## wpiea2020286-print-pdf

## Source details

**Canonical URL:** [wpiea2020286-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020286-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020286-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020286-print-pdf.pdf.json)

---

### I. INTRODUCTION — framing and key quantitative climate markers
- Global annual average surface temperature has already increased by 1.1 degrees Celsius since 1880.
- Extreme weather events are projected to worsen over the next century, as the global annual mean temperature increase by as much as 4 degrees Celsius (IPCC, 2007; Stern, 2007; IPCC, 2014).
- Focus: how climate change may affect sovereign credit ratings using the ND-GAIN dataset on climate change vulnerability and resilience.
- Motivation: rating agencies, market participants and policymakers need guidance on how physical climate risks affect sovereign borrowing costs and access to markets.

### III. DATA OVERVIEW — sample, dependent variable construction, and rating scale
- Panel dataset covering 67 countries over the period 1995–2017.
- Sovereign credit ratings drawn from Fitch, Moody’s and S&P.
- Dependent variable: 푅푅푖푖푖푖∗ — a country’s average credit rating at the end of each calendar year; in ordered-response framing, 푅푅푖푖푖푖∗ is latent and mapped to observed rating categories via cut-off points c1 … c20.
- Ratings grouped into 21 categories: observations below C assigned value 1; AAA observations receive value 21.

### Methodology and aggregation of credit ratings
- Three aggregate measures:
  - Ratings_Avg: simple average across S&P, Moody’s, Fitch.
  - Ratings_PCA: first principal component via PCA.
    - LR test rejects sphericity at the 1 percent level.
    - Kaiser-Meyer-Olkin Measure of Sampling Adequacy = 0.79.
    - First factor explains 98 percent of the variance in the standardized data.
  - MCDeq: PCA on a robust covariance/correlation matrix using the Minimum Covariance Determinant (MCD) estimator (Rousseeuw and Van Driessen, 1999).
    - Correlation coefficient between Ratings_PCA and MDCeq = 99, statistically significant at the 1 percent level.
- Linear transformation of qualitative ratings to an ordinal scale from 21 (AAA/Aaa/AAA) down to 0 (SD/D / DDD/DD/D).

### ND-GAIN indices and climate-change data
- ND-GAIN coverage: 184 countries over 1995–2017.
- Vulnerability: composite of 36 indicators across six sectors—food, water, health, ecosystem services, human habitat and infrastructure; covers exposure, sensitivity, and capacity to adapt.
- Resilience (readiness): composite of 9 indicators across economic, governance and social readiness.
- Note: ND-GAIN indices reflect changes in countries’ levels of vulnerability and resilience; they do not reflect fiscal insurance schemes for natural disasters.
- Summary statistics highlighted:
  - Mean climate change resilience = 33.7.
  - Minimum resilience = 0.2.
  - Maximum resilience = 71.3.
- Spatial/time observations: improvements in vulnerability and resilience in some regions (Canada, Australia, parts of South America and Asia, Europe, Russia, parts of South East Asia); little change in Sub-Saharan Africa for vulnerability; slight deterioration in the US resilience; improvements in Europe, Russia, parts of South East Asia and South America.

### Baseline econometric specification and controls
- Baseline regression (linear panel with fixed effects):
  - RR^∗_iit = α_i + δ_t + β CCC_it + γ X_i,t−1′ + ε_iit
  - CCC_it are climate vulnerability and resilience; X_i,t−1 is a vector of lagged macro controls to address endogeneity.
  - Robust standard errors clustered at the country level.
- Estimation approaches:
  - OLS on transformed numeric ratings.
  - Ordered response models: Ordered Probit and Ordered Logit (maximum likelihood).
  - Two-stage least squares (2SLS) with instrumental variables: lagged climate change variables as instruments.
  - IV diagnostics: Kleibergen-Paap and Hansen statistics reported.
- Controls included (expected sign):
  - Real GDP per capita (+)
  - Real GDP growth (+)
  - Inflation rate (+/-)
  - Debt-to-GDP ratio (-)
  - Foreign currency reserves (+)
  - Terms-of-trade index (+/-)
  - Unemployment rate (-)
- Data sources: IMF IFS, IMF WEO, World Bank WDI.

