## Annex 1. Data Source and Summary Statistics (_wp16204)

## Source details

**Canonical URL:** [Annex 1. Data Source and Summary Statistics (_wp16204)](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16204.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16204.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16204.pdf.json)

---

### I. Introduction: purpose and methodological innovations
- Purpose: offer a template for quantifying benefits and costs of corporate debt restructuring based on a transparent set of assumptions.
- Context: corporate debt overhang is a feature of multiple crises and growth slowdowns, most recently in Europe and some emerging markets.
- Three methodological innovations:
  - Refining the ICR measure:
    - Introduce a “persistent ICR” P-ICR<1 measure capturing firms with ICR<1 over three consecutive years.
    - The P-ICR<1 criterion covers only about a half of the firms with ICR<1.
  - Evidence-based assumptions on creditor losses and employment impact:
    - Refine the standard “rule of thumb” 60 percent loss-given-default assumption to account for ownership of debt (bank-based restructurings more efficient), type of restructuring (out-of-court more efficient), and business/industry cycle stage.
    - Historic corporate restructurings in advanced economies suggest average employment losses of 20-25 percent.
  - Econometric estimation of debt overhang effects:
    - Firms with debt overhang invest less and hire less.
    - Removing debt overhang can boost the investment ratio (as a share of fixed assets) and the hiring ratio (as a share of pre-existing employment) by 2.5-3 percentage points each year.
- Key qualitative insight: corporate debt restructurings “pay off” in the medium-term — economic costs to creditors and employment impacts are fully offset by more rapid output growth and higher hiring under the paper’s assumptions.
- Data source and scope:
  - Use the Orbis dataset; Orbis covers 60 percent of Korean firms by volume of corporate debt in 2012-2014.
  - When both consolidated and unconsolidated statements exist for a firm, only unconsolidated statements are used to avoid double counting.
  - Financials and utilities are excluded (utilities’ “cost-plus” pricing makes ICR uninformative).

### II. Corporate debt in Korea: levels, composition, and vulnerabilities
- Corporate debt level and currency composition:
  - Corporate debt in Korea is about 100 percent of GDP.
  - As of end-2015, only 5 percent of corporate debt was foreign currency-denominated.
  - Comparable economies’ foreign currency-denominated corporate debt median: 10 percent; large emerging market economies: 20-70 percent.
- Export intensity and revenue exposure:
  - Korea’s net exports to GDP was over 6 percent in 2015.
  - Share of export revenue in total revenue of Korean firms: 55 percent in manufacturing overall; 65 percent in export-oriented manufacturing sectors (automobiles, electronics, petrochemicals, shipbuilding and steel).
  - Growth rate of Korean corporate earnings, particularly in export-oriented manufacturing and shipping, has been declining and recently turned negative.
- Interest coverage ratio (ICR) considerations:
  - ICR benchmarks examples:
    - ICR<1.5 might be appropriate for manufacturing that needs free cash flow.
    - ICR<1 might be appropriate for highly leveraged, fixed-asset-intensive firms (shipping, construction).
  - Monetary accommodation example:
    - Bank of Korea policy rate cut from 3.25 percent in 2012 to 2 percent in 2014.
    - Counterfactual increase in corporate borrowing cost from 4.5 to 5.75 percent is a 28 percent difference (assuming immediate pass-through).
  - P-ICR<1 (ICR<1 for three consecutive years) findings:
    - Only about half of manufacturing firms with ICR<1 satisfy P-ICR<1.
    - A majority of ICR<1 firms in shipping also have P-ICR<1.
    - Share of ICR<1 firms in construction and other non-manufacturing industries that are also P-ICR<1 is about 70 percent.

### III. Debt-at-risk and employment-at-risk: definitions, baseline, and stress scenarios
- Definitions:
  - Debt-at-risk: total debt in P-ICR<1 firms.
  - Employment-at-risk: total employment in P-ICR<1 firms.
- Scenario design:
  - Estimates based on 2014 actual P-ICR<1 firms (ICR<1 in 2012-2014).
  - Counterfactuals with 10, 20, and 30 percent decreases in earnings from 2012-2014 levels:
    - 2014 actual and 10 percent decrease interpreted as baseline / confidence range.
    - 20 percent and 30 percent decreases interpreted as stress-tests (moderate and severe).
  - Historical context: 20 and 30 percent declines in corporate earnings have been rare in Korea; negative 20 percent in 2014 was an episode.
- Baseline and stress-test estimates:
  - Baseline: debt-at-risk in the Korean economy is 12-14 percent GDP.
  - Baseline: employment-at-risk is 1.9-2.1 percent of the labor force.
  - Stress scenarios (20 or 30 percent declines in corporate earnings):
    - Debt-at-risk increases to 16-18 percent GDP.
    - Employment-at-risk increases to 2.4-3 percent of the labor force.
  - Debt-at-risk concentrated in manufacturing (including shipbuilding), shipping, and construction.
  - Employment-at-risk significant in services due to higher labor intensity.
- Conversion reference:
  - Based on 2014 nominal GDP (KRW 1485 trillion) and end-2014 labor force (26.4 million persons).

