## 1grcea2021002

## Source details

**Canonical URL:** [1grcea2021002](https://www.imf.org/-/media/files/publications/cr/2021/english/1grcea2021002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2021/english/1grcea2021002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2021/english/1grcea2021002.pdf.json)

---

### Introduction and key findings
- Greek non-financial corporations (NFCs) improved balance sheets over the last decade despite a challenging macroeconomic environment.
- Simulations show that the government’s fiscal support during the COVID-19 shock prevented a deterioration in corporate balance sheets that otherwise would have rivalled the experience during the early phase of the sovereign debt crisis.
- Given narrowing fiscal space, a data-driven approach is recommended to deploy further policy support that targets viable sectors.
- The implementation of the new solvency framework will be key to reducing the incidence of zombie firms.

### Dataset and coverage
- Firm-level dataset: Orbis Amadeus covering 2009–18, about 131,000 firm-by-year observations; cleaned aggregated dataset covers about 46 percent of total operating revenues in Greece (OECD SDBS).
- Sample composition and counts:
  - Total firms: 100541
  - SMEs: 98156
  - Large firms: 2385
  - Total operating revenue (billions of euro): 1136
  - SMEs operating revenue (billions of euro): 636
  - Large operating revenue (billions of euro): 500
- Size representation notes:
  - Underrepresents micro firms (≤ 9 employees).
  - Overrepresents medium-sized firms (50–249 employees) and large firms (≥ 250 employees).
  - Share of small firms (10–49 employees) consistent with SDBS.

### Historical performance of Greek NFCs (2009–18)
- Leverage (gross debt/EBITDA):
  - Share of turnover in firms with gross debt/EBITDA ratio above four fell from 41 percent in 2010 to 28 percent in 2016.
  - Leverage increased slightly beginning in 2016 across firm sizes.
- Investment:
  - Business investment (change in log fixed tangible assets) contracted during 2010–12; recovered since 2013.
  - Investment peaked at roughly 8 percent in 2014.
  - Investment growth in SMEs higher relative to large firms for most of the sample period.
- Profitability:
  - Share of turnover in firms with ROA higher than 5 percent increased from 22 percent in 2011 to close to one-third in 2017–18.
  - Large firms experienced higher earnings volatility than SMEs.
- Financial distress and liquidity buffers:
  - ICR measured as EBIT over interest paid.
  - Share of turnover in firms with ICR below one declined from 31 percent in 2011 to 13 percent in 2018.
  - Financial vulnerability higher in large firms than in SMEs.
- Exporters:
  - About 12.4 percent of firms are exporters; exporters account for about 22.3 percent of sample turnover.
  - Exporters had higher leverage, investment, and profitability, and slightly lower liquidity buffers than non-exporters.
  - High-leverage exporting firms’ share in total turnover about 33 percent in 2018 vs. 30 percent for non-exporters.

### Sectoral vulnerabilities prior to COVID-19
- Liquidity vulnerabilities concentrated mostly in:
  - Construction; Electricity and gas; Transportation; Manufacturing; ICT; Accommodation.
- Wholesale and retail:
  - Lower share of vulnerable firms (8.5 percent of turnover in firms with ICR less than one).
  - Accounted for about 40.6 percent of total turnover in the sample.

### Definitions, classification, and empirical framework
- ICR definition and zombie criteria:
  - ICR = EBIT / interest paid.
  - Financial distress indicator: ICR < 1.
  - Zombie firm definition (Adalet McGowan et al. (2018) approach):
    - ICR below one for three consecutive years.
    - Firm age at least 10 years.
- Dynamic panel model (units: firm i, one-digit NACE j, year t):
  - ICR_ijt = α ICR_ijt−1 + FirmChar_ijt−1′ Γ + β ΔGVA_jt + Macro_t′ Ω + u_i + ε_ijt
  - Macro_t includes output gap and real interest rate.
- Key estimated dynamic coefficients (benchmark two-step difference GMM):
  - A one-point increase in lagged ICR associated with a 0.368-point increase in contemporaneous ICR.
  - A one-point increase in twice lagged ICR associated with a 0.049-point increase in contemporaneous ICR.
- Role of macro variables:
  - Sectoral GVA growth, output gap, and real interest rate all affect ICR positively and are statistically significant.
  - Real interest rate interpreted as measure of aggregate demand rather than actual firm borrowing cost.
- Model diagnostics (selected):
  - Observations: 50,567 (GMM benchmark); Instruments: 11; Arellano-Bond AR(2) p-val: 0.726; Hansen test p-val: 0.389.

### COVID-19 projection methodology and fiscal support scenarios
- Two-step projection (2019–2021):
  1. Estimate dynamic link between firm ICRs and aggregate/sectoral shocks conditional on firm covariates.
  2. Project firm ICRs for 2019–2021 using estimated coefficients with projected macro variables.
- Macroeconomic covariates: output gap, real interest rate, sectoral real GVA growth (one-digit NACE).
- Scenarios:
  - Baseline scenario: macro assumptions consistent with 2021 Article IV projections and includes fiscal support measures.
  - Alternative (counterfactual) scenario: regressors projected without implementation of fiscal support measures, derived using fiscal multipliers and estimates of pandemic-fiscal support measures implemented in 2020–21.
- Fiscal support measures (baseline): 11 percent of GDP in 2020; 7.4 percent of GDP in 2021.
- Note: Support mix includes refundable advances, loan guarantees, rent reductions, intermediate purchase support, and other programs.

### Key simulation findings and numeric outcomes
- Sectoral activity drops (2020–21): wholesale and retail, transportation, and accommodation and food services faced large drops in 2020; under baseline contact-intensive sectors expected to recover half of 2020 output loss.
- Share of firms with ICR < 1:
  - Projected at 10 percent in 2021 under the baseline.
  - Pre-pandemic level: 11 percent.
- Debt trapped in financially distressed firms:
  - Projected to decline from 2019 to 2021 by two percentage points to 12 percent of total debt (baseline).
- Alternative (no fiscal support) scenario:
  - Share of firms under distress would have been roughly four times larger relative to the baseline and exceed the 2010–13 average.
  - Debt held by firms under distress would have been about six times higher than the baseline and above the 2010–13 average.
- Sectoral example — Wholesale and trade:
  - Government support helped reduce number of financially vulnerable firms by 33 percentage points.
  - Helped reduce share of debt held by distressed firms by 65 percent of total debt.
- Zombies and trapped debt (baseline vs alternative):
  - Baseline projected share of zombie firms: 5.5 percent.
  - Alternative projected share of zombie firms: 8.5 percent.
  - Baseline projected trapped debt in zombie firms: 4.6 percent.
  - Alternative projected trapped debt in zombie firms: 13 percent.
  - Under the baseline, trapped debt in zombie firms declines by about half relative to pre-pandemic levels and remains substantially below the 2011–13 average.
- Sectoral and size pockets of vulnerability (2021 projections):
  - Trapped debt highest in accommodation, manufacturing, and construction.
  - Wholesale and retail expected to hold 5.5 percent of trapped debt vs. 25 percent of total debt held by the sector.
  - Large firms projected to account for more than 50 percent of trapped debt in 2021 while holding about 24 percent of total debt.

