## 2. Sovereign Debt Restructurings and Trade Data

## Source details

**Canonical URL:** [2. Sovereign Debt Restructurings and Trade Data](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025240-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025240-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025240-source-pdf.pdf.json)

---

### Introduction and objectives
- Study covers 194 restructurings with private external creditors spanning 76 countries over 1975−2019.
- Methodology:
  - Local Projections with the Augmented Inverse Probability Weighted (AIPW) method proposed by Jordà and Taylor (2016).
  - Classification of restructuring types follows Asonuma and Trebesch (2016) based on whether payments have been missed (post-default) or not (preemptive).
- Purpose:
  - Quantify cumulative effects on exports and imports and their sub-groups (manufactured vs primary commodity goods; capital, intermediate, consumption goods).
  - Use post-default vs preemptive nature of restructuring as an ex ante proxy for depth and duration of economic costs.
  - Apply AIPW estimation on subsamples differentiated by whether domestic aggregate demand-to-GDP in the year prior to restructuring is higher than the median.

### Sample, classification, and data
- Sample composition:
  - Total episodes: 194 across 76 countries, 1975−2019.
  - Post-default restructurings: 115 episodes.
  - Preemptive restructurings: 79 episodes.
  - Most post-default events involve a default at the start of the restructuring or in the same year: 106 episodes, "accounting for 92 percent of post-default restructurings".
- Definitions (Asonuma and Trebesch (2016) classification):
  - "Preemptive restructurings”: implemented when there are no missed payments or payments are missed only temporarily and after the start of negotiations with creditor committees.
  - "Post-default restructurings”: take place after a default event where payments have been missed without the agreement of creditors.
- Trade data and goods classification (UNCTAD production groups):
  - Aggregate groups:
    - "Manufactured goods": (i) chemicals and related products, (ii) manufactured goods excluding pearls, precious and semi-precious stones and non-ferrous metals, (iii) machinery and transport equipment, (iv) miscellaneous manufactured articles.
    - "Primary commodity goods": (i) food and live animals, (ii) beverages and tobacco, (iii) crude materials, inedible, except fuels, (iv) mineral fuels, lubricants and related materials, (v) animal and vegetable oils, fats and waxes, (vi) non-ferrous metals.
    - "Other goods": all remaining goods other than manufactured goods and primary commodity goods.
  - "Other goods" account for around 2 percent of aggregate imports and exports; analysis focuses mainly on manufactured goods and primary commodity goods.
  - Subcomponents:
    - "Capital goods": machinery and transport equipment.
    - "Intermediate goods": crude materials, mineral, and fuels, metals, iron, steel, cement, and all the products, including "parts".
    - "Consumption goods": food, beverages, tobacco, chemicals, textiles, furniture, footwear, office equipment, etc.
    - Manufactured goods can be classified among all three subgroups; primary commodities include only intermediate and consumption goods.
- Non-trade macro variables (annual, mostly from WDI):
  - Banking crisis variable (binary) from Laeven and Valencia (2013, 2020).
  - US federal fund rate from FRED.
  - Freedom House indices from Freedom Houses.
  - Real effective exchange rate (REER) from Darvas (2012, 2021).

### Baseline model and estimation approach
- Baseline model specification:
  - Two measures of trade explored: growth rate and level of cumulative change in exports (imports)-to-GDP ratio from pre-restructuring year (year 푡) to year 푡 + ℎ (measured in percent and percentage points, respectively).
  - Growth (log) specification (Equation 1):
    log(푦푖,푡+ℎ) − log(푦푖,푡) = 훼푖 푆,ℎ + 훽푆,ℎ 퐷푖,푡+1 푆 + 푋푖,푡 훽푆,ℎ + 휇푖,푡+ℎ 푆
  - Level specification (Equation 2):
    푦푖,푡+ℎ − 푦푖,푡 = 훼푖 푆,ℎ + 훽푆,ℎ 퐷푖,푡+1 푆 + 푋푖,푡 훽푆,ℎ + 휇푖,푡+ℎ 푆
  - Horizons: ℎ = 1,2,..,5. Strategies S ∈ {postdefault, preemptive}.
  - Notation preserved exactly for log(푦푖,푡+ℎ), 푦푖,푡+ℎ, 훼푖 푆,ℎ, 퐷푖,푡+1 푺, 퐗풊,풕, and conditional means 푚푗 푆,ℎ.
  - Illustrative example: exports decline from 10 percent of GDP to 8 percent of GDP → growth rate = -20 percent; level change = -2 percentage points.
- Baseline Local Projection controls (following Asonuma et al. (2024)):
  - (i) GDP growth,
  - (ii) government expenditure-to-GDP ratio,
  - (iii) trade openness (exports + imports / GDP ratio),
  - (iv) banking crisis dummy,
  - (v) bank credit-to-GDP ratio,
  - (vi) high inflation dummy (above 50 percent),
  - (vii) domestic aggregate demand-to-GDP ratio,
  - (viii) terms of trade,
  - (ix) REER.
- OLS local projections:
  - OLS estimates growth rate of cumulative change of imports (exports)-to-GDP ratio (aggregate, manufactured goods, primary commodity).
  - Effects of post-default and preemptive restructurings estimated simultaneously.
  - Sample trimming: observations below 1 percentile and above 99 percentile dropped.
  - Key OLS patterns:
    - Aggregate, manufactured goods, and primary commodity imports experience a sharp decline following post-default restructurings; moderate and gradual but not statistically significant decline following preemptive restructurings.
    - Exports remain broadly stable; manufactured exports show a larger increase for post-default restructurings and a larger decline for preemptive restructurings (mostly not statistically significant).
    - When measured as percentage of GDP, import declines are smaller and amount to a few percentage points of GDP on average.