### IV. Benefits of corporate debt restructuring: econometric evidence
- Empirical approach:
  - Firm-level regressions of growth rate of fixed assets and growth rate of employees on P-ICR<1 and ICR<1 dummies, controlling for ROA, past sales growth, and firm size (log total assets).
  - Industry and year fixed effects; sample 2012-2014; 5 percent trim on growth rates.
- Regression findings and quantitative impacts:
  - Persistent debt vulnerability (P-ICR<1) and one-off ICR shocks (ICR<1) negatively affect investment and hiring.
  - Effects for P-ICR<1 are cumulative with ICR<1; curing persistent vulnerability removes both impediments.
  - In manufacturing, investment responds predominantly to P-ICR<1 rather than to ICR<1.
  - Quantitative impacts used:
    - Moving a firm from P-ICR<1 to ICR>1 increases investment by 3.1 percentage points of fixed assets annually.
    - Moving a firm from P-ICR<1 to ICR>1 increases hiring by 2.3 percentage points of the number of employees annually.
- Robustness notes:
  - Coefficients on ICR dummies sensitive to dataset trimming; 5 percent trim chosen.
  - Alternative specifications reported; P-ICR proved a stronger predictor than ICR level.

### V. Costs of corporate debt restructuring: creditor losses and employment effects
- External LGD evidence summarized:
  - Schuermann (2004): Average LGD across all obligors is 60 percent; during recessions or for distressed industries LGD increases to 70 percent.
  - Schuermann (2004): Average LGD for bank loans is 50 percent in normal times, and 60 percent in recessions.
  - Shibut and Singer (2015): Average LGD for commercial and industrial bank loans is 60 percent during the recession (2009-10), and 45 percent outside the recession (2011 onwards).
  - Shibut and Singer (2015): LGD is up to 7.5 percentage points lower in non-judicial (out-of-court) resolutions.
- Assumptions adopted for Korea (most optimistic within plausible range, bank-led and out-of-court):
  - LGD of 40 percent in most industries.
  - LGD of 50 percent in the distressed industries.
  - Distressed industries (benchmark): shipping and shipbuilding only.
  - Distressed industries (stress-tests): shipping and shipbuilding, petrochemicals, and steel.
- Employment-impact evidence and assumptions:
  - Kang and Shivdasani (1997): layoffs on average 20 percent of the workforce, but higher in distressed sectors (up to 30 percent in some cases).
  - Assumed employment loss: 20 percent in most industries; 25 percent in distressed industries.
  - “The impact of corporate restructuring on employment is, conveniently, half of that on the creditors as captured by the LGD.”
- Acknowledged omitted costs: declines in output and consumption, potential credit supply constraints if bank capital is weakened, and possible larger employment losses if corporate restructuring leads to offshoring.
- Risk note: assumptions are optimistic and based on U.S. data; LGD may be materially higher in weaker institutional environments.

### VI. Estimated costs and employment impacts (baseline and stress-tests)
- Applying LGD and employment-impact assumptions to debt-at-risk and employment-at-risk yields:
  - Costs to creditors: 5.3-5.8 percent of GDP in the baseline, increasing to 7-7.6 percent of GDP in stress-tests.
  - Estimated employment impact: 0.4-0.45 percent of the labor force in the baseline, increasing to 0.5-0.6 percent of the labor force in stress-tests.
- Sectoral concentration:
  - Creditor costs predominantly from manufacturing, shipping, and construction.
  - A large part of the employment impact comes from services.
- Distributional caveat:
  - Part of costs may accrue to banks and could be offset against capital and provisions; part may need fiscal support to alleviate bank pressures and incentivize speedy restructuring.

### VII. Method to combine benefits and costs and main quantitative findings
- Method:
  - Costs: one-time costs taken from applied LGD to debt-at-risk and employment-at-risk.
  - Benefits:
    - Reduce 2014 fixed assets and employment in P-ICR<1 firms by 20 or 25 percent (one-off reduction).
    - Apply estimated increased investment and hiring in restructured firms (from regressions) to post-restructuring volumes to obtain repeated, annual benefits.
- Main findings:
  - Baseline one-off costs to creditors of 5.3-5.8 percent of GDP are offset by a future annual increase in corporate investment of about 0.45 percent of GDP.
  - Baseline one-off employment impact of 0.4-0.45 percent of labor force is offset by 0.03-0.04 percent of labor force higher hiring in subsequent years.
  - Both costs and benefits are higher in stress-test scenarios.
  - For any scenario, accumulated benefits offset the one-time costs in about 10-12 years.