### Sectoral incidence of increased ICR<1 (selected figures, % of total firms)
- Increase in firms with ICR<1 by sector (2020, 2021):
  - Manufacturing: 36.3, 55.0
  - Electricity & gas: 26.2, 23.7
  - Construction: 21.2, 33.9
  - Wholesale & retail: 33.3, 65.0
  - Transportation: 30.9, 63.4
  - Accommodation: 39.6, 57.7
  - Real estate: 31.4, 35.2
  - Other: 20.3, 42.3
- By firm size:
  - SME (<250): 32.3, 59.7
  - Large (≥250): 29.6, 45.1

### Zombie firms — selected shares and contributions
- Unweighted % zombies within sector (2016–18 avg., 2020 proj., 2021 proj.):
  - Manufacturing: 7.6, 5.1, 7.1
  - Electricity & gas: 2.5, 8.2, 9.8
  - Construction: 6.4, 6.7, 9.4
  - Wholesale & retail: 4.7, 3.3, 3.9
  - Accommodation: 9.6, 8.3, 8.7
  - Real estate: 12.7, 8.6, 8.6
- Weighted by debt — % contribution to total zombies (2016–18 avg., 2020 proj., 2021 proj.):
  - Manufacturing: 32.8, 26.5, 31.9
  - Construction: 6.3, 13.8, 16.5
  - Wholesale & retail: 23.3, 4.9, 5.5
  - Accommodation: 19.9, 43.9, 33.4
  - Real estate: 9.0, 1.1, 1.2
- By firm size (unweighted % zombies within size category, 2016–18 avg., 2020 proj., 2021 proj.):
  - SME (<250): 5.6, 4.3, 5.7
  - Large (≥250): 5.5, 4.1, 3.3
- Weighted by debt — % zombies within size category (2016–18 avg., 2020 proj., 2021 proj.):
  - SME (<250): 17.6, 3.3, 3.4
  - Large (≥250): 10.5, 9.2, 7.2

### Firm- and macro-level summary statistics (selected, Table 1)
- Firm-level variables (Mean, Median, Std, Count):
  - Employees: 30.9, 12.0, 63.7, 116,541
  - Age: 18.5, 16.0, 14.0, 128,739
  - Log operating revenue: 14.2, 14.1, 1.6, 128,509
  - Debt/EBITDA: 2.7, 0.6, 11.7, 108,075
  - Debt/Assets: 0.2, 0.1, 0.2, 130,915
  - ROA: 2.7, 1.5, 9.0, 130,196
  - ICR: 30.4, 2.8, 89.7, 111,780
- Macro variables (Mean, Median, Std, Count):
  - Sectoral GVA growth: -2.3, -2.4, 11.2, 200
  - GDP growth: -2.6, -1.84, 3.9, 10
  - Output gap: -9.9, -11.18, 4.9, 10
  - Real interest rate: -0.5, -0.37, 1.6, 10
  - Unemployment rate: 20.8, 22.5, 5.9, 10

### Policy recommendations and implications
- Continue a data-driven approach to target further policy support toward viable sectors and firms given narrowing fiscal space.
- Prioritize implementation of the new solvency framework (new Insolvency Code) to reduce incidence and persistence of zombie firms; lift temporary measures that freeze debt resolution activity.
- Use firm-level analysis to improve sectoral and firm-level targeting once the pandemic recedes.
- Allow resolution framework to function and complement with stepped-up enforcement so only viable firms receive state-supported rescheduling options.

### Investment, capital stock, and the investment gap (selected findings)
- Staff baseline projection: real investment rates stabilize between 15–17 percent of GDP.
- Government macro-fiscal baseline: investment-to-GDP ratio increases to around 17 percent by 2025.
- RRP expectation: private investment boosted by 20 percent by 2026.
- Historical and recent investment rates:
  - Pre-sovereign debt crisis peak total investment rate: 24 percent of GDP.
  - Total investment growth close to 0 percent (2015–19); investment-to-GDP ratio 10 percent at end-2019.
  - Greece’s end-2019 investment-to-GDP ratio about 10 percentage points lower than the regional average.
- Capital stock and sectoral composition:
  - Net capital stock per capita in Greece about two thirds of the Euro Area’s average.
  - Net capital stock contracted by an average of 1.2 percent annually in the last decade; output declined by 26 percent over the same period.
  - Private sector holds 70 percent of total capital stock.
  - Public-to-private capital ratio close to 45 percent in 2019 (growth-maximizing public-private ratio in literature for advanced economies: 42 percent).
- Estimated investment gap (end-2019): ranges from 1.6 to 8 percent of GDP depending on method:
  - Golden Rule: about 4 percent of GDP (standard estimate).
  - Historical Benchmark: largest estimated gap of 8 percent of GDP.
  - Predicted Norm: smallest gap, 1.6 percent of GDP in richer specifications.
- Structural drivers of the investment gap:
  - Top three contributors to the fall in investment since 2007: services share, capital account openness, and regulatory quality — together account for over 50 percent of the decline.
  - Other factors: low returns on capital (NFC gross return on capital 13.7 percent in 2019, 10 percentage points below EA average), high debt burden, insufficient domestic savings, reduced access to bank credit.
- Policy recommendations to boost private investment:
  - Product market reforms to encourage economies of scale; trade facilitation; reforms to raise private-sector productivity.
  - Facilitate access to credit, good-paying jobs, and availability of infrastructure and services (transport, childcare, elder care, unemployment benefits) to support housing demand.
  - Implement and sequence Recovery and Resilience Plan reforms (digitalization, green investments, up-/re-skilling).
  - Improve public investment efficiency (Strategic Projects Pipeline, Project Preparation Facility).
  - Consider judicial reform, cadaster completion, pension adjustments to encourage private savings and labor participation, and privatizations that attract greenfield projects.

### Appendix I — Investment Gap Estimations (Golden Rule, Historical Benchmark, Predicted Norm)
- Golden Rule framework: neo-classical (Ramsey-Cass-Koopmans) steady-state interpretation; parameterized with α, p (SRTP), δ, n, g; PWT data used (versions 10.0, 9.0).
- SRTP values tested: SRTP=5 percent (baseline), SRTP=6, SRTP=4.0, SRTP=3.5.
- Golden Rule scenario labels and descriptions (verbatim):
  - GR5-0.8Self-employed income accrues 100 percent to labor.
  - GR41.9SRTP=6 (higher consumption).
  - GR33.1Self-employed income accrues 67 percent to labor.
  - GR13.8No adjustment to PWT values, SRTP=5.
  - GR65.5Adjusted for welfare-relevant TFP (based on absorption).
  - GR26.3SRTP=4.0 (lower consumption).
  - GR76.9SRTP=3.5 (lowest consumption).
- Historical Benchmark: fits transition (“catch-up”) investment path for Germany, France, Italy, Spain (1951–2011), fitted R2=0.87; uses Greece’s K/L and TFP to compute sustainable benchmark investment rate.
- Predicted Norm: panel fixed-effects regression for 27 EU countries over past three decades; controls include regulatory efficiency, trade and financial account openness, robustness checks with alternative indicators.