### Key empirical findings — baseline (full sample, 1995–2017)
- Macroeconomic controls behaved as expected:
  - Higher income and real GDP growth associated with better sovereign credit ratings.
  - Debt-to-GDP increase associated with lower credit ratings.
  - Higher foreign reserves associated with higher credit ratings.
  - Inflation and unemployment generally associated with worse credit ratings.
- Climate change vulnerability (full sample):
  - Mixed results across single-agency specifications; not consistently negative and statistically significant.
  - Using Ratings_Avg: Climate vulnerability coefficient = -0.232**, implying:
    - An increase of one percentage point in climate change vulnerability leads to a reduction of 0.23 percent in credit worthiness in the full sample of countries during the period 1995–2017.
  - Other reported coefficients:
    - Climate vulnerability (Moody’s) = 0.032 (not significant).
    - Climate vulnerability (S&P) = 0.094 (not significant).
    - Climate vulnerability (Fitch) = -0.162 (not significant).
    - Climate vulnerability (Ratings_PCA) = -0.047**.
- Climate change resilience (full sample):
  - Unambiguous and statistically significant positive impact across all specifications.
  - Using Ratings_Avg: Climate resilience coefficient = 0.086*** and summary statement:
    - An improvement of one percentage point in climate change resilience is associated with an increase of 0.09 percent in sovereign credit rating in the full sample of countries during the period 1995–2017.
  - Specifics:
    - Climate resilience (Moody’s) = 0.085**.
    - Climate resilience (S&P) = 0.098***.
    - Climate resilience (Fitch) = 0.082***.
    - Climate resilience (Ratings_PCA) = 0.017***.

### Heterogeneity: advanced economies vs. emerging markets
- Split-sample results:
  - Climate vulnerability:
    - No significant impact on credit ratings in advanced economies (AE).
    - Emerging markets (EM): Ratings_Avg coefficient = -0.690***.
      - Interpretation: An increase of one percentage point in climate change vulnerability leads to a reduction of 0.69 percent in credit worthiness in emerging market economies during 1995–2017.
      - This is three times the full-sample estimate of 0.23.
  - Climate resilience:
    - Positive and significant in both AE and EM.
    - Magnitude nearly three times greater in EM than AE:
      - EM effect on Ratings_Avg = 0.202***.
      - AE effect on Ratings_Avg = 0.083**.
    - Summary: resilience effect = 0.20 in emerging markets vs 0.08 in advanced economies.

### Robustness and endogeneity checks
- Ordered Probit and Ordered Logit, and IV estimates confirm baseline patterns:
  - Ordered Probit: Climate resilience = 0.127***; Climate vulnerability = -0.384***.
  - Ordered Logit: Climate resilience = 0.263***; Climate vulnerability = -0.627**.
  - IV (1 lag) and IV (2 lag) specifications:
    - IV resilience coefficients ≈ 0.057*–0.058*.
    - IV vulnerability coefficients ≈ -0.267** to -0.201* depending on lag.
  - Diagnostic statistics for IV:
    - Kleibergen-Paap statistic (p-value) reported as 0.000 for IV specifications.
    - Hansen statistic (p-values) reported as 0.434, 0.641, 0.352, and 0.147 across IV specifications.
- Overall robustness conclusion: results robust to ordered-response estimators and to IV approaches addressing potential reverse causality and endogeneity.

### Quantitative sample, fit and summary statistics
- Sample size and coverage:
  - Full-sample number of countries: up to 67 (varies by specification; baseline panel: 67 countries over 1995–2017).
  - Number of observations varies across specifications: examples include 980, 1,140, 1,068, 898.
- Fit statistics:
  - R-squared values in baseline OLS tables around 0.930 to 0.951; Ratings_Avg and Ratings_PCA models show R-squared = 0.944 in some specifications.
  - Country-group regressions: Number of countries AE = 25, EM = 28; observations 492 (AE) and 406 (EM); R-squared ≈ 0.866–0.878.
- Appendix summary statistics (selected):
  - Fitch: 1476; 14.50; 5.22; 1; 21
  - Moody’s: 1782; 14.09; 5.27; 0; 21
  - S&P: 1418; 14.12; 5.08; 2; 21
  - Real GDP per capita (log): 1771; 4.19; 2.27; -0.22; 10.46
  - Real GDP growth: 1770; 3.40; 4.54; -96.95; 71.53
  - Inflation rate: 1580; 6.78; 52.74; -4.86; 2075.88
  - Terms-of-Trade: 1756; 104.31; 27.90; 19.69; 321.35
  - Debt-to-GDP ratio: 1359; 52.91; 32.01; 2.48; 245.48
  - Foreign reserves (log): 1753; 9.37; 1.73; 3.87; 15.16
  - Unemployment rate: 1589; 8.00; 4.81; 0.30; 28.1
  - Climate change vulnerability: 1602; 39.05; 7.17; 25.98; 62.25
  - Climate change resilience: 1602; 47.94; 15.34; 19.10; 80.05