### Augmented Inverse Probability Weighted (AIPW) methodology
- Purpose: attenuate bias from selection on observables since restructurings are associated with negative shocks and pre-existing conditions.
- Estimation steps:
  1. Apply OLS on baseline model (equations 1 and 2) to obtain estimated conditional means 푚푗 푆,ℎ.
  2. Apply probit regression (equation 3) to obtain predicted likelihood of each debt restructuring strategy 푝̂푖,푡+1 푆.
  3. Define AIPW estimator using both estimated conditional means and predicted likelihood as in Equation 4.
- AIPW estimator structure:
  - First square bracket inside summation = inverse probability weighting (IPW).
  - Second term inside summation = weighted average of the two predicted dependent variables.
- Probit first stage predictors (exclusive restrictions per Asonuma et al. (2024)):
  - (i) US federal funds rate;
  - (ii) contagion measure constructed from debt restructuring dummies of other debtor countries, weighted by geographical distances;
  - (iii) number of past preemptive restructuring choice.
- Probit performance:
  - ROC areas under the curve: 0.86 and 0.94.
  - ROC areas with and without predictors:
    - Post-default: 0.87 vs. 0.79.
    - Preemptive: 0.94 vs. 0.85.
  - Kernel density: predicted probabilities for treatment group are more equally distributed, implying weights (inverse predicted probability) assigned more equally.

### AIPW results — macroeconomic outcomes
- GDP and REER:
  - AIPW confirms a much higher decline in GDP following post-default restructurings.
  - REER: persistent and deep decline following post-default restructurings.
  - Preemptive restructurings: large initial REER impact but recovery, with effect mostly vanishing by the fifth year.

### AIPW results — imports (aggregate and by goods)
- Aggregate imports:
  - A 10-15 percent decline in aggregate imports-to-GDP ratio by year 3 following post-default restructurings.
  - Import compression measured as percentage of GDP: about 2 percentage points of GDP by year 2 following post-default restructurings.
  - Preemptive restructurings: imports remain stable; aggregate imports experience a modest increase by the end of the horizon (a 7 percent increase in the aggregate imports-to-GDP ratio).
  - Statistical significance:
    - Difference in response between post-default and preemptive restructurings is statistically significant under the Clogg et al (1995) test, but not statistically significant under the bootstrap approach.
- Manufactured vs primary commodity imports:
  - Initial declines by year 3 comparable for both categories following post-default restructurings.
  - Manufactured imports tend to recover by year 4.
  - Mechanisms:
    - Sharp decline in GDP reduces import demand.
    - Large REER depreciation increases import price in local currency, further reducing imports.
- Subcomponents (Section 4.3, Figure 5):
  - Manufactured goods imports compression following post-default restructurings mainly driven by capital goods and, to a smaller extent, intermediate goods; consumption goods remain more stable.
  - Primary commodity import declines following post-default restructurings driven by both intermediate and consumption goods.
  - For preemptive restructurings, imports of sub-groups remain more stable; manufactured goods increase toward end of horizon driven by capital goods.

### AIPW results — exports (aggregate and by goods)
- Aggregate exports:
  - Relatively stable following post-default restructuring; AIPW finds a small but significant increase of about 11 percent by year 4 when measured as percent change from initial GDP ratio.
  - Preemptive restructurings: aggregate exports experience a small but significant decline of about 5 percent (driven mostly by manufactured exports).
  - When measured as percentage change of GDP:
    - Effect of post-default on exports is much more muted and largely not statistically significant.
    - Preemptive restructurings associated with a significant decline in exports of about 1.7 percentage points of GDP, mostly driven by primary commodity exports.
- Manufactured vs primary commodity exports:
  - Manufactured exports: post-default shows a more pronounced increase; preemptive shows decline driven mostly by manufactured exports.
  - Primary commodity exports: broadly stable for post-default; preemptive declines driven mostly by primary commodity exports.
- Subcomponents (Section 4.3):
  - Increase in manufactured goods exports following post-default restructurings driven by capital goods.
  - Primary commodity exports remain stable post-default despite temporary increases in primary consumption good exports.
  - Preemptive restructurings lead to comparable declines across goods; primary commodity decline driven by intermediate goods.
- Statistical testing:
  - Differences between post-default and preemptive restructurings are statistically significant for all three export measures over most or the entire horizon under the Clogg et al. (1995) test but not statistically significant under bootstrapping.

### Role of domestic aggregate demand
- Definition and threshold:
  - Domestic aggregate demand = consumption + investment + government spending (equivalently, GDP minus net exports).
  - Sample median of domestic aggregate demand-to-GDP in the year prior to restructuring: 102 percent of GDP.
  - That threshold classifies half of post-default restructurings and one third of preemptive restructurings as starting from a high domestic aggregate demand.
- Interaction specification:
  - Equations (5) and (6) interact restructuring strategies with dummy 퐷퐴퐷 indicating high domestic aggregate demand (퐷
    푖,푡
    퐷퐴퐷
    = 1 if high, 0 otherwise).
  - 훤
    퐷퐴퐷,ℎ reflects impact when domestic aggregate demand-to-GDP is high.
  - 훤
    푁표푛_퐷퐴퐷,ℎ reflects impact when domestic aggregate demand-to-GDP is low.
- Main findings (Figure 7; Tables 3 and H6):
  - Imports:
    - Debtor countries with high domestic aggregate demand-to-GDP experience a larger contraction in imports.
    - Decline is stronger for post-default restructurings; preemptive restructurings with high domestic demand also experience a decline but smaller.
    - Difference driven mainly by both manufactured goods and primary commodities for post-default restructurings, and by primary commodities for preemptive restructurings.
    - Statistical significance: differences significant under Clogg et al (1995) for most horizons but not under bootstrapping.
  - Exports:
    - Increase in exports following post-default restructurings driven by the low domestic demand subsample; countries with high domestic demand experience a decline in exports following post-default restructurings.
    - Differences driven mostly by manufactured goods exports.
    - Statistical significance: differences significant under Clogg et al (1995) for most horizons but not under bootstrapping.
  - Robustness: OLS estimation results based on Table G6 (Figure G5 in Appendix G) are similar to baseline AIPW results.

### Robustness and inference
- Robustness checks reported:
  - Including hybrid post-default cases yields similar results to baseline strict classification.
  - Treating default year as start of restructuring yields comparable AIPW results.
  - Alternative approaches for overlapping restructurings reported.
- Inference and bootstrap details (Table 2 notes):
  - ℎ = 1 is the restructuring year.
  - 95% confidence intervals constructed by bootstrap with 1000 resamples using Stata’s bsample; some iterations dropped due to perfect prediction or insufficient observations.
  - Confidence intervals based on 962 sets of coefficients for GDP and 992 sets for investment for post-default; other variables based on 1000 sets.
  - Significance notation: * = 10% level, ** = 5% level, *** = 1% level.