*Prepared by IMF staff (content excerpted from IMF staff analysis in "Corporate Vulnerabilities in the Wake of COVID-19" and related investment-gap appendix and chapter).*

### 1. Sectoral and Size Representation of the Analysis Sample  ________________________________ 6

### 1. Sectoral and Size Representation of the Analysis Sample

### Introduction and key findings
- Greek non-financial corporations (NFCs) improved balance sheets over the last decade despite a challenging macroeconomic environment.
- Simulations show that the government’s fiscal support during the COVID-19 shock prevented a deterioration in corporate balance sheets that otherwise would have rivalled the experience during the early phase of the sovereign debt crisis.
- Given narrowing fiscal space, a data-driven approach is recommended to deploy further policy support that targets viable sectors.
- The implementation of the new solvency framework will be key to reducing the incidence of zombie firms.

### Dataset and coverage
- Firm-level dataset based on Orbis Amadeus covering the period 2009–18, containing about 131,000 firm by year observations.
- After cleaning, the aggregated dataset covers about 46 percent of the total operating revenues in Greece as reported by the OECD Structural and Demographic Business Statistics (SDBS).
- The sample broadly represents sectoral and size distributions relative to SDBS but:
  - Underrepresents micro firms (≤ 9 employees).
  - Overrepresents medium-sized firms (50–249 employees) and large firms (≥ 250 employees).
  - The share of small firms (10–49 employees) is consistent with SDBS.
- Sample count and operating revenue breakdown (as presented):
  - Total firms: 100541
  - SMEs: 98156
  - Large firms: 2385
  - Total operating revenue (billions of euro): 1136
  - SMEs operating revenue (billions of euro): 636
  - Large operating revenue (billions of euro): 500

### Historical performance of Greek NFCs (2009–18)
- Leverage
  - Measured as gross debt/EBITDA.
  - The share of turnover in firms with gross debt/EBITDA ratio above four fell from 41 percent in 2010 to 28 percent in 2016.
  - Leverage increased slightly irrespective of firm size beginning in 2016 in line with economic recovery.
- Investment
  - Business investment measured as the change in log fixed tangible assets contracted during 2010–12 but recovered since 2013 across all size categories.
  - Investment peaked at roughly 8 percent in 2014.
  - Investment growth in SMEs has been higher relative to large firms for most of the sample period.
- Profitability
  - The share of turnover in firms with ROA higher than 5 percent increased from 22 percent in 2011 to close to one-third in 2017–18.
  - Large firms experienced higher volatility in earnings than SMEs.
- Financial distress and liquidity buffers
  - ICR (interest coverage ratio) measured as EBIT over interest paid.
  - The share of turnover in firms with ICR below one declined from 31 percent in 2011 to 13 percent in 2018.
  - Financial vulnerability is higher in large firms than in SMEs, possibly reflecting higher leverage.
- Exporters vs. non-exporters
  - Exporting firms had higher leverage, investment, and profitability, and slightly lower liquidity buffers than non-exporting firms.
  - About 12.4 percent of firms in the sample are exporters.
  - Exporters account for about 22.3 percent of sample turnover.
  - High-leverage exporting firms’ share in total turnover was about 33 percent in 2018, compared to 30 percent for non-exporters.
  - Exporting firms had higher investment rates than non-exporters for most of the sample period.

### Sectoral vulnerabilities prior to COVID-19
- Liquidity vulnerabilities concentrated mostly in:
  - Construction
  - Electricity and gas
  - Transportation
  - Manufacturing
  - Information and communication technology (ICT)
  - Accommodation
- Wholesale and retail sector:
  - Had a lower share of vulnerable firms (8.5 percent of turnover in firms with ICR less than one).
  - Accounted for about 40.6 percent of total turnover in the sample.

### Simulations of the COVID-19 shock and policy impact
- Counterfactual simulations indicate that absent the sizable and targeted policy support measures by the Greek authorities, the adverse impact of the pandemic on firms’ financial health would have rivalled that of the sovereign debt crisis.
- Fiscal support was particularly effective in mitigating financial distress in:
  - Accommodation
  - Manufacturing
  - Wholesale and retail
  - Transportation
- Distributional implications (debt-weighted basis)
  - Electricity and gas, construction sectors, and large firms (≥250 employees) would contribute disproportionately to the incidence of firms with liquidity vulnerabilities relative to their activity shares.
- SME outcomes under no-policy scenarios
  - Gourinchas et al. (2020) estimate SME bankruptcy rate would increase by 57 percent under COVID-19 and in the absence of policy intervention.
  - The paper’s results suggest the zombie share among SMEs would rise by two-thirds in 2021 in a no-policy scenario.

### Policy recommendations and implications
- Continue to use a data-driven approach to target further policy support toward viable sectors and firms given narrowing fiscal space.
- Prioritize implementation of the new solvency framework to reduce the incidence and persistence of zombie firms.
- Use firm-level analysis to improve sectoral and firm-level targeting once the pandemic recedes.

*Prepared by Efthymios Argyropoulos, Yi Liu, and Francisco Parodi; excerpted from the IMF staff analysis in "Corporate Vulnerabilities in the Wake of COVID-19."*

### 11.      Our analysis of financial distress is based on the ICR. More specifically, ICR is defined as

### 11.      Our analysis of financial distress is based on the ICR. More specifically, ICR is defined as

### Definitions and classification
- ICR is defined as the ratio of earnings before interest and taxes (EBIT) to interest paid and indicates the ability of a firm to pay interest expenses on outstanding debt using contemporaneous earnings.
- An ICR less than one is typically used as an indicator of financial distress because earnings are insufficient to cover interest payments and additional financing may be needed to sustain operations.
- Zombie firm definition (following Adalet McGowan et al. (2018)):
  - ICR below one for three consecutive years.
  - Firm aged at least 10 years.
  - Rationale: three-year window captures persistent financial weakness beyond business cycle fluctuations; age restriction distinguishes from young, innovative firms.