### VIII. Adjustments for sample coverage and layoffs without debt restructuring
- Two omissions addressed:
  - Sample coverage: 40 percent of firms by volume of corporate debt are not in Orbis (omitted firms disproportionately smaller).
  - Layoffs without debt restructuring: some ICR<1 but P-ICR>1 firms may undertake corporate actions involving layoffs.
- Adjustment assumptions:
  - Extrapolate results to the 40 percent of firms not in Orbis by inflating results by 100/60 = 66 percent.
  - For omitted (smaller) firms, apply lower LGD: 20 percent overall and 25 percent for distressed industries (instead of 40 and 50 percent), keep original employment-impact estimates.
  - For layoffs without debt restructuring: assume half of firms with ICR<1 but P-ICR>1 undertake corporate actions; assume employment losses half those of debt restructuring: 10 percent for all industries and 12.5 percent for distressed industries.
  - Apply smaller benefit estimate for these firms (coefficient on the ICR<1 dummy) and inflate results by 66 percent for sample coverage.
- Adjusted results:
  - Baseline cost to creditors: 7-7.75 percent of GDP.
  - Future increase in corporate investment: about 0.85-0.9 percent of GDP (annual).
  - Baseline employment impact: 0.9-0.95 percent of labor force (one-off) offset by 0.1 percent of labor force higher hiring in subsequent years.
  - Both costs and benefits higher in stress-tests.
  - For any scenario, accumulated benefits offset the one-time costs in about 10 years.

### IX. Illustrative comparisons and bank-fiscal implications
- Shipping and shipbuilding illustrative comparison:
  - Estimated cost (adjusted baseline) ≈ KRW 31 trillion.
  - Comparable policy supports: KRW 12 trillion support package for state-owned KDB and KEXIM banks and about KRW 10 trillion internal loss-absorption capacity available in those banks.
  - Employment-impact estimate for shipbuilding only: 10 thousand persons (method gives 10 thousand, compared to 20 thousand suggested by some analyses).
- Comparison with Bank of Korea (2015) FSR:
  - Sample sizes: this study 106,870 firms vs FSR 27,995 firms.
  - Share of distressed firms: FSR 10.5 percent of firms by number vs this study 9.2 percent by number and 22 percent by debt volume.
  - FSR reports about 30 percent of distressed firms’ debt is from commercial banks (another 40 percent from policy banks, 30 percent from other creditors).
  - Implied: from 5.5-7.5 percent of GDP baseline cost estimate, about 1.6-2.25 percent of GDP may accrue to commercial banks.
  - This is higher than current total provisions in commercial banks of 1.2 percent of GDP (staff calculations based on Bank of Korea (2015)).
  - Losses exceeding provisions would reduce bank risk-weighted capital ratios from current average of 14.5 percent. Bank capital ratios likely remain above the 11.5 percent minimum (8 percent Basel minimum requirement, 1 percent D-SIB surcharge, 2.5 percent capital conservation buffer), but monitoring of bank financial conditions is important to maintain credit supply.
- Bank-specific data (as of end-2015):
  - KDB provisions KRW 3 trillion and KEXIM provisions KRW 0.5 trillion.
  - KDB could use KRW 7 trillion of “excess” Tier 1 capital (which would reduce its Tier 1 capital ratio from 12 to 9 percent).

### X. Limitations, caveats, and conclusions
- Limitations and caveats:
  - Evidence-driven estimates based on 60 percent of firms by value of corporate debt covered in Orbis.
  - Some P-ICR>1 firms still have ICR<1 and may need corporate actions (possibly layoffs) even without debt restructuring; accounting for these would raise both cost and benefit estimates.
  - Use of U.S.-derived assumptions implies strong institutional environment; LGD and costs could be higher elsewhere.
  - Analysis focuses on creditor losses; shareholder losses are not quantified.
  - Omitted effects include declines in output and consumption, potential credit constraints, and offshoring-related employment losses.
- Policy implications and conclusion:
  - Method summary: identify firms likely to require restructuring via P-ICR<1 criterion; estimate effects of debt overhang on investment and hiring; draw on LGD and employment-impact literature; apply judgmental adjustments for sample coverage and layoffs without debt restructuring.
  - Baseline conclusion: corporate debt restructuring entails material one-off economic costs that are recouped by stronger economic activity over about 10 following years.
  - Fiscal and banking implications:
    - Economic cost of restructuring is not directly a fiscal cost but may imply need for fiscal support to incentivize restructuring; compensate affected individuals; maintain bank solvency; and offset short-term output and consumption drag.
    - Higher fiscal revenue from accelerated medium-term output growth and hiring may help offset short-term fiscal costs.
    - Corporate restructurings are more successful when banks are well-capitalized.
  - Qualitative insight reiterated: corporate debt restructurings “pay off.” Quantitative results depend on assumptions but robustly indicate that one-time restructuring costs to creditors and employment impacts are offset in the medium term by higher investment, output growth, and hiring.