### Conclusions and policy implications
- Core empirical conclusions:
  - Climate change vulnerability has adverse effects on sovereign credit ratings; climate change resilience improves credit ratings.
  - Magnitudes reported:
    - Full sample: +1 percentage point vulnerability → -0.23 percent credit worthiness (Ratings_Avg = -0.232**).
    - Full sample: +1 percentage point resilience → +0.09 percent sovereign credit rating (Ratings_Avg = 0.086***).
    - Emerging markets: +1 percentage point vulnerability → -0.69 percent credit worthiness (Ratings_Avg = -0.690***).
    - Resilience effect ≈ 0.20 in emerging markets vs 0.08 in advanced economies.
- Policy recommendations (especially for developing countries):
  - Enhance structural resilience through cost-effective mitigation and adaptation.
  - Strengthen financial resilience via fiscal buffers and insurance schemes for climate-related disasters.
  - Improve economic diversification and policy management to reduce public-finance exposure to climate risks and thereby reduce borrowing costs associated with lower credit ratings.
- Final synthesis: econometric evidence supports that improving climate resilience can yield measurable sovereign credit benefits, while elevated climate vulnerability imposes sovereign-credit costs—effects are economically and statistically significant and disproportionately large for emerging market and developing economies.

*Source: wpiea2020286-print-pdf - References (excerpt provided).*

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

### wpiea2020286-print-pdf - References .............................................................................................................

### I. INTRODUCTION — framing and key quantitative climate markers
- The global annual average surface temperature has already increased by 1.1 degrees Celsius since 1880.
- Extreme weather events are projected to worsen over the next century, as the global annual mean temperature increase by as much as 4 degrees Celsius (IPCC, 2007; Stern, 2007; IPCC, 2014).
- Focus of the paper: how climate change may affect sovereign credit ratings using the ND-GAIN dataset on climate change vulnerability and resilience.
- Motivation: rating agencies, market participants and policymakers need guidance on how physical climate risks affect sovereign borrowing costs and access to markets.

### III. DATA OVERVIEW — sample, dependent variable construction, and rating scale
- Panel dataset covering 67 countries over the period 1995–2017.
- Sovereign credit ratings drawn from Fitch, Moody’s and S&P.
- Dependent variable: 푅푅푖푖푖푖∗ — a country’s average credit rating at the end of each calendar year; in ordered-response framing, 푅푅푖푖푖푖∗ is latent and mapped to observed rating categories via cut-off points c1 … c20.
- Ratings are grouped into 21 categories: observations below C are assigned the value of 1, while AAA observations receive the value of 21.

### Empirical approach and controls
- Estimation methodologies: ordinary least squares (OLS) and ordered response models.
- Controls: conventional determinants of credit ratings (economic factors and institutional characteristics), following the literature (Cantor and Packer, 1996; Mulder and Monfort, 2000; Amstad and Packer, 2015).
- Robustness: battery of sensitivity checks, including alternative measures of debt default, model specifications and estimation methodologies.

### Main empirical findings
- Climate change vulnerability has adverse effects on sovereign credit ratings after controlling for conventional macroeconomic determinants.
- Climate change resilience has a positive effect on sovereign credit ratings.
- Heterogeneity by country group:
  - Climate change vulnerability has no significant impact on credit ratings in advanced economies.
  - The magnitude and statistical significance of the vulnerability coefficient are much greater in developing countries due to weaker capacity to adapt and mitigate consequences.
  - Climate change resilience is statistically significant in both advanced and developing countries, but the magnitude of this effect is almost three times greater in emerging markets than in advanced economies.
- Results remain robust to alternative specifications and sensitivity checks.