### Empirical framework and estimation
- Estimation period and scope:
  - Dynamic panel model estimated over the period 2009–2018.
  - Units: firm i, one-digit NACE sector j, year t.
- Model (equation 1) structure (variables preserved as in source):
  - ICR_ijt = α ICR_ijt−1 + FirmChar_ijt−1′ Γ + β ΔGVA_jt + Macro_t′ Ω + u_i + ε_ijt
  - FirmChar_ijt−1: lagged firm-level variables such as leverage and profitability.
  - ΔGVA_jt: growth rate of sectoral real gross value added at the one-digit NACE level.
  - Macro_t: output gap and real interest rate (captures common aggregate demand shocks).
  - u_i: unobserved firm fixed effects; ε_ijt: disturbances assumed serially uncorrelated.
- Identification and assumptions:
  - ICR lagged terms and firm characteristics treated as predetermined; ΔGVA_jt and Macro_t treated as strictly exogenous.
  - Orthogonality of sectoral and aggregate shocks to the unobserved firm-specific error term is formally tested.
- Estimation approach and diagnostics:
  - Benchmark difference GMM estimation includes lagged ICR, contemporaneous sectoral GVA growth rate, output gap, and real interest rate.
  - Twice lagged ICR added to satisfy the Arellano-Bond AR(2) test in first differences.
  - All instruments jointly valid; IV-style instruments for sectoral GVA growth rate, output gap, and real interest rate satisfy difference-in-Hansen exogeneity test.
  - Estimated coefficients on lagged ICR lie between OLS and fixed effects counterparts, suggesting correct specification.
- Key estimated dynamic coefficients:
  - A one-point increase in lagged ICR is associated with a 0.368-point increase in contemporaneous ICR.
  - A one-point increase in twice lagged ICR is associated with a 0.049-point increase in contemporaneous ICR.
- Role of macro variables:
  - Sectoral GVA growth, output gap, and real interest rate all affect ICR positively and are statistically significant.
  - The real interest rate is interpreted more as a measure of aggregate demand than the actual cost of borrowing for firms.
- Robustness and limitations:
  - Augmented specifications including additional firm-level covariates (profitability, leverage) failed Hansen and difference-in-Hansen tests.
  - Benchmark model robust to alternative lag lengths, weighting matrices, and transformation methods.
  - In-sample fit for realized 2018 ICRs: adequate predictive power for firms with ICR below 10; underpredicts ICR values higher than 10 (less relevant for simulations focused on low ICRs).
  - Suggested extensions: address possible sample selection bias (extend data to begin in 2001), Orbis data shortcomings (underestimates firm exit and micro firms), and omitted variable bias (employment and consumption variables).

### COVID-19 projection methodology and scenarios
- Two-step projection procedure for 2019–2021:
  1. Estimate dynamic panel link between firm ICRs and aggregate/sectoral shocks conditional on firm covariates.
  2. Project firm ICRs for 2019–2021 using estimated coefficients with projected macro variables for 2021 and realized values for 2019 and 2020.
- Macroeconomic covariates used:
  - Output gap and real interest rate for aggregate demand component.
  - Growth rate of sectoral real GVA at one-digit NACE level for sector-specific shocks.
- Scenarios:
  - Baseline scenario: macro assumptions consistent with 2021 Article IV projections and includes fiscal support measures.
  - Alternative (counterfactual) scenario: values of regressors projected without implementation of fiscal support measures; derived using only fiscal multipliers and estimates of pandemic-fiscal support measures implemented in 2020–21.
  - Baseline captures other financial sector and EU-wide measures indirectly and does not change them in the alternative scenario.
- Fiscal support measures (baseline):
  - 11 percent of GDP in 2020.
  - 7.4 percent of GDP in 2021.
- Note on support measures composition:
  - Mix of liquidity loans conditionally becoming grants (“refundable advances”), loan guarantees, rent reductions, intermediate purchase support, and other programs. Modelling individual measures is outside scope.

### Key simulation findings and numeric outcomes
- Sectoral activity drops (2020–21):
  - Wholesale and retail, transportation, and accommodation and food services faced large drops in sectoral activity in 2020.
  - Under the baseline, contact-intensive sectors expected to recover half of the output lost during 2020.
- Aggregate and firm-level distress metrics:
  - Share of firms with ICR < 1:
    - Projected at 10 percent in 2021 under the baseline.
    - Pre-pandemic level: 11 percent.
  - Debt trapped in financially distressed firms:
    - Projected to decline from 2019 to 2021 by two percentage points to 12 percent of total debt (baseline).
  - Alternative scenario (no fiscal support) outcomes:
    - Share of firms under distress would have been roughly four times larger relative to the baseline and exceed the 2010–13 average.
    - Debt held by firms under distress would have been about six times higher than the baseline and above the 2010–13 average.
- Sectoral impact example (interpreted as reduction in vulnerabilities by end-2021):
  - Wholesale and trade sector:
    - Macro impact of government support policies helped reduce the number of financially vulnerable firms by 33 percentage points.
    - Helped reduce the share of debt held by distressed firms by 65 percent of total debt.
- Zombie firms and trapped debt:
  - Projected change in zombie firms (baseline vs alternative):
    - Baseline projected share of zombie firms: 5.5 percent.
    - Alternative projected share of zombie firms: 8.5 percent.
    - Baseline is about one-third lower than the alternative.
  - Trapped debt in zombie firms:
    - Baseline projected share: 4.6 percent.
    - Alternative projected share: 13 percent.
    - Under the baseline, share of trapped debt in zombie firms would decline by about half relative to pre-pandemic levels and remain substantially below the 2011–13 average.
- Sectoral and size pockets of vulnerability:
  - Projected trapped debt for 2021 highest in accommodation, manufacturing, and construction sectors.
  - In manufacturing, the composition effect explains trapped debt share; in accommodation and construction, within-sector effects explain increases.
  - Wholesale and retail expected to hold a relatively small share of trapped debt (5.5 percent) compared to total debt held by the sector (25 percent).
  - Large firms projected to account for more than 50 percent of trapped debt in 2021, while holding about 24 percent of total debt.

### Conclusions and policy implications
- Simulations indicate that Greek government fiscal support measures helped contain the COVID-19 pandemic’s impact on the NFC sector relative to a counterfactual of no policy support, which would have produced distress similar to the Greek sovereign crisis period.
- Improvements in NFC financial performance over the past decade resulted in a large reduction of ‘trapped debt’ held by zombie firms.
- Implication: The impact of the crisis on NFCs is adequately addressed by temporary support measures (as modelled in the baseline), reducing the risk of widespread trapped debt accumulation akin to 2010–2015 levels.

*Source: IMF staff estimates based on Bureau van Dijk Orbis and IMF databases, as presented in the supplied content.*

### 25.      Going forward, a data-driven approach  could help improve the targeting of measures

### 25.      Going forward, a data-driven approach  could help improve the targeting of measures

### Key recommendation and policy implications
- A data-driven approach can help identify firms that face temporary liquidity problems versus those with persisting financial distress (‘zombie’ firms).
- Allowing the resolution framework to function would help address the persistence of ‘zombie’ firms.
- The new Insolvency Code and the lifting of temporary measures that “freeze” debt resolution activity (measures adopted because of pandemic disruptions to court operations) are a step in the right direction.
- Implementation of a data-driven approach can complement stepped-up enforcement procedures to ensure that only viable firms are provided with state-supported rescheduling options.