*Source: Annex 1. Data Source and Summary Statistics (content unit _wp16204).*

### Annex 1. Data Source and Summary Statistics ...................................................................... 22

### Annex 1. Data Source and Summary Statistics

### I. Introduction: purpose and methodological innovations
- Purpose: offer a template for quantifying benefits and costs of corporate debt restructuring based on a transparent set of assumptions.
- Context: corporate debt overhang is a feature of multiple crises and growth slowdowns, most recently in Europe and some emerging markets.
- Three methodological innovations highlighted:
  - Refining the ICR measure:
    - Introduce a “persistent ICR” P-ICR<1 measure capturing firms with ICR<1 over three consecutive years.
    - The P-ICR<1 criterion covers only about a half of the firms with ICR<1.
  - Evidence-based assumptions on creditor losses and employment impact:
    - Refine the standard “rule of thumb” 60 percent loss-given-default assumption to account for ownership of debt (bank-based restructurings more efficient), type of restructuring (out-of-court more efficient), and business/industry cycle stage.
    - Historic corporate restructurings in advanced economies suggest average employment losses of 20-25 percent.
  - Econometric estimation of debt overhang effects:
    - Show firms with debt overhang invest less and hire less.
    - Removing debt overhang can boost the investment ratio (as a share of fixed assets) and the hiring ratio (as a share of pre-existing employment) by 2.5-3 percentage points each year.
- Key qualitative insight: corporate debt restructurings “pay off” in the medium-term — economic costs to creditors and employment impacts are fully offset by more rapid output growth and higher hiring under the paper’s assumptions.
- Data source and scope:
  - Use the Orbis dataset; Orbis covers 60 percent of Korean firms by volume of corporate debt in 2012-2014.
  - When both consolidated and unconsolidated statements exist for a firm, only unconsolidated statements are used to avoid double counting.
  - Financials and utilities are excluded (utilities’ “cost-plus” pricing makes ICR uninformative).

### II. Corporate debt in Korea: levels, composition, and vulnerabilities
- Corporate debt level and currency composition:
  - Corporate debt in Korea is about 100 percent of GDP.
  - As of end-2015, only 5 percent of corporate debt was foreign currency-denominated.
  - Comparable economies’ foreign currency-denominated corporate debt median: 10 percent; large emerging market economies: 20-70 percent.
- Export intensity and revenue exposure:
  - Korea’s net exports to GDP was over 6 percent in 2015.
  - Share of export revenue in total revenue of Korean firms: 55 percent in manufacturing overall; 65 percent in export-oriented manufacturing sectors (automobiles, electronics, petrochemicals, shipbuilding and steel).
  - Growth rate of Korean corporate earnings, particularly in export-oriented manufacturing and shipping, has been declining and recently turned negative.
- Interest coverage ratio (ICR) considerations:
  - No single best ICR benchmark; examples:
    - ICR<1.5 might be appropriate for manufacturing that needs free cash flow.
    - ICR<1 might be appropriate for highly leveraged, fixed-asset-intensive firms (shipping, construction).
  - Monetary accommodation example:
    - Bank of Korea policy rate cut from 3.25 percent in 2012 to 2 percent in 2014.
    - Counterfactual increase in corporate borrowing cost from 4.5 to 5.75 percent is a 28 percent difference (assuming immediate pass-through).
  - P-ICR<1 (ICR<1 for three consecutive years) is used to identify firms likely unable to grow out of distress:
    - Only about half of manufacturing firms with ICR<1 satisfy P-ICR<1.
    - A majority of ICR<1 firms in shipping also have P-ICR<1.
    - Share of ICR<1 firms in construction and other non-manufacturing industries that are also P-ICR<1 is about 70 percent.

