table 9):
  - World Bank sample (469 pairs): Duration of prior recession coefficient on duration of expansion = -3.27 *** (t-stat -3.6).
  - Penn World Tables (747 pairs): Duration of prior recession coefficient on duration of expansion = -1.98 *** (t-stat -5.7).
  - Conclusion: expansions are weaker or shorter when preceding recessions are longer; no evidence that deep/prolonged recessions lead to rebound or growth takeoff.
- Tests of booms prior to recessions (Table 10):
  - Dummy (P3) for peak year and previous two years: coefficient = -0.27 *** (t-stat -2.7) in World Bank — years immediately prior to recessions tend to experience significantly lower growth, contradicting hypothesis that strong booms trigger recessions.
  - Strong expansions precede significantly shorter recessions; prolonged expansions precede significantly shorter and shallower recessions.

### Persistence, unit roots, and Monte Carlo evidence
- Panel unit root testing (Table 11):
  - Hadri panel test rejects null of no unit root in favor of unit root for both datasets (Hadri Z-statistics reported, e.g., 53.5 with Prob. 0.01 for ln(GDP per capita) -- PWT).
  - Pesaran CIPS test statistics (World Bank: –1.09, –1.33, –1.86 across deterministic specifications) do not reject null of unit root given critical values cited; conclusion: null of a unit root cannot be rejected.
- Serial correlation (Table 12):
  - Growth Rate (-1) coefficient (FE, World Bank) = 0.29 *** (t-stat 12.0).
  - D(Growth Rate (-1)) coefficient (FE) = 0.37 *** (t-stat 7.7).
  - Serial correlation in growth is positive and increases across samples.
- Monte Carlo experiments:
  - DGP parameters set to μ = a = 0.03, φ = 0.2 and σ = 0.07 to match first two moments.
  - Data generated as a stochastic trend yield a significantly negative coefficient on the dummy for an expansion year after a trough; deterministic trend DGP yields a significantly positive coefficient.
  - Conclusion: stochastic trend DGP is consistent with the properties of the data; deterministic trend DGP is not.

### Shocks, crises, political change, and growth
- Regressions linking shocks to growth:
  - Currency crisis and banking crisis coefficients on growth are large and negative across multiple specifications (examples from Table 15: Currency crisis coefficients ~ -0.8 *** to -2.3 ***, Currency crisis(-1) ~ -1.0 *** to -2.1 ***, Banking crisis ~ -1.5 *** to -2.5 *** depending on sample and dates).
  - Civil War coefficient (Table 16) = -2.7 *** (World Bank) and -1.4 *** (Penn World Tables).
  - Change in government coefficient = -1.5 *** (World Bank) and -1.3 *** (Penn World Tables).
  - Trade liberalization (Tradelib) coefficient = 0.4 *** (World Bank) and -0.1 (Penn World Tables) — positive long-run effect in World Bank regressions.
  - Change in terms of trade has negligible quantitative effect (e.g., 0.001 ** in one specification).
- Interaction of shocks with recoveries (Table 17 and Table 18):
  - Recoveries are weaker when output contraction is associated with a financial crisis:
    - Example (Penn World Tables): Trough(-1)*Bankcrisis(-1) = -0.96 *** (t-stat -3.3); Trough(-1)*Currency crisis(-1) = -0.58 ** (t-stat -2.1).
  - Recoveries are weak in countries with more liberalized capital accounts and financial markets:
    - Examples: Trough(-1)*Capacctlib(-2) = -0.46 ** (Penn World Tables); Trough(-1)*Finlib(-2) = -0.09 ** (World Bank).
  - Higher inflows of aid (Aid2GNI) boost growth in recovery: Trough(-1)*Aid2GNI(-1) = 0.06 *** (Penn World Tables) and 0.08 *** (World Bank).
  - Political transitions:
    - Trough(-1)*Autocratic change in government(-1) = -2.55 *** (Penn World Tables) and -1.04 * (World Bank) — large negative effects on recovery.
    - Trough(-1)*Democratic change in government(-1) not significantly different from typical recoveries in pooled samples; in Africa, democratic change is associated with positive and significant recovery effects in Penn World Tables.
  - Civil wars lead to weak recoveries in general; civil wars in Africa that coincide with change to autocratic governments generate especially weak recoveries (interaction term Trough(-1)*Civil war(-1)*Africa*Autocratic change = -5.32 *** in one specification).