### Firm- and macro-level summary statistics (Table 1)
- Firm-level variables (Mean, Median, Std, Count):
  - Employees: 30.9, 12.0, 63.7, 116,541
  - Age: 18.5, 16.0, 14.0, 128,739
  - Log operating revenue: 14.2, 14.1, 1.6, 128,509
  - Debt/EBITDA: 2.7, 0.6, 11.7, 108,075
  - Debt/Assets: 0.2, 0.1, 0.2, 130,915
  - Debt/Equity: 0.9, 0.2, 1.9, 130,914
  - Investment growth: 11.7, -4.4, 111.9, 98,364
  - ROA: 2.7, 1.5, 9.0, 130,196
  - ROE: 6.1, 3.7, 29.6, 130,258
  - ICR: 30.4, 2.8, 89.7, 111,780
- Macro variables (Mean, Median, Std, Count):
  - Sectoral GVA growth: -2.3, -2.4, 11.2, 200
  - GDP growth: -2.6, -1.84, 3.9, 10
  - Output gap: -9.9, -11.18, 4.9, 10
  - Real interest rate: -0.5, -0.37, 1.6, 10
  - Unemployment rate: 20.8, 22.5, 5.9, 10

Notes: EBITDA = earnings before interest, taxes, depreciation, and amortization. ROA = return on assets. ROE = return on equity. ICR = interest coverage ratio measured as EBIT over interest paid. Sectoral GVA growth at one-digit NACE level. Real interest rate = euro area short-term deposit rate minus inflation rate in Greece. Nominal firm-level variables converted to 2010 values and winsorized at top and bottom 0.5%.

### Dynamic panel regression results (Table 2) — key estimated coefficients
- L.ICR (once lagged ICR):
  - OLS: 0.542*** (0.015)
  - FE: 0.202*** (0.017)
  - Diff (benchmark two-step difference GMM): 0.368*** (0.034)
  - Diff 2 (adds lagged ROA and leverage): 0.410*** (0.033)
  - System (two-step system GMM): 0.432*** (0.028)
- L2.ICR (twice lagged ICR):
  - OLS: 0.139*** (0.011)
  - FE: -0.019 (0.014)
  - Diff: 0.049*** (0.016)
  - Diff 2: 0.054*** (0.016)
  - System: 0.074*** (0.015)
- GVA: 0.113*** (0.033), 0.098*** (0.032), 0.152*** (0.037), 0.184*** (0.034), 0.154*** (0.038)
- Output gap: 0.326*** (0.123), 0.860*** (0.123), 0.326** (0.158), 0.268** (0.123), 0.232 (0.159)
- Real rate: 1.944*** (0.260), 2.870*** (0.249), 2.067*** (0.303), 0.924*** (0.247), 1.907*** (0.305)
- Additional variables:
  - L.ROA reported in column (4): -0.025 (0.126)
  - L.leverage in column (4): 21.766*** (5.967)

Model diagnostics and sample:
- Observations: 65,407 (OLS, FE); 50,567 / 50,566 / 50,567 (GMM variants)
- R-squared: 0.43 (OLS), 0.045 (FE)
- Instruments reported for GMM specifications: 11, 27, 20
- Arellano-Bond test for AR(2) in differences (p-val): 0.726, 0.892, 0.499 (for GMM columns)
- Hansen test of joint validity (p-val): 0.389, 0.000, 0.089 (various GMM columns)
- Difference-in-Hansen tests reported as shown.

Notes: L. = once lagged; L2. = twice lagged. Column (3) is benchmark two-step difference GMM with Windmeijer-corrected standard errors clustered at firm level, treating lagged ICRs as predetermined and sectoral GVA growth, output gap, and real interest rate as exogenous/IV-style. Column (5) is two-step system GMM.

### Robustness checks (Table 3)
- Benchmark (column 1) L.ICR: 0.368*** (0.034); L2.ICR: 0.049*** (0.016); GVA: 0.152*** (0.037); Output: 0.326** (0.158); Real Interest Rate: 2.066*** (0.303). Observations: 50,567; Instruments: 11; Arellano-Bond AR(2) p-val: 0.726; Hansen test p-val: 0.389; Difference-in-Hansen p-val: 0.192.
- Variants include just-identified, one-step, and forward orthogonal deviations transformations with broadly similar estimates for L.ICR, L2.ICR, GVA, and Real Interest Rate.

### Impact of government measures on Greek NFCs — selected sectors (Table 4)
- Increase in firms with ICR<1 (% of total firms):
  - By sector (2020, 2021):
    - Manufacturing: 36.3, 55.0
    - Electricity & gas: 26.2, 23.7
    - Construction: 21.2, 33.9
    - Wholesale & retail: 33.3, 65.0
    - Transportation: 30.9, 63.4
    - Accommodation: 39.6, 57.7
    - Real estate: 31.4, 35.2
    - Other: 20.3, 42.3
  - By firm size:
    - SME (<250): 32.3, 59.7
    - Large (≥250): 29.6, 45.1

### Share and composition of zombie firms (Tables 5 and 6)
- Definition: A firm is classified as a zombie firm if it is aged 10 years or older and has an ICR less than one for three consecutive years. Underlying firm-level ICRs winsorized at top and bottom 0.5%. Each firm's debt level in 2020 and 2021 is assumed to remain the same as 2018. Tables restricted to firms with non-missing relevant information.
- Table 5 — % zombies within sector (Unweighted, 2016–18 avg., 2020 proj., 2021 proj.) and % contribution to total zombies:
  - Unweighted % zombies within sector (2016–18 avg., 2020 proj., 2021 proj.):
    - Manufacturing: 7.6, 5.1, 7.1
    - Electricity & gas: 2.5, 8.2, 9.8
    - Construction: 6.4, 6.7, 9.4
    - Wholesale & retail: 4.7, 3.3, 3.9
    - Transportation: 5.9, 4.0, 7.6
    - Accommodation: 9.6, 8.3, 8.7
    - Real estate: 12.7, 8.6, 8.6
    - Other: 2.1, 2.7, 4.2
  - Unweighted % sectoral firm in total (2016–18 avg.):
    - Manufacturing: 23.5
    - Electricity & gas: 1.5
    - Construction: 5.3
    - Wholesale & retail: 41.4
    - Transportation: 4.7
    - Accommodation: 7.0
    - Real estate: 1.9
    - Other: 14.7
  - Unweighted % contribution to total zombies (2016–18 avg., 2020 proj., 2021 proj.):
    - Manufacturing: 31.9, 29.6, 31.5
    - Electricity & gas: 0.6, 2.3, 2.2
    - Construction: 6.1, 7.9, 8.6
    - Wholesale & retail: 34.5, 35.6, 32.3
    - Transportation: 4.9, 4.2, 6.1
    - Accommodation: 12.0, 10.2, 8.2
    - Real estate: 4.3, 2.8, 2.2
    - Other: 5.6, 7.4, 9.0
- Table 5 — Weighted by Debt (selected figures):
  - % zombies within sector (2016–18 avg., 2020 proj., 2021 proj.):
    - Manufacturing: 14.5, 3.9, 4.2
    - Construction: 21.4, 20.6, 22.0
    - Wholesale & retail: 14.8, 1.0, 1.0
    - Accommodation: 19.2, 15.4, 10.4
    - Real estate: 26.1, 1.1, 1.1
  - % contribution to total zombies weighted by debt (2016–18 avg., 2020 proj., 2021 proj.):
    - Manufacturing: 32.8, 26.5, 31.9
    - Construction: 6.3, 13.8, 16.5
    - Wholesale & retail: 23.3, 4.9, 5.5
    - Accommodation: 19.9, 43.9, 33.4
    - Real estate: 9.0, 1.1, 1.2
- Table 6 — Share of zombie firms by firm size:
  - Unweighted % zombies within size category (2016–18 avg., 2020 proj., 2021 proj.):
    - SME (<250): 5.6, 4.3, 5.7
    - Large (≥250): 5.5, 4.1, 3.3
  - % sectoral firm in total (2016–18 avg.):
    - SME (<250): 96.9
    - Large (≥250): 3.1
  - % contribution to total zombies (unweighted, 2016–18 avg., 2020 proj., 2021 proj.):
    - SME (<250): 97.0, 95.3, 97.1
    - Large (≥250): 3.0, 4.7, 2.9
  - Weighted by Debt — % zombies within size category (2016–18 avg., 2020 proj., 2021 proj.):
    - SME (<250): 17.6, 3.3, 3.4
    - Large (≥250): 10.5, 9.2, 7.2
  - Weighted by Debt — % contribution to total zombies (2016–18 avg., 2016–18 avg. share, 2020 proj., 2021 proj.):
    - SME (<250): 76.1 (sectoral firm in total), contribution: 85.7, 41.0, 48.2
    - Large (≥250): 23.9 (sectoral firm in total), contribution: 14.3, 59.0, 51.8