### Main high-level findings and interpretation
- Imports:
  - Experience a large compression following a post-default restructuring: "15 percent decline in the ratio to GDP, or about 2 percentage points of GDP".
  - Remain broadly stable following a preemptive restructuring; preemptive restructurings can be associated with modest import increases (e.g., 7 percent increase in aggregate imports-to-GDP ratio by end of horizon).
  - Import compression is stronger when domestic aggregate demand is initially high.
- Exports:
  - Impact tends to be smaller in magnitude:
    - Post-default restructurings experience a small increase "of 0.4 percentage points of GDP" in some measures and about "11 percent by year 4" when measured as percent change from initial GDP ratio.
    - Preemptive restructurings experience a decline "of about 1.5 percentage points of GDP" in some measures and about "5 percent" decline in others.
  - Differential patterns suggest substitution from domestic to external markets can offset negative supply shocks in post-default cases.
  - Effects on exports are sensitive to measurement: percent-change metrics can show large increases from a small base, while percentage-point-of-GDP metrics are more muted.
- Goods composition:
  - Imports of manufactured goods and primary commodities experience similar declines following post-default restructurings.
  - Declines largely driven by:
    - Manufactured goods: capital goods and intermediate goods.
    - Primary commodities: consumption goods and intermediate goods.
  - Exports response more pronounced for manufactured goods, with the sharpest differential across restructuring types among capital goods.
- Mechanisms and interpretation:
  - Sharp GDP declines reduce import demand; REER depreciation raises local-currency import prices, amplifying import compression.
  - REER adjustment and reallocation to external markets can partially compensate for negative macro shocks, explaining export stability or increases post-default.
  - Deeper, more protracted crises (post-default) can lead to improvement in trade balance via import compression and export responses, but at welfare costs associated with large import declines and output losses.
  - Preemptive restructuring tends to require smaller external adjustment and may attenuate welfare-costly forced adjustment.

### Contributions to literature and policy implication
- Contributions:
  - Highlights differential impacts on imports vs exports, across types of goods, and dependence on whether restructuring involved missed payments.
  - Extends industry- and firm-level findings (e.g., financial dependence, firm-level product mix changes) to a large cross-country sample.
  - Links country-specific crisis drivers and the economy’s reliance on trade to amplify trade-driver effects.
- Policy implication:
  - How a debtor country handles its sovereign debt crisis has significant implications for trade and welfare.
  - Large swings in net imports following restructuring can proxy welfare impact and align with estimated output costs in the literature.
  - By restructuring preemptively, countries may attenuate impacts that would otherwise require welfare-costly forced external adjustment.

*Source: IMF Working Paper — Trade Dynamics in Sovereign Debt Crises (chapter "2. Sovereign Debt Restructurings and Trade Data" and start of section "3. Stylized Facts: Ordinary Least Square Local Projection Estimation").*

### 2. Sovereign Debt Restructurings and Trade Data ........................................................................

### 2. Sovereign Debt Restructurings and Trade Data

### Introduction and key objectives
- Study covers 194 restructurings with private external creditors spanning 76 countries over 1975−2019.
- Methodology:
  - Local Projections with the Augmented Inverse Probability Weighted (AIPW) method proposed by Jordà and Taylor (2016).
  - Classification of restructuring types follows Asonuma and Trebesch (2016) based on whether payments have been missed (post-default) or not (preemptive).
- Purpose:
  - Quantify cumulative effects on exports and imports and their sub-groups (manufactured vs primary commodity goods; capital, intermediate, consumption goods).
  - Use post-default vs preemptive nature of restructuring as an ex ante proxy for depth and duration of economic costs.
  - Apply AIPW estimation on subsamples differentiated by whether domestic aggregate demand-to-GDP in the year prior to restructuring is higher than the median.

### Main high-level findings (summary statements from the text)
- Imports:
  - Experience a large compression following a post-default restructuring: "15 percent decline in the ratio to GDP, or about 2 percentage points of GDP".
  - Remain broadly stable following a preemptive restructuring.
- Exports:
  - Impact tends to be smaller in magnitude:
    - Post-default restructurings experience a small increase "of 0.4 percentage points of GDP".
    - Preemptive restructurings experience a decline "of about 1.5 percentage points of GDP".
  - Differential patterns suggest substitution from domestic to external markets can offset negative supply shocks in post-default cases.
- Goods composition:
  - Imports of manufactured goods and primary commodities experience similar declines following restructurings.
  - Declines are largely driven by:
    - Manufactured goods: capital goods and intermediate goods.
    - Primary commodities: consumption goods and intermediate goods.
  - Exports response is more pronounced for manufactured goods, with the sharpest differential across restructuring types among capital goods.
  - When measured as a percentage of GDP, effects and differences across goods become more muted (large responses from a relatively small base).
- Role of domestic aggregate demand:
  - Import compression is stronger when domestic aggregate demand is initially high.

### Contributions to literature (as stated)
- Sharpens focus on differential impacts on imports vs exports, across types of goods, and dependence on whether restructuring involved missed payments.
- Extends industry- and firm-level findings (e.g., financial dependence, firm-level product mix changes) to a large cross-country sample.
- Links country-specific crisis drivers and the economy’s reliance on trade to amplify trade-driver effects.

### Classification of sovereign debt restructurings (definitions and sample)
- Definitions (Asonuma and Trebesch (2016) classification):
  - "Preemptive restructurings” are implemented when there are no missed payments or payments are missed only temporarily and after the start of negotiations with creditor committees.
  - "Post-default restructurings” take place after a default event where payments have been missed without the agreement of creditors.
- Sample composition:
  - Total episodes: 194 across 76 countries, 1975−2019.
  - Post-default restructurings: 115 episodes.
  - Preemptive restructurings: 79 episodes.
  - Most post-default events involve a default at the start of the restructuring or in the same year: 106 episodes, "accounting for 92 percent of post-default restructurings".
- Episode treatment:
  - The start of debt restructuring is defined as the earlier of (i) the month in which a default occurs or (ii) the month when a distressed restructuring is publicly announced.
  - For preemptive debt restructurings, the year that debtor countries announce debt restructuring is the start (start of estimation horizon in analysis t=1).
  - Baseline sample drops hybrid cases (where restructuring is announced first and default occurs later) and focuses on post-default episodes where default happens in the year of start of the restructuring.
  - Robustness checks consider alternative classifications, including hybrid episodes and alternative handling of overlapping episodes (e.g., "initial strategy" vs "worst strategy").