### Interpretation and policy-relevant implications
- Empirical properties summarized:
  - (1) Negative shocks to level of output typically do not dissipate (persistent level losses).
  - (2) Growth is positively serially correlated.
- The data are inconsistent with temporary demand-driven, trend-stationary business cycles (which would produce negative serial correlation) and with simple neoclassical predictions that a reduction in capital stock should raise growth back rapidly (except in some low-income/African cases where partial rebounds are observed).
- Financial crises and political instability are costly at all horizons and contribute to persistent output losses; they explain a large share of negative-growth years.
- Policy-relevant findings:
  - Limiting susceptibility to financial and political crises could materially affect long-run income levels by reducing persistent output losses.
  - Aid inflows can materially improve recovery (Trough(-1)*Aid2GNI(-1) positive and significant).
  - Democratic political change is associated with stronger recoveries relative to autocratic shifts.
  - Trade liberalization tends to raise long-run growth but may weaken recovery in contexts of liberalized capital accounts (possible mechanism: financing constraints for imported intermediate inputs).
- Implication for convergence: conditional on expansion phases, there is statistically significant evidence of convergence. Absolute divergence across countries arises because poor countries suffer more frequent and deeper recessions; poor countries do not appear to be stuck in a savings trap but many are subject to a crisis trap.

*Italic source: Excerpt from the IMF working paper (content unit provided).*

### References

### _wp05147 - References

### Cross-country growth and convergence literature
- Acemoglu, Daron, Simon Johnson, and James Robinson, 2001, “Colonial Origins of Comparative Development: An Empirical Investigation,” American Economic Review, Vol. 91, pp. 1369–1401.
- Barro, Robert J., 1997, Determinants of Economic Growth: A Cross-Country Empirical Study (Cambridge, Massachusetts: MIT Press).
- Barro, Robert J., 1991, “Economic Growth in a Cross Section of Countries,” Quarterly Journal of Economics Vol. 106, No. 2, pp. 407–43.
- Barro, Robert J., 2003, “Determinants of Economic Growth in a Panel of Countries,” Annals of Economics and Finance, Vol. 4, No. 2, pp. 231–74.
- Barro, Robert J., and Xavier Sala-i-Martin, 2004, Economic Growth, Second Edition, (New York: McGraw Hill).
- Caselli, Francesco, Gerardo Esquivel and Fernando Lefort, 1996, ”Reopening the Convergence Debate: A New Look at Cross-Country Growth Empirics,” Journal of Economic Growth, Vol. 1, No. 3, pp. 363–89.
- Doppelhofer, Gernot, Ronald Miller, and Xavier Sala-i-Martin, 2000, “Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach,” NBER Working Paper 7750.
- Islam, Nazrul, 1995, “Growth Empirics: A Panel Data Approach,” Quarterly Journal of Economics, Vol. 110 (November), pp. 1127–70.
- Levine, Ross, and David Renelt, 1992, “A Sensitivity Analysis of Cross-Country Growth Regressions,” American Economics Review, Vol. 82, No. 4, pp. 942–63.
- Lucas, Robert E., Jr. 1988, “On the Mechanics of Economic Development,” Journal of Monetary Economics, Vol. 22, pp. 2–42.
- Mankiw, Gregory N., David Romer, and David Weil, 1992, “A Contribution to the Empirics of Economic Growth,” Quarterly Journal of Economics, Vol. 107 (May), pp. 407–38.
- Pritchett, Lant, 2002, “Understanding Patterns of Economic Growth: Searching for Hills among Plateaus, Mountains, and Plains,” World Bank Economic Review, Vol. 14, No. 2, pp. 221-50.
- Sala-i-Martin, Xavier, 1997, “I Just Ran Two Million Regressions,” American Economic Review, Papers and Proceedings, Vol. 87, No. 2, pp. 178–83.
- Durlauf, Steven, 2003, “The Convergence Hypothesis After Ten Years,” University of Wisconsin at Madison, manuscript.
- Caselli et al., and other cross-country empirical contributions listed above.