### Data and methodology highlights (Appendix I: Cleaning of Orbis-Amadeus Data)
- Sample limited to unconsolidated firm accounts in non-farm, non-financial, private business sectors; drop NACE letters A, K, O, P, Q, R, S, and manually identified state-owned enterprises.
- Populate missing string variables (NACE code, firm founding year) using non-missing firm-year information for the same firm.
- Year adjustment rule: if reported “Closing Date” is after or on June 1st, assign current year; otherwise assign previous year.
- Duplicate firm-year records removed by keeping records with (i) “local registry filing”, (ii) “local GAAP”, or (iii) closing date closest to December 31st.
- Data quality filters:
  - Drop entire company if total assets or tangible fixed assets negative in any year.
  - Drop entire company if number of employees negative or greater than 2 million in any year.
  - Drop entire company if sales are negative in any year (no filter on revenue).
  - Drop entire company if total assets missing or if total assets, operating revenue, sales, and number of employees are missing simultaneously.
  - Drop companies with extreme ratios (employees per million assets, employees per million sales, sales over assets) above 99.9th percentile.
  - Drop firm-year observations with negative firm age.
  - Drop company if growth rate of total assets in any year is above 98th or below 2nd percentiles (to catch unit switches).
- Further trimming: drop top and bottom one percent of observations in liquidity ratio, asset-based solvency ratio, return on equity, and leverage ratio.
- Additional consistency checks ensuring non-negativity of current liabilities, current assets, loans, long-term debt, interest paid, trade creditors; and various ratios below unity as specified.

*IMF staff estimates (from the provided content).*

### 2019. Structural factors seem  to be driving  the low  level  of corporate investment,  while  household

### 1grcea2021002 - 2019. Structural factors seem  to be driving  the low  level  of corporate investment,  while  household

### Background
- Objective: raise potential growth and living standards in Greece through stronger capital accumulation and productivity.
- Staff baseline projection: real investment rates stabilize between 15–17 percent of GDP.
- Greek Government macro-fiscal baseline (in line with its new Stability and Growth Pact, SGP): projects the investment-to-GDP ratio will increase to around 17 percent by 2025.
- Recovery and Resilience National Plan (RRP) expectation: combination of funding and reforms will boost the level of private investment by 20 percent by 2026.
- Historical peak: pre-sovereign debt crisis total investment rate peak of 24 percent of GDP (accompanied by an unsustainable current account deficit).
- Recent performance: total investment growth close to 0 percent (2015–19), resulting in an investment-to-GDP ratio of 10 percent at end-2019.
- Comparative position: end-2019 investment-to-GDP ratio about 10 percentage points lower than the regional average; Greece ranks low compared to income peers in capital stock per capita.

### Capital Stock and Investment: Where Does Greece Stand?
- Total capital stock:
  - Total capital stock to GDP ratio: close to the euro area’s average.
  - Net capital stock per capita in Greece: about two thirds of the Euro Area’s average.
  - Net capital stock contracted by an average of 1.2 percent annually in the last decade; output declined by 26 percent over the same period.
- Sectoral composition:
  - Private sector holds 70 percent of total capital stock.
  - Public-to-private capital ratio: estimated close to 45 percent in 2019.
  - Growth-maximizing public-private ratio in literature for advanced economies: 42 percent.
- Infrastructure gaps and strengths:
  - Transportation infrastructure: nearly 30 percent lower road density and 60 percent lower railroad density compared to peers (partly due to geography).
  - Utility infrastructure: closer to EU average; electricity generation capacity and renewable energy production notable.
  - Water supply reliability: ranks high; water transport has gaps.
  - ICT and health infrastructure: slightly higher than regional peers except mobile phone penetration (7 percent below EU average).
  - Medical devices and equipment: higher-than-average; number of hospital beds as of 2019 was 16 percent below the EU.
  - Global Health Security Index (2018): Greece lower than regional average (noted that pandemic investments likely improved this).
- Investment rates by sector and industry:
  - Greece’s investment rate ranked among the lowest in the world by end-2019: investment-to-GDP ratio of 10 percent.
  - Industry drivers:
    - Real estate historically dominated investment (at peak represented 40 percent of total investment and nearly 10 percent of GDP).
    - From 2007 to 2019, investment in dwellings and other building structures dropped by more than 80 percent.
    - Real estate industry investment currently represents less than 1 percent of GDP (EU average 4.5 percent of GDP).
  - Cross-sector summary:
    - Business investment: widest gap (about 10 percentage points lower) compared to the EU.
    - Household investment: sharpest drop (over 10 percentage points since pre-crisis); now the lowest investment rate across sectors.
    - Government investment: relatively stable and higher than the EU average in relative terms over the past two decades; public capital expenditure boosted by EU funds and remained around 2 percentage points higher than EU level until 2011.

### Is Greece underinvesting? What is the Size of the Investment Gap?
- Analytical approaches applied to estimate total investment benchmark rates:
  - Golden Rule: neo-classical steady-state equilibrium level of investment.
  - Historical Benchmark: stylized transition investment dynamics from advanced European countries (Germany, France, Spain, Italy).
  - Predicted Norm: panel regression predicted investment rate determined by economic fundamentals and structural characteristics across 27 EU countries over three decades.
- Pros and cons (summary):
  - Golden Rule: steady-state benchmark invariant to initial conditions; requires knowledge of unobservable variables (e.g., social rate of time preference, SRTP).
  - Historical Benchmark: proven achievable and sustainable based on peers; assumes similar economic structure and balanced growth independent of external balance considerations.
  - Predicted Norm: grounded in Greece’s own fundamentals and structural characteristics; sample and model specification dependent.
- Estimated investment gap (as of 2019): ranges from 1.6 to 8 percent of GDP depending on method and specification.
  - Golden Rule: standard estimate points to an investment gap of about 4 percent of GDP (standard calculation for period 2015–19 adjustment; notes adjustments to labor share, savings preferences, or TFP methodology change gap size).
  - Historical Benchmark: largest estimated gap of 8 percent of GDP at end-2019 (represents an upper bound; assumes full convergence to mature economy characteristics and balanced growth).
  - Predicted Norm: smallest gap; suggests actual investment is only slightly below the predicted value based on Greece’s income level, fundamentals, and structural characteristics.