### Trade data and goods classification
- Data source: annual trade dataset from UNCTAD (production groups and composition classification).
- Aggregate groups:
  - "Manufactured goods":
    - (i) chemicals and related products,
    - (ii) manufactured goods excluding pearls, precious and semi-precious stones and non-ferrous metals,
    - (iii) machinery and transport equipment,
    - (iv) miscellaneous manufactured articles.
  - "Primary commodity goods":
    - (i) food and live animals,
    - (ii) beverages and tobacco,
    - (iii) crude materials, inedible, except fuels,
    - (iv) mineral fuels, lubricants and related materials,
    - (v) animal and vegetable oils, fats and waxes,
    - (vi) non-ferrous metals.
  - "Other goods": all remaining goods other than manufactured goods and primary commodity goods.
- Share and focus:
  - "Other goods" account for around 2 percent of aggregate imports and exports, so analysis mainly focuses on manufactured goods and primary commodity goods.
- Subcomponents (further distinctions):
  - "Capital goods": machinery and transport equipment.
  - "Intermediate goods": crude materials, mineral, and fuels, metals, iron, steel, cement, and all the products, including "parts" in the descriptions.
  - "Consumption goods": food, beverages, tobacco, chemicals, textiles, furniture, footwear, office equipment, etc.
  - Manufactured goods can be classified among all three subgroups; primary commodities include only intermediate and consumption goods.
  - No subcomponents available for "other goods".
- Non-trade macro variables (annual, mostly from WDI):
  - Banking crisis variable (binary) from Laeven and Valencia (2013, 2020).
  - US federal fund rate from FRED.
  - Freedom House indices from Freedom Houses.
  - Real effective exchange rate (REER) from Darvas (2012, 2021).

### Stylized facts and OLS Local Projection estimation (setup and sample patterns)
- Outcome presentation:
  - Averages of cumulative percentage changes from the level in the pre-restructuring year (t=0, one year before start of debt restructuring) for aggregate imports and exports, GDP and REER are reported.
  - Series are adjusted for differences in country means (equivalent to a country-specific linear trend in levels) for the sample period.
  - Aggregate imports and exports measured as a share of GDP.
  - Debt restructurings occur at t=1.
- Sample mean patterns (Figure 1 summary):
  - Both aggregate imports and GDP (measured relative to pre-crisis trend) experience a severe decline after post-default restructurings, while decline is milder following preemptive restructurings (especially for GDP).
  - Aggregate exports decline in the first 2 years after post-default restructurings but stabilize afterwards; for preemptive restructurings the decline continues.
  - REER experiences a severe decline over 5 years in both post-default and preemptive restructurings, sharper in the first two years after preemptive restructurings.
- Estimation approach:
  - Local Projection approach (Jordà (2005)) used to estimate total cumulative impact over a horizon after sovereign debt crisis events.
  - Exports and imports are influenced directly and indirectly through macroeconomic variables (GDP, REER) that are driven by the debtor’s restructuring choice.
- Baseline Local Projection controls (following Asonuma et al. (2024)):
  - (i) GDP growth,
  - (ii) government expenditure-to-GDP ratio,
  - (iii) trade openness (exports + imports / GDP ratio),
  - (iv) banking crisis dummy,
  - (v) bank credit-to-GDP ratio,
  - (vi) high inflation dummy (above 50 percent),
  - (vii) domestic aggregate demand-to-GDP ratio,
  - (viii) terms of trade,
  - (ix) REER.
  - Controls (i)–(vi) are the same as in Asonuma et al. (2024); controls (vii)–(ix) are main trade drivers used in the literature.

*Source: IMF Working Paper — Trade Dynamics in Sovereign Debt Crises (chapter "2. Sovereign Debt Restructurings and Trade Data" and start of section "3. Stylized Facts: Ordinary Least Square Local Projection Estimation").*

### Appendix A2 reports summary statistics of the controls. Table F1 in Appendix F shows differences in

### wpiea2025240-source-pdf - Appendix A2 reports summary statistics of the controls. Table F1 in Appendix F shows differences in

### Baseline model specification
- Baseline follows Jordà and Taylor (2016), Jordà et al. (2016), and Kuvshinov and Zimmermann (2018).
- Two measures of trade are explored: growth rate and level of cumulative change in exports (imports)-to-GDP ratio from the pre-restructuring year (year 푡) to year 푡 + ℎ (measured in percent and percentage points, respectively).
- Estimation equations:
  - Growth (log) specification (Equation 1):
    log(푦푖,푡+ℎ) − log(푦푖,푡) = 훼푖 푆,ℎ + 훽푆,ℎ 퐷푖,푡+1 푆 + 푋푖,푡 훽푆,ℎ + 휇푖,푡+ℎ 푆
  - Level specification (Equation 2):
    푦푖,푡+ℎ − 푦푖,푡 = 훼푖 푆,ℎ + 훽푆,ℎ 퐷푖,푡+1 푆 + 푋푖,푡 훽푆,ℎ + 휇푖,푡+ℎ 푆
  - Horizons: ℎ = 1,2,..,5. Strategies S ∈ {postdefault, preemptive}.
- Definitions and notation preserved exactly:
  - log(푦푖,푡+ℎ) and 푦푖,푡+ℎ denote log and level of exports- and imports-to-GDP ratio of country 푖 in year 푡 + ℎ.
  - 훼푖 푆,ℎ denotes country fixed effects.
  - 퐷푖,푡+1 푺 is a dummy taking unity if there is a type 푆 debt restructuring in year 푡+1; 훽푆,ℎ is its coefficient.
  - 푿풊,풕 denotes the vector of control variables with coefficients 휷푺,풉.
  - Conditional mean of dependent variable denoted by 푚푗 푆,ℎ for state 푗 = 1, 0 corresponding to “restructuring” and “no restructuring”, respectively.
- Illustrative numeric example preserved:
  - When exports decline from 10 percent of GDP to 8 percent of GDP, the growth rate of the cumulative change in exports-to-GDP ratio is -20 percent, while the change in the level of exports-to-GDP corresponds to -2 percentage points.