### Business cycles, volatility, and long-run effects
- Aghion, Philippe, and Gilles Saint-Paul, 1991, “On the Virtue of Bad Times: An Analysis of the Interaction Between Economic Fluctuations and Productivity Growth,” CEPR Working Paper, No. 578.
- Caballero, Ricardo J., and Mohamad Hammour, 1994, “The Cleansing Effect of Recession,” American Economic Review, Vol. 84, No. 5, pp. 1350–68.
- Caballero, Ricardo J., and Mohamad Hammour, 2005, “The Cost of Recessions Revisited: A Reverse-Liquidationist View,” Review of Economic Studies 72, (March), pp. 313–41.
- Friedman, Milton, 1993, “The ‘Plucking Model’ of Business Fluctuations Revisited,” Economic Inquiry, Vol. 31 (April), pp. 171–77.
- Gali, Jordi, and Mohamad Hammour, 1993, “Long-Run Effects of Business Cycles” (New York: Columbia University Graduate School of Business), unpublished manuscript.
- Hamilton, James D., 1989, “A New Approach to the Economic Analysis of Nonstationary Times Series and the Business Cycle,” Econometrica, Vol. 57 (March), pp. 357–84.
- Harding Don, and Adrian Pagan, 2002, ‘Dissecting the Cycle: A Methodological Investigation’, Journal of Monetary Economics, Vol. 49, pp. 365–81.
- Martin, Philippe, and Carol Ann Rogers, 1997, “Stabilization Policy, Learning-By-Doing, and Economic Growth,” Oxford Economic Papers, Vol. 49, No. 1, pp. 152–66.
- Nelson, Charles R., and Charles I. Plosser, 1982, “Trends and Random Walks in Macroeconomic Time Series: Some Evidence and Implications,” Journal of Monetary Economics, Vol. 10, No. 2, pp. 139–62.
- Ramey, G. and Ramey, V. A., 1995, “Cross-Country Evidence on the Link Between Volatility and Growth,” American Economic Review, Vol. 85, pp. 1138–51.
- Dawson, J. W., and F.E. Stephenson, 1997, “The Link Between Volatility and Growth: Evidence from the States,” Economic Letters, Vol. 55, pp. 365–69.
- Siegler, Mark, 2005, “International Growth and Volatility in Historical Perspective,” Applied Economics Letters, Vol. 12, No. 2, pp. 67–71.

### Crises, financial instability, and sudden stops
- Arellano, M., and S.R. Bond, 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations,” Review of Economic Studies, No. 58, pp. 277–97.
- Caprio, Gerard, and Daniela Klingebiel, 2003, “Episodes of Systemic and Borderline Financial Crises,” World Bank data set available via the internet: http://econ.worldbank.org/view.php?id=23456.
- Cerra, Valerie, and Sweta Saxena, 2005a, “Did Ouput Recovery from the Asian Crisis?,” IMF Staff Papers, Vol. 54 (1), pp. 1–23.
- Cerra, Valerie, and Sweta Saxena, 2005b, “Eurosclerosis or Financial Collapse: Why Did Swedish Incomes Fall Behind?,” IMF Working Paper 05/29 (Washington: International Monetary Fund).
- Chari, V.V., Patrick Kehoe, and Ellen McGrattan, 2005, “Sudden Stops and Output Drops,” Federal Reserve Bank of Minneapolis, Research Department Staff Report 353 (Minneapolis).
- Kaminsky, Graciela, and Carmen Reinhart, 1999, “The Twin Crises: The Causes of Banking and Balance of Payments Problems,” American Economic Review, Vol. 89, No. 3 (June), pp. 473–500.
- Omori, Sawa, 2005, “Financial Reform Database: Measuring Seven Dimensions of Financial Liberalization” (Washington: International Monetary Fund), unpublished manuscript.
- Morsink, James, Thomas Helbling, and Stephen Tokarick, 2002, “Recessions and Recoveries,” World Economic Outlook, Chapter III, (Washington: International Monetary Fund).