### Key implications and analytical focus
- The analysis focuses on total investment given measurement caveats:
  - Prevalence of micro family-owned firms: investment by unincorporated enterprises can be accounted as household investment.
  - Public capital expenditure can include capital transfers destined for SOE operations.
  - Arms or ship purchases cause volatile quarterly swings not necessarily indicative of private sector dynamics.
- Subsequent sections (in source) provide three estimation approaches (detailed in Annex I) and sectoral constraints to investment, concluding with policy implications.

*Prepared by Shiqing Hua, María Méndez, and Xin Cindy Xu; Ritzy Dumo and Daniel Murphy Pineda provided assistance. The paper benefitted from comments from Dimitris Malliaropulos, Marianthi Anastasatou, and Filippos Petroulakis (all BoG).*

### 12.      Greece’s low investment rate is an outlier compared against its peer group, but it

### 12.      Greece’s low investment rate is an outlier compared against its peer group, but it seems to be in line with its economic  structure.

### Investment gap: magnitude and benchmarking
- Historical benchmark indicates Greece’s investment gap has been negative (under-investing) through the period under consideration, signaling Greece as an outlier versus advanced EA peers.
- Two granular approaches that allow for country-specific characteristics suggest Greece sustained a positive gap (over-investment) prior to the sovereign debt crisis averaging between 4 and 8 percentage points.
- Since the crisis, Greece has under-invested, but the gap is narrower than its advanced economy status would suggest.
- At end-2019 the calculated gap ranges between 1.6 (predicted norm) and about 4 percent of GDP (golden rule), broadly in line with the lower bound of staff’s baseline investment rate.
- Absent structural factors, the investment benchmark predicted by Greece’s income level and economic fundamentals is close to 20 percent of GDP, yielding a large gap relative to actual investment at 8 percent of GDP. After adding structural constraints, the predicted investment rate gets closer to actual levels, with a much smaller gap of 1.6 percent of GDP once all structural factors are controlled for.

### Structural determinants of the investment gap
- The analysis suggests structural constraints explain the bulk of the investment gap.
- Golden rule sensitivity:
  - The size of the investment gap is sensitive to assumptions about the share of self-employed workers (and how much of their income accrues to labor instead of to capital) and savings preferences.
- Predicted-norm panel regression sensitivity:
  - The size of the investment gap is sensitive to model specifications; as more controls are added, the gap becomes smaller (see Table 2 specifications and Inv. Gap in 2019: -11.7, -10.2, -8.1, -6.4, -4.7, -3.9, -1.6 across models m1–m7).
- Main contributors to the sharp fall in the investment rate from the pre-crisis peak in 2007 to the record-low in 2019:
  - Top three drivers are structural factors: the services share, capital account openness (capturing the introduction of capital flow management measures in 2015), and regulatory quality.
  - These three structural factors together account for over 50 percent of the decline.
- Other control variables (income level and economic fundamentals) are statistically significant but have smaller explanatory power than structural factors.

### Sectoral drivers of investment

Business investment
- Corporate investment averaged 7 percent of GDP in the decade preceding the GFC and around 5½ percent of GDP since.
- 2020 EIB Survey findings:
  - 71 percent of firms indicate they invested the “right” amount in the past three years (EU average: 80 percent).
  - Two thirds of firms were operating at or about capacity in 2019.
  - Highest-cited barrier to investment: business regulation (improved from 95 percent of firms in 2019 to 86 percent in 2020).
  - Other cited barriers: energy costs (70 percent, down from 79 percent in 2019) and labor market regulations (71 percent, down from 78 percent in 2019).
- Additional business investment constraints:
  - Low returns on capital: gross return on capital of Non-Financial Corporations (NFCs) dropped from a pre-crisis peak of 34 percent in 2008 and 38 percent in 2011 to a historic low of 13.7 percent in 2019, 10 percentage points below the EA average.
  - High debt burden: average net debt-to-income ratio of Greek NFCs increased since 2013 and by 2016 had more than doubled compared to its 2008 level. Greek companies still feature the second highest gross debt burden for the EA-15 region.
  - Insufficient domestic savings and poor channeling as credit: NFC saving-to-GDP ratio was lower than the EU average pre-crisis and the gap widened; in 2018 NFC saving-to-GDP dipped into single-digit percentage points for the first time.
  - Reduced access to bank credit: pre-crisis credit boom vanished since the GFC, with negative credit growth until 2020; banks are deleveraging with high NPEs constraining credit supply.
  - EIB Survey: firms more likely to lack access to finance (13 percent) than EU average (6 percent); high cost of finance (12 percent vs EU average 5 percent) and collateral rules (10 percent) cited as barriers.

Household investment
- Household investment decline closely linked to income and savings dynamics.
- Greece has the lowest saving rate among EU countries by a wide margin: average national saving from 2010 to 2019 was 9.2 percent of GDP, 13 percentage points below the regional average.
- Household savings entered and have stayed in negative territory since 2012 despite recovery in disposable income and employment (driven in large part by low-paying, part-time, or seasonal jobs).
- Wealth effects and high degree of home ownership:
  - High home ownership and investment through real estate, combined with plummeting housing prices after the sovereign debt crisis, have driven the drop in investment as households dis-save to cope with reduced disposable income and reduced credit for housing expenditures.
  - Real estate sector started to recover prior to the pandemic (partly driven by tourism and Airbnb); transaction volume hard to ascertain due to halted statistics on housing starts/transactions after the GFC. Housing prices continued increasing through 2020.
- Household lending indicators shown in figures point to constrained loan growth and loan-to-disposable-income dynamics.

Public investment
- Public investment-to-GDP ratio has been broadly stable and slightly higher than the regional average despite limited fiscal space.
- Possible reasons: measurement issues (transfers classified as investment), rising importance of EU structural funds in a shrinking economy, and necessary investments to maintain a relatively large public capital stock held by State-Owned Enterprises.
- Greece’s public-to-private capital ratio stands above the regional average at 45 percent at end-2019.