### OLS local projections (Figure 2 and Appendices)
- Methodology:
  - OLS local projections estimate growth rate of cumulative change of imports (exports)-to-GDP ratio (aggregate, manufactured goods, primary commodity).
  - Effects of post-default (red lines) and preemptive restructurings (blue lines) estimated simultaneously by including all dummies in one regression.
  - Restructuring samples across panels are balanced; 95 percent confidence intervals shown; observations below 1 percentile and above 99 percentiles dropped.
- Key OLS findings (Figure 2; Tables G1−G3; Figures E4−E7):
  - Aggregate, manufactured goods, and primary commodity imports:
    - All experience a sharp decline following a post-default restructuring (red lines in Panels A−C).
    - Negative values indicate cumulative percent change of imports-to-GDP ratio compared to country-specific pre-restructuring level.
    - Magnitude of initial decline is similar across both groups of goods following post-default restructurings.
    - Imports experience a moderate and gradual but not statistically significant decline following preemptive restructurings (blue lines).
  - Note: Equation 1 estimates a log change, so aggregate import decline is a weighted average (not a simple sum) of subcomponents.
  - When measured as percentage of GDP (Figure G2), results are similar but magnitude of decline is smaller; import declines only amount to a few percentage points of GDP on average.
  - Exports:
    - Remain broadly stable (Panel D), with a smaller decline for post-default restructurings; most changes not statistically significant except a small decline among preemptive restructurings at the end of the horizon.
    - Manufactured exports (Panel E): point to a larger increase for post-default restructurings and a larger decline for preemptive restructurings, albeit not statistically significant.
    - Primary commodity exports (Panel F): largely mirror aggregate exports.
    - When measured by level of cumulative change (Figure G2), aggregate export results similar but difference between manufactured and primary commodity exports is muted with wide confidence intervals.
- Statistical testing reported in Tables G1−G3:
  - Conventional t-tests for significance of coefficients.
  - Two comparison tests for differences between post-default and preemptive point estimates: bootstrap estimation and a Clogg et al (1995) statistic.
  - Bootstrapped results constructed by resampling original data and reporting percentile-based confidence intervals across 1000 draws.

### Augmented Inverse Probability Weighted (AIPW) estimation — Overview
- Purpose: attenuate bias from selection on observables because restructurings are associated with negative shocks and pre-existing conditions.
- AIPW differs from OLS by assigning weights across observations: lower weights on observations more likely to be associated with a restructuring and higher weights to those less likely.
- Estimation steps:
  1. Apply OLS on baseline model (equations 1 and 2) to obtain estimated conditional means 푚푗 푆,ℎ.
  2. Apply probit regression (equation 3) to obtain predicted likelihood of each debt restructuring strategy 푝̂푖,푡+1 푆.
  3. Define AIPW estimator using both estimated conditional means and predicted likelihood as in Equation 4 (preserved exactly in notation).
- AIPW estimator (Equation 4) structure preserved:
  - First square bracket inside summation = inverse probability weighting (IPW).
  - Second term inside summation = weighted average of the two predicted dependent variables.

### Probit regression on debt restructurings (first stage for AIPW)
- Probit specification uses data in the year of the start of restructuring and the previous year.
- Predictors (exclusive restrictions) follow Asonuma et al. (2024):
  - (i) the US federal funds rate;
  - (ii) a contagion measure constructed from the debt restructuring dummies of other debtor countries, weighted by geographical distances;
  - (iii) the number of past preemptive restructuring choice.
- Characteristics:
  - Both the US federal funds rate and other countries’ restructurings are orthogonal to debtor i’s restructuring choice in year t; past preemptive restructurings are predetermined in year t.
- Probit regression results and predictive power:
  - Estimated probabilities from Equation 3 produce predicted likelihood 푝̂푖,푡+1 푆.
  - Receiver Operating Characteristic (ROC) areas under the curve: 0.86 and 0.94, indicating fairly good classification power.
  - ROC areas with and without predictors:
    - Post-default: 0.87 vs. 0.79.
    - Preemptive: 0.94 vs. 0.85.
  - Predictors significantly improve classification power for both restructuring strategies.
  - Kernel density estimates show predicted probabilities for treatment group are more equally distributed, implying weights (inverse predicted probability) assigned more equally across predicted probabilities.
- Table 1 and related figures:
  - Table 1 reports probit regression results; Columns 3−5 present probit regression results for alternative classifications.
  - Panels (I) and (II) in Figure B1 in Appendix B1 show ROC and kernel density for three alternative classifications: (a) post-default episodes based on default year, (b) post-default episodes including those after the start of restructuring, (c) all debt restructuring episodes.

### AIPW results on macroeconomic and trade outcomes (Figures 4, H1; Table 2; Tables H1−H2)
- GDP and REER:
  - AIPW confirms a much higher decline in GDP following post-default restructurings.
  - REER: AIPW points to a persistent and deep decline in REER following post-default restructurings.
  - Preemptive restructurings: large initial REER impact but recovery, with effect mostly vanishing by the fifth year.
- Imports (AIPW Panels I-A−I-C and Panel II):
  - Aggregate, manufactured goods, and primary commodity imports all experience a sharp decline following post-default restructuring.
  - Quantified magnitudes:
    - A 10-15 percent decline in aggregate imports-to-GDP ratio by year 3 following post-default restructurings.
    - Import compression when measured as percentage of GDP: about 2 percentage points of GDP by year 2 following post-default restructurings.
  - Manufactured vs primary commodity imports:
    - Initial declines by year 3 comparable for both; manufactured imports tend to recover by year 4.
  - Mechanisms:
    - Sharp decline in GDP reduces import demand.
    - Large REER depreciation increases import price in local currency, further reducing imports.
  - Preemptive restructurings:
    - Imports remain stable; aggregate imports and manufactured imports experience a modest increase by the end of the estimation horizon, with a 7 percent increase in the aggregate imports-to-GDP ratio.
  - Statistical significance:
    - Difference in response between post-default and preemptive restructurings is statistically significant under the Clogg et al (1995) test, but not statistically significant under the bootstrap approach.
- Exports (AIPW Panels I-D, I-E, I-F and Panel II):
  - Aggregate exports:
    - Relatively stable following post-default restructuring, but eventually experience a small but significant increase of about 11 percent by year 4 when measured as percent change from initial GDP ratio (red line in Panel I-D).
    - For preemptive restructurings, aggregate exports experience a small but significant decline of about 5 percent (driven mostly by manufactured exports).
  - Manufactured exports:
    - Post-default: more pronounced increase.
    - Preemptive: decline driven mostly by manufactured exports.
  - Primary commodity exports:
    - Remain broadly stable for post-default; preemptive declines in exports as percent of GDP driven mostly by primary commodity exports.
  - When measured as percentage change of GDP (Panel II):
    - Effect of post-default on exports is much more muted and largely not statistically significant.
    - Preemptive restructurings associated with a significant decline in exports of about 1.7 percentage points of GDP (blue line in Panel II-D), mostly driven by primary commodity exports (blue line in Panel II-F).
  - Statistical testing:
    - Differences between post-default and preemptive restructurings are statistically significant for all three export measures over most or the entire horizon under the Clogg et al. (1995) test but not statistically significant under bootstrapping approach.
  - Interpretation:
    - Results suggest negative macroeconomic impacts (sharp GDP decline) can be more than compensated by REER adjustment and reallocation to external sector where possible.
    - Findings for exports are less robust than for imports due to competing channels and sensitivity to measurement (percent change vs percentage points of GDP) and the relatively closed nature of the economies in the sample.
- Robustness checks:
  - Using full sample of post-default restructurings including hybrid cases (Figure B4 in Appendix B1) yields results similar to baseline strictly post-default cases.
  - Treating the default year as the start of restructuring (Figure B7 in Appendix B2) yields comparable AIPW results.
  - Alternative approaches for overlapping debt restructurings (Figures C1–C2 in Appendix C) also reported.

*Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025240-source-pdf.pdf*

### Section 2.1. Both figures are similar to our baseline results. Lastly, when we use a median regression, the

### Section 2.1.

### AIPW Local Projection Results (Figure 4 and Table 2)
- The analysis presents AIPW local projection estimates; Year 1 is the restructuring year.
- Dependent variables:
  - Panel (I): growth rate of cumulative change of imports (exports)-to-GDP ratio (aggregate, manufactured goods, primary commodity).
  - Panel (II): level of cumulative change of imports (exports)-to-GDP ratio (aggregate, manufactured goods, primary commodity).
- Effects of post-default (red lines) and preemptive restructurings (blue lines) are estimated simultaneously by including all restructuring dummies in one regression equation.
- Restructuring samples across all panels are balanced to examine effects of the same restructuring episodes.
- Confidence and sample trimming:
  - Gray bands and dashed lines indicate the 95 percent confidence intervals.
  - Observations below the 1 percentile and above the 99 percentile are dropped.
- Table 2 notes on estimation and inference:
  - ℎ = 1 is the restructuring year.
  - All regressions include the control variables shown in Table 1.
  - Effects of the two restructurings are estimated separately by estimating AIPW for each type (respective other restructuring observations are dropped so control group includes only non-restructuring observations).
  - 95% confidence intervals are constructed by bootstrap, resampling the sample used for the corresponding AIPW 1000 times using Stata’s bsample code and reporting percentiles from the distribution of estimates.
  - Out of the 1000 iterations, a few iterations led to no estimates due to either perfect prediction in the first stage or insufficient observations; as a result:
    - Confidence intervals are based on 962 sets of coefficients for GDP and 992 sets of coefficients for investment for post-default.
    - Other variables are based on 1000 sets of coefficients.
  - Significance notation:
    - * indicates statistical significance at the 10% level.
    - ** indicates statistical significance at the 5% level.
    - *** indicates statistical significance at the 1% level.

### Subcomponents of Manufactured and Primary Commodity Goods (Section 4.3, Figure 5)
- Imports (growth rate of cumulative change):
  - Larger compression in manufactured goods imports following a post-default restructuring is mostly driven by the decline in capital goods and to a smaller extent by intermediate goods; imports of consumption goods remain more stable (red lines in Panels B-i−iii in section (I)).
  - Manufactured goods imports remain stable following preemptive restructurings, except for a dip in intermediate goods; bulk of the increase in manufactured goods imports towards end of estimation horizon driven by capital good imports (blue line in Panel B-i in section (I)).
  - Decline in primary commodity imports following post-default restructurings is driven by both intermediate and consumption goods; imports of all sub-groups remain more stable following preemptive restructuring (blue lines in panel C in section (I)).
  - Results are broadly similar when measured as a percentage change in GDP terms (Figure H2 in Appendix H).
- Exports (growth rate of cumulative change):
  - Increase in manufactured goods exports following post-default restructurings is driven by capital goods (red line in Panel E-I in section (II)).
  - All groups of goods show a comparable decline following preemptive restructurings (blue lines in Panels E-i−iii).
  - Primary commodity exports following a post-default restructuring remain stable despite a temporary increase in primary consumption good exports (red line in Panel F-iii in section (II)).
  - Primary commodity exports experience a small decline following preemptive restructurings, driven by intermediate good exports (blue line in Panels F-ii in section (II)).
  - When change in exports is measured as a percentage change in GDP terms (Figure H2 in Appendix H), the impact of post-default restructurings becomes fairly muted; the only clear pattern involves a significant decline in primary commodity exports (about 1 percentage point of GDP) driven by intermediate goods following preemptive restructurings.