### III. Debt-at-risk and employment-at-risk: definitions, baseline, and stress scenarios
- Definitions:
  - Debt-at-risk: total debt in P-ICR<1 firms.
  - Employment-at-risk: total employment in P-ICR<1 firms.
  - These are volumes likely to be affected by corporate restructuring; not the cost or employment impact themselves.
- Scenario design:
  - Estimates based on 2014 actual P-ICR<1 firms (ICR<1 in 2012-2014).
  - Counterfactual scenarios where ICRs are calculated after 10, 20, and 30 percent decreases in earnings from 2012-2014 levels:
    - 2014 actual and 10 percent decrease interpreted as baseline / confidence range.
    - 20 percent and 30 percent decreases interpreted as stress-tests (moderate and severe).
  - Historical context: 20 and 30 percent declines in corporate earnings have been rare in Korea; negative 20 percent in 2014 was an episode.
- Key baseline and stress-test estimates:
  - Baseline: debt-at-risk in the Korean economy is 12-14 percent GDP.
  - Baseline: employment-at-risk is 1.9-2.1 percent of the labor force.
  - Stress scenarios (20 or 30 percent declines in corporate earnings):
    - Debt-at-risk increases to 16-18 percent GDP.
    - Employment-at-risk increases to 2.4-3 percent of the labor force.
  - Debt-at-risk is concentrated in manufacturing (including shipbuilding), shipping, and construction.
  - Employment-at-risk is significantly in services due to higher labor intensity.

- Conversion reference used:
  - Based on 2014 nominal GDP (KRW 1485 trillion) and end-2014 labor force (26.4 million persons).

### IV. Benefits of corporate debt restructuring: econometric evidence
- Empirical approach:
  - Firm-level regressions linking corporate investment (growth rate of fixed assets) and hiring (growth rate of employees) to P-ICR<1 and ICR<1 dummies, controlling for return on assets (ROA), past sales growth, and firm size (log total assets).
  - Regressions use industry and year fixed effects, cover 2012-2014, and employ a 5 percent trim to reduce outlier influence.
  - Specification similar to Kalemli-Ozcan et al. (2015A).
- Regression findings (high-level summary of coefficients and interpretation):
  - Both persistent debt vulnerability (P-ICR<1) and one-off ICR shocks (ICR<1) have negative effects on investment and hiring.
  - Effects for P-ICR<1 are cumulative with ICR<1; curing persistent vulnerability removes both impediments.
  - In manufacturing, investment responds predominantly to P-ICR<1 rather than to ICR<1.
  - Quantitative impact used in the paper:
    - Moving a firm from P-ICR<1 to ICR>1 increases investment by 3.1 percentage points of fixed assets annually.
    - Moving a firm from P-ICR<1 to ICR>1 increases hiring by 2.3 percentage points of the number of employees annually.
- Robustness and specification notes:
  - Coefficients on ICR dummies sensitive to dataset trimming; 5 percent trim chosen.
  - Alternative specifications (ICR level, leverage variables, lagged explanatory variables) reported as robustness checks; P-ICR proved a stronger predictor than ICR level.
  - In some specifications the ICR level coefficient was insignificant, validating use of ICR and P-ICR dummies.

### V. Cost considerations referenced
- Loss-given-default benchmark and refinements:
  - The paper refines the standard 60 percent loss-given-default assumption to reflect debt ownership, restructuring form, and cycle stage.
- Employment impact of restructurings:
  - Historic corporate restructurings in advanced economies suggest average employment losses of 20-25 percent (used to assess social implications and fiscal costs).

*Source: Annex 1. Data Source and Summary Statistics (content unit _wp16204).*

### Section III-C to quantify the benefits of corporate debt restructuring in Korea.

### Section III-C to quantify the benefits of corporate debt restructuring in Korea

### Additional benefits of restructuring
- Restructured firms may have higher R&D expenditure and positive fiscal implications because restructured firms are expected to make positive profits (as opposed to loss-making pre-restructured firms).  
- Some additional benefits can be estimated using Orbis data, but the analysis focuses on investment and hiring because these are most comparable with the costs described.

### Costs of corporate debt restructuring (creditor losses and employment effects)
- Key external LGD evidence:
  - Schuermann (2004): Average LGD across all obligors is 60 percent; during recessions or for distressed industries LGD increases to 70 percent.
  - Schuermann (2004): Average LGD for bank loans is 50 percent in normal times, and 60 percent in recessions.
  - Shibut and Singer (2015): Average LGD for commercial and industrial bank loans is 60 percent during the recession (2009-10), and 45 percent outside the recession (2011 onwards).
  - Shibut and Singer (2015): LGD is up to 7.5 percentage points lower in non-judicial (out-of-court) resolutions.
- Assumptions adopted for Korea (most optimistic within plausible range, bank-led and out-of-court):
  - LGD of 40 percent in most industries.
  - LGD of 50 percent in the distressed industries.
  - Distressed industries (benchmark): shipping and shipbuilding only.
  - Distressed industries (stress-tests): shipping and shipbuilding, petrochemicals, and steel.
- Employment-impact evidence and assumptions:
  - Kang and Shivdasani (1997): When layoffs take place they are on average 20 percent of the workforce, but higher in distressed sectors (up to 30 percent in some cases).
  - Assumed employment loss: 20 percent in most industries; 25 percent in distressed industries.
  - “The impact of corporate restructuring on employment is, conveniently, half of that on the creditors as captured by the LGD.”
- Acknowledged omitted costs: declines in output and consumption, potential credit supply constraints if bank capital is weakened, and possible larger employment losses if corporate restructuring leads to offshoring.
- Risks: estimates tilted toward higher costs because assumptions are optimistic and based on U.S. data; LGD may be materially higher in weaker institutional environments (example: 60 percents LGD used for China in IMF (2016A)). Shareholders would also incur likely steeper costs.