### Relation to broader literature
- Connects two strands:
  - Determinants of sovereign credit ratings (Cantor and Packer, 1996; Afonso, 2003; Mulder and Monfort, 2000; Bissoondoyal-Bheenick, 2005; Mellios and Paget-Blanc, 2006; Amstad and Packer, 2015).
  - Macroeconomic impact of climate change and weather anomalies (Nordhaus, 1991; 1992; Cline, 1992; Gallup et al., 1999; Nordhaus, 2006; Dell et al., 2012; Burke et al., 2015; Acevedo et al., 2018; Burke and Tanutama, 2019; Kahn et al., 2019; Loyaza et al., 2012; Noy, 2009; Raddatz, 2009; Skidmore and Toya, 2002; Rasmussen, 2004; Cuaresma, 2010; Gassebner et al., 2010).
- Positions the paper as extending prior work showing climate impacts on government bond yields, spreads, and sovereign default probability (Cevik and Jalles, 2020a; 2020b).

### Policy implications and actionable recommendations
- For developing countries (relatively more vulnerable to climate risks):
  - Enhance structural resilience through cost-effective mitigation and adaptation.
  - Strengthen financial resilience through fiscal buffers and insurance schemes.
  - Improve economic diversification and policy management to cope with consequences of climate change for public finances and reduce borrowing costs associated with lower credit ratings.
- Rationale: Improving resilience can help raise sovereign creditworthiness and lower the cost of borrowing.

*Source: wpiea2020286-print-pdf - References (excerpt provided).*

### 1. In addition to using each rating agency ́s assessment separately, we also take three aggregate

### 1. In addition to using each rating agency ́s assessment separately, we also take three aggregate

### Methodology and aggregation of credit ratings
- Three aggregate measures of credit ratings:
  - Ratings_Avg: simple average across S&P, Moody’s, Fitch.
  - Ratings_PCA: first principal component extracted via Principal Component Analysis (PCA).
    - A likelihood ratio (LR) test rejects sphericity at the 1 percent level.
    - Kaiser-Meyer-Olkin Measure of Sampling Adequacy = 0.79.
    - The first factor explains 98 percent of the variance in the standardized data.
  - MCDeq: PCA repeated on a robust covariance/correlation matrix using the Minimum Covariance Determinant (MCD) estimator implemented via Rousseeuw and Van Driessen (1999).
    - Correlation coefficient between Ratings_PCA and MDCeq = 99, statistically significant at the 1 percent level.
- Linear transformation of qualitative ratings to an ordinal scale:
  - Ordinal scale 21 (AAA/Aaa/AAA) down to 0 (SD/D / DDD/DD/D) as shown in the source table mapping S&P, Moody’s, Fitch to integer scores.

### Data and climate-change indices
- ND-GAIN indices:
  - Coverage: 184 countries over the period 1995–2017.
  - Vulnerability: composite of 36 indicators covering exposure, sensitivity, and capacity to adapt across six sectors—food, water, health, ecosystem services, human habitat and infrastructure.
  - Resilience (readiness): composite of 9 indicators across three areas—economic, governance and social readiness.
  - Note: ND-GAIN indices reflect changes in countries’ levels of vulnerability and resilience (not necessarily forward-looking projections); they do not reflect fiscal insurance schemes for natural disasters.
- Summary statistics highlighted:
  - Mean climate change resilience = 33.7.
  - Minimum resilience = 0.2.
  - Maximum resilience = 71.3.
- Spatial/time observations:
  - Improvements in vulnerability and resilience in some regions (Canada, Australia, parts of South America and Asia, Europe, Russia, parts of South East Asia).
  - Little change in Sub-Saharan Africa for vulnerability; slight deterioration in the US resilience; improvements in Europe, Russia, parts of South East Asia and South America.