### Role of Domestic Aggregate Demand (Section 4.4, Figure 6 and 7; Tables 3 and H5–H6)
- Definition and distribution:
  - Domestic aggregate demand = consumption + investment + government spending (equivalently, GDP minus net exports).
  - Figure 6 reports distribution of domestic aggregate demand-to-GDP ratio in the year prior to debt restructuring for the two types of restructurings.
  - Sample median of domestic aggregate demand of all restructuring observations is 102 percent of GDP (shown in red dashed lines in Figure 6).
  - That threshold classifies half of post-default restructurings and one third of preemptive restructurings as starting from a high domestic aggregate demand.
  - The threshold of 102 percent of GDP remains robust when computed using the three-year average of domestic aggregate demand-to-GDP prior to restructuring.
- Regression specification (interaction with high vs low domestic aggregate demand):
  - Equations (5) and (6) interact restructuring strategies with a dummy 퐷퐴퐷 indicating high domestic aggregate demand (퐷
    푖,푡
    퐷퐴퐷
    = 1 if high, 0 otherwise).
  - Coefficients:
    - 훤
      퐷퐴퐷,ℎ
      reflects impact of restructuring strategy S when domestic aggregate demand-to-GDP is high.
    - 훤
      푁표푛_퐷퐴퐷,ℎ
      reflects impact when domestic aggregate demand-to-GDP is low.
  - Estimation procedure mirrors baseline AIPW: OLS for conditional means, probit for predicted likelihoods (equation 3), then AIPW combining both.
- Main AIPW local projection findings (Figure 7; Tables 3 and H6):
  - Imports (Panel I):
    - Debtor countries with high domestic aggregate demand-to-GDP (purple lines) experience a larger contraction in imports.
    - Decline is stronger for post-default restructurings; preemptive restructurings with high domestic demand also experience a decline but smaller in magnitude.
    - Difference in import compression between high vs low domestic demand is driven mainly by both manufactured goods and primary commodities for post-default restructurings, and by primary commodities for preemptive restructurings.
    - Statistical significance:
      - Difference in response between high vs low aggregate demand subsamples for post-default and preemptive restructurings is statistically significant for most estimation horizons (except years 3 and 5 for preemptive restructurings) under the Clogg et al (1995) tests, but not statistically significant under bootstrapping.
      - Difference in response between post-default and preemptive restructurings for both high and low aggregate demand countries is statistically significant for most estimation horizons (except year 1 for high aggregate demand countries, and years 1 and 3 for low aggregate demand countries) under Clogg et al (1995), but not under bootstrapping.
    - Effects are broadly similar when change in imports is measured as a percentage of GDP (Figure H3 in Appendix H), with a more pronounced increase in manufacturing imports following preemptive restructurings with high domestic aggregate demand.
  - Exports (Panel II):
    - Increase in exports following post-default restructurings is driven by the low domestic demand subsample (black line); countries with high domestic demand experience a decline in exports following post-default restructurings.
    - Preemptive restructurings show more stable exports in their aftermath.
    - Differences are driven mostly by manufactured goods exports (Panel II-B); primary commodities show a more similar response initially but divergence becomes pronounced toward the end of the estimation horizon for post-default restructurings.
    - Statistical significance:
      - Difference in response between high vs low aggregate demand subsamples is statistically significant for most horizons (except years 2, 4 and 5 for preemptive restructurings) under Clogg et al (1995), but not under bootstrapping.
      - Difference between post-default and preemptive restructurings for high and low aggregate demand countries is statistically significant for most horizons (except years 1 and 5 for high aggregate demand countries, and years 1−2 for low aggregate demand countries) under Clogg et al (1995), but not under bootstrapping.
    - Effects are qualitatively similar when change in exports is measured as a percentage of GDP (Figure H3 in Appendix H).
- Robustness:
  - Figure G5 in Appendix G reports OLS estimation results based on Table G6 in Appendix G; OLS results are similar to baseline AIPW results.

### Conclusion (Section 5)
- Sovereign debt restructurings have large adverse impacts on GDP, aggregate demand, exchange rates, and trade.
- Main trade findings:
  - Imports (relative to GDP) suffer a large compression following a post-default restructuring but remain broadly stable following a preemptive restructuring.
  - Import compression is stronger when domestic aggregate demand is initially high.
  - Exports: post-default restructurings tend to be associated with an increase in exports as a percent increase from the initial ratio to GDP, though that effect becomes much smaller when measured in percentage terms of GDP.
    - Interpretation: large percent increases in exports often occur from a relatively low base because many economies experiencing restructurings had relatively small trade to begin with.
- Policy implication:
  - How a debtor country handles its sovereign debt crisis has significant implications for trade and welfare.
  - Large swings in net imports following restructuring can proxy welfare impact and align with estimated output costs in the literature.
  - Deeper, more protracted crises can lead to an improvement in the trade balance, but typically at the cost of higher import compression and exports supported by large real depreciation.
  - Milder crises, proxied here by preemptive restructurings, tend to require smaller external adjustment and lead to lower net exports changes.
  - By restructuring preemptively, countries may attenuate impacts that would otherwise require welfare-costly forced external adjustment.

*IMF Working Paper excerpt from the provided content unit.*

### References

### wpiea2025240-source-pdf - References

### Major thematic clusters represented
- Sovereign default costs, restructuring, and renegotiation
  - Borenstein, Eduardo, and Ugo Panizza, 2009, “The Costs of Sovereign Default,” IMF Staff Papers, Vol.56(4), pp.683–741.
  - Asonuma, Tamon, Marcos Chamon, Aitor Erce, and Akira Sasahara, 2024, “Costs of Sovereign Debt Crises: Restructuring Strategies and Bank Intermediation,” Journal of International Economics. Vol.152, pp.104002.
  - Asonuma, Tamon, and Christoph Trebesch, 2016, “Sovereign Debt Restructurings: Preemptive or Post-default,” Journal of the European Economic Association, Vol.14(1), pp.175–214.
  - Hébert, Benjamin, and Jesse Schreger, 2017, “The Costs of Sovereign Default: Evidence from Argentina,” American Economic Review, Vol.107(10), pp.3119–3145.
  - Trebesch, Christoph, and Michael Zabel, 2017, “The Output Costs of Hard and Soft Sovereign Default,” European Economic Review, Vol.92, pp.416–432.
  - Yue, Vivian. Z., 2010, “Sovereign Default and Debt Renegotiation,” Journal of International Economics, Vol.80(2), pp.176–187.
  - Kuvshinov, Dmitry, and Kaspar Zimmermann, 2019, “Sovereign Going Bust: Estimating the Cost of Default,” European Economic Review, Vol.119, pp.1–21.
  - Gordon, Grey, and Pablo Guerron-Quintana, 2019, “A Quantitative Theory of Hard and Soft Sovereign Defaults,” Working Paper, Federal Reserve Bank of Richmond and Boston College.
  - Bricongne, Jean-Charles, Juan Carluccio, Lionel Fontagné, Guillaume Gaulier, and Sebastian Stumpner, 2025, “From Macro to Micro: Large Exporters Coping with Global Crises,” Journal of International Economics, Vol.153, pp.104037.
  - Caselli, Francesca, Matilde Faralli, Paolo Manasse, and Ugo Panizza, forthcoming, “On the Benefits of Repaying,” Journal of the European Economic Association.