### Estimated costs and employment impacts (baseline and stress-tests)
- Applying LGD and employment-impact assumptions to debt-at-risk and employment-at-risk from Figure 5 yields:
  - Costs to creditors: 5.3-5.8 percent of GDP in the baseline, increasing to 7-7.6 percent of GDP in stress-tests.
  - Estimated employment impact: 0.4-0.45 percent of the labor force in the baseline, increasing to 0.5-0.6 percent of the labor force in stress-tests.
- Sectoral concentration:
  - Costs to creditors stem predominantly from manufacturing, shipping, and construction.
  - A large part of the employment impact comes from services.
- Distributional caveat: part of costs may accrue to banks and could be offset against capital and provisions; part may need fiscal support to alleviate bank pressures and incentivize speedy restructuring.

### Method to combine benefits and costs
- Costs: one-time costs taken from Figure 6 (applied LGD and employment assumptions to debt-at-risk and employment-at-risk).
- Benefits: computed in two steps:
  - Reduce 2014 fixed assets and employment in P-ICR<1 firms by 20 or 25 percent (consistent with employment-impact assumptions) to obtain post-restructuring fixed assets and employment volumes (one-off reduction).
  - Apply estimates of increased investment and hiring in restructured firms from Table 1 (columns 1 and 3) to the post-restructuring volumes to obtain repeated, annual benefits.
- Main quantitative findings (baseline and stress-tests, summarized from Table 2 narrative):
  - Baseline one-off costs to creditors of 5.3-5.8 percent of GDP are offset by a future annual increase in corporate investment of about 0.45 percent of GDP.
  - Baseline one-off employment impact of 0.4-0.45 percent of labor force is offset by 0.03-0.04 percent of labor force higher hiring in subsequent years.
  - Both costs and benefits are higher in stress-test scenarios.
  - For any scenario, accumulated benefits offset the one-time costs in about 10-12 years.

### Adjustments for sample coverage and layoffs without debt restructuring
- Two omissions addressed:
  - Sample coverage: 40 percent of firms by volume of corporate debt are not in Orbis (omitted firms disproportionately smaller).
  - Layoffs without debt restructuring: some ICR<1 but P-ICR>1 firms may undertake corporate actions (efficiency improvements or consolidation) that involve layoffs.
- Adjustment procedure and assumptions:
  - Extrapolate results to the 40 percent of firms not in Orbis by inflating results by 100/60 = 66 percent.
  - For omitted (smaller) firms, apply a lower LGD: 20 percent overall and 25 percent for distressed industries (instead of 40 and 50 percent), while keeping the original employment-impact estimates.
  - For layoffs without debt restructuring: assume half of firms with ICR<1 but P-ICR>1 undertake corporate actions; assume employment losses half those of debt restructuring: 10 percent for all industries and 12.5 percent for distressed industries.
  - Apply a smaller estimate for benefits to these firms (coefficient on the ICR<1 dummy in Table 1 only) and inflate results by 66 percent for sample coverage.
- Adjusted results (from Table 3 narrative):
  - Baseline cost to creditors: 7-7.75 percent of GDP.
  - Future increase in corporate investment: about 0.85-0.9 percent of GDP (annual).
  - Baseline employment impact: 0.9-0.95 percent of labor force (one-off) offset by 0.1 percent of labor force higher hiring in subsequent years.
  - Both costs and benefits higher in stress-tests.
  - For any scenario, accumulated benefits offset the one-time costs in about 10 years.