### Conflict, political variables, and development collapses
- Collier, Paul, 1999, “On the Economic Consequences of Civil War,” Oxford Economic Papers, Vol. 51, pp. 168–83.
- Rodrik, Dani, 1999, “Where Did All the Growth Go?: External Shocks, Social Conflict, and Growth Collapses” Journal of Economic Growth, Vol.4, No. 4, pp. 385–412.
- Sachs, Jeffrey, John W. Mcarthur, Guido Schmidt-Traub, Margaret Kruk, Chandrika Bahadur, Michael Faye, and Gordon Mccord, 2004, “Ending Africa’s Poverty Trap,” Brookings Papers on Economic Activity, Vol. 1, pp. 117–240.
- Sarkees, Meredith Reid, 2000, “The Correlates of War Data on War: An Update to 1997,” Conflict Management and Peace Science, Vol. 18, No. 1, pp. 123–44. Available at: www.correlatesofwar.com
- Singer, J. David, and Melvin Small, 1994, “Correlates of War Project: International and Civil War Data, 1816-1992” [Computer file], Inter-university Consortium for Political and Social Research [distributor], Ann Arbor, Michigan. Available at: http://ssdc.ucsd.edu/ssdc/icp09905.html
- Polity IV Project, Integrated Network for Societal Conflict Research (INSCR) Program, Center for International Development and Conflict Management (CIDCM), University of Maryland, College Park 20742. Available at: www.cidcm.umd.edu/inscr/polity

### Trade, liberalization, and structural determinants
- Cashin, Paul A., Luis Cespedes, and Ratna Sahay, 2002. "Keynes, Cocoa, and Copper: In Search of Commodity Currencies," IMF Working Paper No. 02/223 (Washington: International Monetary Fund).
- Wacziarg, Romain and Karen Horn Welch, 2003. "Trade Liberalization and Growth: New Evidence," NBER Working Paper No. w10152 (Cambridge, Massachusetts: National Bureau of Economic Research).
- Hausmann, Ricardo, Lant Pritchett, and Dani Rodrik, 2004, “Growth Accelerations,” Harvard University, unpublished manuscript.

### Data sources, econometric methods, and technical tools
- Heston, Alan, Robert Summers and Bettina Aten, 2002, Penn World Table Version 6.1 Center for International Comparisons at the University of Pennsylvania (CICUP), October.
- White, Halbert, 1980, “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity,” Econometrica 48, pp. 817–38.
- Pesaran, M. Hashem, 2003, “A Simple Panel Unit Root Test in the Presence of Cross Section Dependence,” University of Southern California and Cambridge University, unpublished manuscript.
- Arellano and Bond (1991) on panel specification tests; White (1980) on heteroskedasticity-consistent covariance estimation; Pesaran (2003) on panel unit root testing.

### Other notable methodological and theoretical contributions
- Grier, K.B. and G. Tullock, 1989, “An Empirical Analysis of Cross-National Economic Growth, 1951-80,” Journal of Monetary Economics, Vol. 24, pp. 259–76.
- Kormendi, R., and Meguire, P., 1985, “Macroeconomic Determinants of Growth: Crosscountry Evidence,” Journal of Monetary Economics, Vol. 16, pp. 141–63.
- Sachs et al. (2004) and Schumpeter, Joseph A., 1942, Capitalism, Socialism, and Democracy (New York, New York: Harper and Brothers).
- Sikimilar methodological and theoretical studies listed above including work by Gali and Hammour (1993), Siegler (2005), and others.

*References list for IMF Working Paper _wp05147 (References section).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05147.pdf_