### Baseline econometric specification
- Baseline regression (linear panel with fixed effects):
  - RR^∗_iit = α_i + δ_t + β CCC_it + γ X_i,t−1′ + ε_iit
  - Where CCC_it are climate vulnerability and resilience; X_i,t−1 is a vector of lagged macro controls to address endogeneity.
  - Robust standard errors clustered at the country level.
- Estimation approaches employed:
  - OLS on transformed numeric ratings.
  - Ordered response models: Ordered Probit and Ordered Logit (maximum likelihood).
  - Two-stage least squares (2SLS) with instrumental variables (IV): lagged climate change variables used as instruments.
  - IV diagnostics: Kleibergen-Paap and Hansen statistics reported for validity of instruments.

### Controls included (expected sign)
- Real GDP per capita (+)
- Real GDP growth (+)
- Inflation rate (+/-)
- Debt-to-GDP ratio (-)
- Foreign currency reserves (+)
- Terms-of-trade index (+/-)
- Unemployment rate (-)
- Data sources: IMF IFS, IMF WEO, World Bank WDI.

### Key empirical findings — baseline (full sample, 1995–2017)
- General macro controls behaved as expected:
  - Higher income and real GDP growth associated with better sovereign credit ratings.
  - Debt-to-GDP increase associated with lower credit ratings.
  - Higher foreign reserves associated with higher credit ratings.
  - Inflation and unemployment generally associated with worse credit ratings.
- Climate change vulnerability (full sample):
  - Mixed results across single-agency specifications; not consistently negative and statistically significant.
  - Using Ratings_Avg: Climate vulnerability coefficient = -0.232**, implying:
    - An increase of one percentage point in climate change vulnerability leads to a reduction of 0.23 percent in credit worthiness in the full sample of countries during the period 1995–2017.
  - Other reported coefficients (Table 2):
    - Climate vulnerability (Moody’s) = 0.032 (not significant).
    - Climate vulnerability (S&P) = 0.094 (not significant).
    - Climate vulnerability (Fitch) = -0.162 (not significant).
    - Climate vulnerability (Ratings_PCA) = -0.047**.
- Climate change resilience (full sample):
  - Unambiguous and statistically significant positive impact across all specifications.
  - Using Ratings_Avg: Climate resilience coefficient = 0.086*** (Table 3) and summary statement:
    - An improvement of one percentage point in climate change resilience is associated with an increase of 0.09 percent in sovereign credit rating in the full sample of countries during the period 1995–2017.
  - Specifics (Table 3):
    - Climate resilience (Moody’s) = 0.085**.
    - Climate resilience (S&P) = 0.098***.
    - Climate resilience (Fitch) = 0.082***.
    - Climate resilience (Ratings_PCA) = 0.017***.

### Heterogeneity: advanced economies vs. emerging markets
- Split-sample results reveal substantial differences (Table 4):
  - Climate vulnerability:
    - No significant impact on credit ratings in advanced economies (AE).
    - Emerging markets (EM): Ratings_Avg coefficient = -0.690***.
      - Interpretation: An increase of one percentage point in climate change vulnerability leads to a reduction of 0.69 percent in credit worthiness in emerging market economies during 1995–2017.
      - This is three times the full-sample estimate of 0.23.
  - Climate resilience:
    - Positive and significant in both AE and EM.
    - Magnitude nearly three times greater in EM than AE:
      - EM effect on Ratings_Avg = 0.202*** (Table 4).
      - AE effect on Ratings_Avg = 0.083** (Table 4).
    - Summary earlier: resilience effect = 0.20 in emerging markets vs 0.08 in advanced economies.

### Robustness and endogeneity checks
- Ordered Probit and Ordered Logit, and IV estimates confirm baseline patterns (Table 5):
  - Ordered Probit: Climate resilience = 0.127***; Climate vulnerability = -0.384***.
  - Ordered Logit: Climate resilience = 0.263***; Climate vulnerability = -0.627**.
  - IV (1 lag) and IV (2 lag) specifications:
    - IV results show climate resilience coefficients around 0.057*–0.058* and climate vulnerability coefficients around -0.267** to -0.201* depending on lag specification.
  - Diagnostic statistics for IV:
    - Kleibergen-Paap statistic (p-value) reported as 0.000 for IV specifications.
    - Hansen statistic (p-values) reported as 0.434, 0.641, 0.352, and 0.147 across IV specifications, indicating instrument validity in those cases.
- Overall robustness conclusion:
  - Results robust to ordered-response estimators and to IV approaches addressing potential reverse causality and endogeneity.