### Illustrative comparisons and bank-fiscal implications
- Shipping and shipbuilding illustrative comparison:
  - Estimated cost (adjusted baseline) ≈ KRW 31 trillion.
  - Comparable policy supports: KRW 12 trillion support package for state-owned KDB and KEXIM banks and about KRW 10 trillion internal loss-absorption capacity available in those banks.
  - Employment-impact estimate for shipbuilding only: 10 thousand persons (method gives 10 thousand, compared to 20 thousand suggested by some analyses).
- Comparison with Bank of Korea (2015) FSR:
  - Sample sizes: this study 106,870 firms vs FSR 27,995 firms.
  - Share of distressed firms: FSR 10.5 percent of firms by number vs this study 9.2 percent by number and 22 percent by debt volume.
  - FSR reports about 30 percent of distressed firms’ debt is from commercial banks (another 40 percent from policy banks, 30 percent from other creditors).
  - Implied: from 5.5-7.5 percent of GDP baseline cost estimate, about 1.6-2.25 percent of GDP may accrue to commercial banks.
  - This is higher than current total provisions in commercial banks of 1.2 percent of GDP (staff calculations based on Bank of Korea (2015)).
  - Losses exceeding provisions would reduce bank risk-weighted capital ratios from current average of 14.5 percent. Bank capital ratios likely remain above the 11.5 percent minimum (8 percent Basel minimum requirement, 1 percent D-SIB surcharge, 2.5 percent capital conservation buffer), but monitoring of bank financial conditions is important to maintain credit supply.
- Bank-specific data (as of end-2015):
  - KDB provisions KRW 3 trillion and KEXIM provisions KRW 0.5 trillion.
  - KDB could use KRW 7 trillion of “excess” Tier 1 capital (which would reduce its Tier 1 capital ratio from 12 to 9 percent).

### Limitations and caveats
- Evidence-driven estimates based on 60 percent of firms by value of corporate debt covered in Orbis.
- Some P-ICR>1 firms still have ICR<1 and may need corporate actions (possibly layoffs) even without debt restructuring; accounting for these would raise both cost and benefit estimates.
- Use of U.S.-derived assumptions implies strong institutional environment; LGD and costs could be higher elsewhere.
- Analysis focuses on creditor losses; shareholder losses are not quantified.
- Omitted effects include declines in output and consumption, potential credit constraints, and offshoring-related employment losses.

### Conclusions and policy implications
- Method summary: identify firms likely to require restructuring via P-ICR<1 criterion; estimate effects of debt overhang on investment and hiring; draw on LGD and employment-impact literature; apply judgmental adjustments for sample coverage and layoffs without debt restructuring.
- Baseline conclusion: corporate debt restructuring entails material one-off economic costs that are recouped by stronger economic activity over about 10 following years.
- Fiscal and banking policy implications:
  - Economic cost of restructuring is not directly a fiscal cost but may imply the need for fiscal support to incentivize restructuring; to compensate affected individuals (unemployment benefits, retraining); to maintain bank solvency; and to offset short-term output and consumption drag.
  - Higher fiscal revenue from accelerated medium-term output growth and hiring may help offset short-term fiscal costs.
  - Corporate restructurings are more successful when banks are well-capitalized.
- Qualitative insight: corporate debt restructurings “pay off.” Quantitative results depend on assumptions but robustly indicate that one-time restructuring costs to creditors and employment impacts are offset in the medium term by higher investment, output growth, and hiring.

*Source: _wp16204 - Section III-C to quantify the benefits of corporate debt restructuring in Korea.*