- Trade dynamics and adjustment during crises, devaluations, and defaults
  - Eaton, Jonathan, Samuel Kortum, Brent Neiman, and John Romalis, 2016, “Trade and the Global Recession.” American Economic Review, Vol.106(11), pp.3401–3438.
  - Benguria, Felipe, and Alan M. Taylor, 2020, “After the Panic: Are Financial Crises Demand or Supply Shocks? Evidence from International Trade,” American Economic Review: Insights, Vol.2(4), pp.509–526.
  - Gopinath, Gita, and Brent Neiman, 2014, “Trade Adjustment and Productivity in Large Crises,” American Economic Review, Vol.104(3), pp.793–831.
  - Kohn, David, Fernando Leibovici, Michal Szkup, 2020, “Financial Frictions and Export Dynamics in Large Devaluations,” Journal of International Economics, Vol.122, pp.103257.
  - Alessandria, George, Sangeeta Pratap, and Vivian Z. Yue, 2020, “Export Growth and Aggregate Dynamics in Large Devaluations,” Working Paper, University of Rochester, Hunger Colleague and Graduate Center CUNY, and Emory University.
  - Martinez, Jose Vicente, and Guido Sandleris, 2011, “Is It Punishment? Sovereign Defaults and the Decline in Trade,” Journal of International Money and Finance Vol.30(6), pp.909–930.
  - Gu, Grace W., 2019, “Sovereign Default, Trade, and Terms of Trade,” Macroeconomic Dynamics, Vol.25(6), pp.1346–1380.
  - Essers, Dennis, Silvia Marchesi, Nejat Okatan, 2025, “Heterogeneous Trade Outcomes of Sovereign Defaults: External versus Domestic,” Working Paper, National Bank of Belgium and University of Milano-Bicocca.
  - Behrens, Kristian, Gregory Corcos, and Giordano Mion, 2023, “Trade Crisis? What Trade Crisis?” The Review of Economics and Statistics, Vol.95(2), pp.702–709.
  - Hummels, David, and Kwan Yong Lee, 2018, “The Income Elasticity of Import Demand: Micro Evidence and an Application,” Journal of International Economics, Vol.113, pp.20−34.

- Banking, financial crises, and interactions with sovereign risk
  - Laeven, Luc, and Fabian Valencia, 2013, “Systemic Banking Crises Database,” IMF Economic Review, Vol.61(2), pp.225–270.
  - Laeven, Luc, and Fabian Valencia, 2020, “Systemic Banking Crises Database II,” IMF Economic Review, Vol.68, pp.307–361.
  - Jordá, Óscar, Moritz Schularick, and Alan M. Taylor, 2016, “Sovereigns versus Banks: Credit, Crises, and Consequences,” Journal of the European Economic Association, Vol.14(1), pp.45–79.
  - Mendoza, Enrique G., and Vivian Z. Yue, 2012, “A General Equilibrium Model of Sovereign Default and Business Cycles,” The Quarterly Journal of Economics, Vol.127(2), pp.889–946.
  - Auerbach, Alan J., and Yuriy Gorodnichenko, 2016, “Effects of Fiscal Shocks in a Globalized World,” IMF Economic Review, Vol.64(1), pp.177–215.
  - Manasse, Paolo, and Nouriel Roubini, 2009, ““Rules of Thumb” for Sovereign Debt Crises,” Journal of International Economics Vol.78(2), pp.192–205.

- Empirical methods, inference, and data resources
  - Jordá, Óscar, 2005, “Estimation and Inference of Impulse Responses by Local Projections,” American Economic Review, Vol.95(1), pp.161–182.
  - Horowitz, Joel L., 2019, “Bootstrap Methods in Econometrics,” Annual Review of Economics, Vol.11, pp.193–224.
  - Clogg, Clifford C., Eva Petkova, and Adamantios Haritou, 1995, “Statistical Methods for Comparing Regression Coefficients Between Models,” American Journal of Sociology, Vol.100(5), pp.1261−1293.
  - Darvas, Zsolt, 2012, “Real Effective Exchange Rates for 178 Countries: A New Database,” Working Paper 06/2012, Bruegel.
  - Darvas, Zsolt, 2021, “Timely Measurement of Real Effective Exchange Rates,” Working Paper 15/2021, Bruegel.
  - Zymek, Robert, 2012, “Sovereign Default, International Lending, and Trade,” IMF Economic Review, Vol.60(3), pp.365–394.
  - Kohlscheen, Emanuel, and Stephen A. O’Connell, 2008, “Trade Credit and Sovereign Debt,” Working Paper, University of Warwick.
  - Tomz, Michael, and Mark L.J. Wright, 2007, “Do Countries Default in “Bad Times”?” Journal of the European Economics Association, Vol.5(2–3), pp.352–360.
  - Jordá, Óscar, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, and Alan M. Taylor, 2019, “The Rate of Return on Everything, 1870–2015,” Quarterly Journal of Economics, Vol.134(3), pp.1225–1298.

### Methodological and disciplinary breadth signaled by the references
- Empirical micro and macro studies: e.g., Altonji et al., 2005; Hummels and Lee, 2018.
- General equilibrium and quantitative modeling: e.g., Mendoza and Yue, 2012; Gordon and Guerron-Quintana, 2019.
- Databases and systematic crisis documentation: Laeven and Valencia, 2013 and 2020; Darvas, 2012 and 2021.
- Historical and cross-country perspectives: Reinhart and Rogoff, 2009; Jordá et al., 2019.

### Representative papers on international trade and sovereign risk linkages
- Rose, Andrew K., 2005, “One Reason Countries Pay Their Debts: Renegotiation and International Trade,” Journal of Development Economics, Vol.77(1), pp.189–206.
- Levy-Yeyati, Eduardo, and Ugo Panizza, 2011, “The Elusive Costs of Sovereign Defaults,” Journal of Development Economics, Vol.94(1), pp.95–105.
- Bricongne et al., 2025, “From Macro to Micro: Large Exporters Coping with Global Crises,” Journal of International Economics, Vol.153, pp.104037.
- Serfaty, Charles, 2024, “Sovereign Default and International Trade,” IMF Economic Review, Vol.72, pp.1449–1501.

*Source: wpiea2025240-source-pdf - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025240-source-pdf.pdf_