### Quantitative sample and fit
- Sample size and coverage reported across tables:
  - Full-sample number of countries: up to 67 (varies by specification; baseline panel: 67 countries over 1995–2017).
  - Number of observations varies across specifications (examples): 980, 1,140, 1,068, 898.
  - R-squared values in baseline OLS tables around 0.930 to 0.951; Ratings_Avg and Ratings_PCA models show R-squared = 0.944 in some specifications.
  - Country-group regressions: Number of countries AE = 25, EM = 28; observations 492 (AE) and 406 (EM) in those splits; R-squared ≈ 0.866–0.878.

### Conclusions and policy implications
- Core empirical conclusions:
  - Climate change vulnerability has adverse effects on sovereign credit ratings; climate change resilience improves credit ratings.
  - Magnitudes:
    - Full sample: +1 percentage point vulnerability → -0.23 percent credit worthiness.
    - Full sample: +1 percentage point resilience → +0.09 percent sovereign credit rating.
    - Emerging markets: +1 percentage point vulnerability → -0.69 percent credit worthiness.
    - Resilience effect approximately 0.20 in emerging markets vs 0.08 in advanced economies.
- Policy implications (especially for developing countries):
  - Enhance structural resilience through cost-effective mitigation and adaptation.
  - Strengthen financial resilience via fiscal buffers and insurance schemes for climate-related disasters.
  - Improve economic diversification and policy management to reduce public-finance exposure to climate risks and thereby reduce borrowing costs associated with lower credit ratings.
- Final synthesis:
  - The econometric evidence supports that improving climate resilience can yield measurable sovereign credit benefits, while elevated climate vulnerability imposes sovereign-credit costs—effects are economically and statistically significant and disproportionately large for emerging market and developing economies.

*Source: wpiea2020286-print-pdf - authors’ calculations.*

### Appendix Table A1. List of Countries

### Appendix Table A1. List of Countries

### List of Countries by Region
- Africa: Cameroon, Gabon, Ghana, Kenya, Libya, Mali, Morocco, Mozambique, Nigeria, Seychelles, South Africa, Tunisia, Uganda
- Americas: Argentina, Barbados, Belize, Brazil, Canada, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Jamaica, Mexico, Panama, Peru, Trinidad and Tobago, Uruguay, United States
- Asia: Australia, China, Fiji, India, Indonesia, Japan, Korea, Malaysia, Mongolia, New Zealand, Papua New Guinea, Thailand, Vietnam
- Europe: Albania, Austria, Azerbaijan, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, France, Georgia, Germany, Hungary, Italy, Kazakhstan, Luxembourg, Netherlands, Norway, Russia, Slovenia, Sweden, Switzerland, Finland, Greece, Iceland, Ireland, Philippines, Poland, Portugal, Spain, Turkey, Ukraine, United Kingdom
- Middle East: Bahrain, Egypt, Israel, Jordan, Kuwait, Lebanon, Oman, Pakistan, Qatar, Saudi Arabia, Sri Lanka, United Arab Emirates

### Appendix Table A2. Summary Statistics
- Variables: Observations; Mean; Std. Dev.; Min.; Max.
- Credit ratings
  - Fitch: 1476; 14.50; 5.22; 1; 21
  - Moody’s: 1782; 14.09; 5.27; 0; 21
  - S&P: 1418; 14.12; 5.08; 2; 21
- Real GDP per capita (log): 1771; 4.19; 2.27; -0.22; 10.46
- Real GDP growth: 1770; 3.40; 4.54; -96.95; 71.53
- Inflation rate: 1580; 6.78; 52.74; -4.86; 2075.88
- Terms-of-Trade: 1756; 104.31; 27.90; 19.69; 321.35
- Debt-to-GDP ratio: 1359; 52.91; 32.01; 2.48; 245.48
- Foreign reserves (log): 1753; 9.37; 1.73; 3.87; 15.16
- Unemployment rate: 1589; 8.00; 4.81; 0.30; 28.1
- Climate change vulnerability: 1602; 39.05; 7.17; 25.98; 62.25
- Climate change resilience: 1602; 47.94; 15.34; 19.10; 80.05

*Source: Appendix Table A1 and Appendix Table A2, wpiea2020286-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020286-print-pdf.pdf_