### REFERENCES

### _wp16204 - REFERENCES

### References
- Aiyar, Shekhar, Wolfgang Bergthaler, Jose M. Garrido, Anna Ilyina, Andreas Jobst, Kenneth Kang, Dmitriy Kovtun, Yan Liu, Dermot Monaghan, and Marina Moretti, 2015, A Strategy for Resolving Europe’s Problem Loans. IMF Staff Discussion Note 15/19.
- Bank of Korea, 2015, Status of Chronically Marginal Firms, and Assessment. Financial Stability Report, December.
- Bergthaler, Wolfgang, Kenneth Kang, Yan Liu, and Dermot Monaghan, 2015, Tackling Small and Medium Sized Enterprise Problem Loans in Europe. IMF Staff Discussion Note 15/04.
- Chow, Julian, and Fabian Valencia, 2015, Corporate Vulnerabilities and Impact on the Real Economy. In: Mexico – Selected Issues. IMF Country Report 15/314.
- Elekdag, Selim, Adrian Alter, Nicolas Arregui, Hibiki Ichiue, Oksana Khadarina, Ayumu Ken Kikkawa, Suchitra Kumarapathy, Machiko Narita, and Jinfan Zhang, 2015, Corporate Leverage in Emerging Markets—A Concern? IMF Global Financial Stability Report, October, Chapter 3.
- Fukuda, Shin-ichi and Jun-ichi Nakamura, 2011, Why Did 'Zombie' Firms Recover in Japan? The World Economy 34(7), pp. 1124–37.
- Gianetti, Mariassunta and Andrei Simonov, 2013, On the Real Effects of Bank Bailouts: Micro Evidence from Japan. American Economic Journal: Macroeconomics 5(1), pp. 135–67.
- Grunert, Jens and Martin Weber, 2009, Recovery Rates of Commercial Lending: Empirical Evidence for German companies. Journal of Banking and Finance 33, pp. 505–513.
- IMF, 2016A, Potent Policies for a Successful Normalization. IMF Global Financial Stability Report, April, Chapter 1.
- IMF, 2016B, Press Release: IMF Staff Concludes 2016 Article IV Mission to Korea; http://www.imf.org/external/np/sec/pr/2016/pr16265.htm.
- Kalemli-Ozcan, Sebnem, Luc Laeven, and David Moreno, 2015A, "Debt Overhang in Europe: Evidence from Firm-Bank-Sovereign Linkages."
- Kalemli-Ozcan, Sebnem, Bent Sorensen, Carolina Villegas-Sanchez, Vadym Volosovych, and Sevcan Yesiltas, 2015B, How to Construct Nationally Representative Firm Level Data from the ORBIS Global Database. NBER Working Paper 21558.
- Kang, Jun-Koo and Anil Shivdasani, 1997, Corporate Restructuring during Performance Declines in Japan. Journal of Financial Economics 46 (1), pp. 29–65.
- Mano, Rui, 2015, Are Philippines Nonfinancial Firms Vulnerable to Tighter Financial Conditions?—Firm-Level Analysis and Stress Testing. In: Philippines – Selected Issues. IMF Country Report 15/247.
- Schuermann, Til, 2004, What Do We Know about Loss Given Default? Wharton Financial Institutions Working Paper 04-01.
- Shibut, Lynn and Ryan Singer, 2015, Loss Given Default for Commercial Loans at Failed Banks. FDIC CFR Working Paper 2015–06.

### Annex 1. DATA SOURCE AND SUMMARY STATISTICS

#### A. Firm-Level Data Source
- Annual data for firm-level analysis were downloaded from Orbis.
- Orbis obtains balance sheet and income statement data from NICE Information Service Co., a credit ratings agency and a supplier of financial data on Korean firms.
- Due to a roughly two-year reporting lag in Orbis, the analysis restricts attention to 2006-2014.
- Starting 2010, Orbis covers approximately 60 percent of non-financial corporate debt in Korea by debt volume.
- Data cleaning and sample construction steps:
  - Downloaded from Orbis financial accounts of all non-financial, non-government companies.
  - Kept only the unconsolidated accounts (U1 + U2) to avoid double counting, following Kalemli-Ozcan et al. (2015B).
  - Dropped utilities.
  - Dropped duplicate observations.
  - Dropped observations for firms with negative total assets.
  - Dropped firms that had less than four data-years available.
  - Assigned industry dummies to each account according to either KSIC-9 or US SIC industry classifications, whichever was available in Orbis.

#### C. ICR Regressions: Summary Statistics
- The growth rates of fixed assets and number of employees were trimmed at 5 and 95 percentiles. (Trimming at lower percentiles is insufficient as it retains many outlier observations).
- Key sample counts and aggregates by year:
  - Number of companies in the sample:
    - 2006: 122,205
    - 2007: 115,758
    - 2008: 120,124
    - 2009: 118,141
    - 2010: 133,280
    - 2011: 144,523
    - 2012: 142,213
    - 2013: 131,810
    - 2014: 106,870
  - Total debt of companies in the sample (percent of GDP), Orbis data:
    - 2006: 44
    - 2007: 45
    - 2008: 55
    - 2009: 56
    - 2010: 59
    - 2011: 63
    - 2012: 65
    - 2013: 63
    - 2014: 60
  - Total credit to non-financial sector (percent of GDP), BIS data:
    - 2006: 86
    - 2007: 91
    - 2008: 104
    - 2009: 106
    - 2010: 102
    - 2011: 104
    - 2012: 106
    - 2013: 102
    - 2014: 105
- Observations and summary statistics for trimmed variables:
  - Fixed asset growth rate:
    - Obs: 321,837
    - Mean: 0.14
    - Std Dev: 0.39
    - Min: -0.35
    - Max: 2.30
    - Min* for 1 percent trim: -0.71
    - Max* for 1 percent trim: 12.54
  - Number of employees growth rate:
    - Obs: 164,905
    - Mean: 0.03
    - Std Dev: 0.17
    - Min: -0.38
    - Max: 0.70
    - Min* for 1 percent trim: -0.69
    - Max* for 1 percent trim: 2.33

*Source: _wp16204 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16204.pdf_