### Policy conclusions and recommendations
- There is scope for expanding private investment through structural reforms and sound macro-financial policies.
- Methodological note: predicted norm approach, golden-rule equilibrium investment rate, and historical benchmark complement each other to gauge investment gap magnitude.
- Empirical findings: countries tend to invest more with lower debt burden, smaller service share, higher capital account openness, and better regulatory quality. High degree of self-employment appears to hold back investment, likely reflecting prevalence of micro firms constrained by market size, financing, or technology.
- Policy measures to boost private investment:
  - Product market reforms to encourage economies of scale.
  - Trade facilitation to integrate firms into global markets.
  - Reforms that raise private-sector productivity to improve earnings and encourage savings, thereby increasing household investment as disposable income grows.
  - Facilitate access to credit, good-paying jobs, and availability of infrastructure and services (transport, childcare, elder care, unemployment benefits) to support housing demand, especially among young people.
- Authorities’ actions already underway:
  - Recovery and Resilience Plan (underpinned by the 2020 National Growth Strategy) with reforms focusing on digitalization, green investments, up- and re-skilling the labor force, and encouraging economies of scale.
  - Near-term supportive policies: lower tax wedge and cuts to red-tape in business processes.
  - Efforts to improve public investment efficiency: interoperability of public investment budget, establishment of a Strategic Projects Pipeline and a Project Preparation Facility.
  - NGEU-grant financed investment, if fully executed, should boost growth impact of public investment.
  - Authorities requested, but then delayed, Public Investment Management Assessment technical assistance from the IMF.
- More ambitious reforms that could unlock higher private investment without endangering external sustainability:
  - Overhaul burdensome judicial system.
  - Complete delayed cadaster reform.
  - Adjust pensions to encourage private savings and labor force participation in older cohorts.
  - Tackle on-the-ground barriers to more competitive product markets and closed professions (as recommended by the Hellenic Competition Commission).
  - Complete long-promised privatizations, provided they attract greenfield projects that strengthen corporate governance and competition.
- Sequencing and prioritization:
  - Adequate sequencing of reforms and prioritization of public investment would maximize the investment potential of NGEU resources in the near term.
  - Analytical work to explore interaction of public and private capital in Greece could determine if supplementary or substitution effects dominate, helping policymakers allocate public investment to areas that encourage and maximize productive private sector investment.

*Source: IMF staff estimates and analysis as presented in the chapter.*

### Appendix I. Investment Gap Estimations

### Appendix I. Investment Gap Estimations

### Golden Rule framework and methodological approach
- The Golden Rule approach is interpreted as a lower bound to which a country’s investment rate eventually converges as it approaches its steady state level, underpinned by its structural characteristics and exogenous parameters.
- A neo-classical growth model (Cass-Koopmans), modified to allow for exogenous growth of labor-augmenting productivity, predicts convergence to a steady-state equilibrium in which income, consumption, and capital grow at a fixed rate equal to the sum of the growth rates of the labor force and labor-augmenting productivity.
- In the augmented neoclassical (Ramsey-Cass-Koopmans) growth model, the saving/investment rate is constant at its “golden-rule” value, with income, consumption, and capital growing at a fixed rate equal to the sum of the exogenous growth rates of the labor force and labor-augmenting productivity.
- The “golden-rule” is parameterized using α (capital share of output), p (social rate of time preference; SRTP), δ (depreciation rate), n (growth of the labor force), and g (rate of technical progress). (The source text contains the formula reference but the explicit formula line is not included in the supplied content.)

### Greece-specific estimation setup and data choices
- The estimate replicated for Greece follows IMF (2016) methodology used for EU countries, using Penn World Tables (PWT) data, updated to version 10.0.
- Adjustments to the capital share of output are based on Eurostat balance sheet data, reducing it by either 100 or 63 percent of self-employed income to reflect Greece’s prevalence of micro firms and large segment of self-employed.
- In IMF (2016), the SRTP is constant and set equal to 5 percent for all countries. The authors test alternative SRTPs given observed dis-saving in Greece over the past decade and potential future changes in saving behavior.
- The new PWT 9.0 data set now includes total factor productivity based on absorption (the “Welfare-relevant TFP”), and the authors note that the results do not vary much when using this measure.
- Scenarios for alternative parameter choices and adjustments are summarized in Table A (see Scenario list below).

### Golden Rule variations and scenario outcomes (Table A)
- The table of scenarios and gaps is presented verbatim in the source (format retained):

  - GR5-0.8Self-employed income accrues 100 percent to labor.  
  - GR41.9SRTP=6 (higher consumption).  
  - GR33.1Self-employed income accrues 67 percent to labor.  
  - GR13.8No adjustment to PWT values, SRTP=5.  
  - GR65.5Adjusted for welfare-relevant TFP (based on absorption).  
  - GR26.3SRTP=4.0 (lower consumption).  
  - GR76.9SRTP=3.5 (lowest consumption).

### Historical benchmark approach
- The Historical Benchmark provides a yardstick investment rate (for a given K/L ratio and technology) consistent with capital accumulation paths of selected advanced European economies during 1951–2011 that proved to be sustainable.
- Advantage: does not require assumptions about SRTP and position on the saddle-path. Disadvantage: assumes similarity in economic structures between selected countries and the country of interest.
- Benchmark values are calculated for each country and each point in time using TFP and population growth rates and the country’s capital-labor ratio to proxy a sustainable path of the investment rate during transition to steady state.
- The transition (“catch-up”) is driven by differences in real interest rates that affect intertemporal consumption and saving choices (Euler equation). When relative capital scarcity makes capital more productive, higher real returns stimulate saving and investment; as K/L rises, returns decline and investment rates fall toward steady state.
- The authors fit an approximated transition path for the investment rate using a Cobb-Douglas marginal product of capital specification and estimate parameters by fitting the transition path on a panel for Germany, France, Italy, and Spain over 1951–2011 (R2=0.87; asterisks denote statistical significance with *** at 1 percent and ** at 5 percent).
- Using fitted parameters and Greece’s K/L ratio and labor-augmenting productivity, they compute a sustainable “historical benchmark” investment rate that mimics earlier transition dynamics of advanced economies, while noting that this approach may be unsuitable for Greece given past periods where investment rates closer to the benchmark corresponded to unsustainable external and fiscal positions.

### Predicted Norm (panel regression) approach
- The “Predicted Norm” is estimated using a panel fixed-effects regression model for 27 EU countries (actual sample size varies by data availability) over the past three decades.
- Specification includes country and year fixed effects and controls for countries’ economic fundamentals, structural characteristics, and external conditions identified in the literature as significant determinants of investment.
- The model explicitly controls for structural factors such as regulatory efficiency, trade, and financial account openness; robustness checks include random effects, dynamic panel models, and alternative control variables (e.g., OECD Product Market Regulation indicators, World Economic Forum Global Competitiveness Index, World Uncertainty Index, Doing Business Index, NPL ratio in lieu of private sector/GDP, and export share GDP).
- Possible multicollinearity among regressors is detected but reported to not affect overall fit and predictive power. The final model specification is driven by data availability and best fit.

### Key quantitative points and parameter values referenced
- PWT data versions cited: PWT version 10.0 and PWT 9.0.
- SRTP values tested: SRTP=5 percent (baseline), SRTP=6, SRTP=4.0, SRTP=3.5.
- Historical panel fit statistic: R2=0.87 (statistical significance indicated with *** at 1 percent and ** at 5 percent).
- Scenario gap values as reported in Table A (embedded verbatim above).

_1grcea2021002 - Appendix I. Investment Gap Estimations_

---


_Source: https://www.imf.org/-/media/files/publications/cr/2021/english/1grcea2021002.pdf_
