## References (wpiea2020079-print-pdf)

## Source details

**Canonical URL:** [References (wpiea2020079-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020079-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020079-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020079-print-pdf.pdf.json)

---

### I. Introduction — framing and scope
- Climate change characterized as a systemic risk; global average surface temperature has risen by 1.1 degrees Celsius since 1880 and projected increases in global annual mean temperatures by as much as 4 degrees Celsius over the next century (IPCC, 2007; Stern, 2007; IPCC, 2014).
- Study scope:
  - Sample: 98 advanced and developing countries, 1995–2017.
  - Focus: how vulnerability and resilience to climate change affect the cost of sovereign borrowing (10-year foreign-currency-denominated government bond yields and spreads; EMBIG spreads for broader emerging market coverage).
- Key definitions (ND-GAIN):
  - Vulnerability: exposure, sensitivity, and capacity to adapt (36 indicators).
  - Resilience (ND-GAIN “readiness”): capacity to apply economic investments into adaptation actions (9 indicators).
- Main empirical summary:
  - Climate vulnerability: statistically and economically significant positive impact on government bond yields and spreads.
  - Climate resilience: statistically and economically significant negative (dampening) effect on government borrowing costs.
  - Effects larger and more significant in developing countries with weaker adaptive capacity.
- Policy takeaway: enhancing resilience (mitigation and adaptation), financial resilience (fiscal buffers, insurance), economic diversification, and policy management can reduce the climate-related premium on sovereign borrowing.

### II. Literature overview — two threads synthesized
- Determinants of sovereign bond yields and spreads:
  - Government borrowing costs depend on macroeconomic fundamentals and institutional factors (numerous cited studies).
- Macroeconomic impact of climate change:
  - Higher temperatures and weather anomalies reduce economic growth, especially in developing/hotter countries.
  - Natural disasters, human capital accumulation, and trade balance effects documented in the literature.
- Pricing of climate-related risks:
  - Scarcity of research; some evidence that physical risk and temperature shifts affect asset valuations and bond pricing.

### III. Data overview — sources and construction
- Sample and period:
  - Panel of annual observations covering 98 advanced and developing countries over 1995–2017.
- Dependent variables:
  - 10-year foreign-currency-denominated government bond yields and spreads from Bloomberg.
  - EMBIG sovereign bond spreads from J.P. Morgan Emerging Market Bond Index Global for broader emerging market coverage.
- Main explanatory variables:
  - ND-GAIN indices of vulnerability and resilience (45 indicators: 36 vulnerability; 9 resilience). ND-GAIN database covers 184 countries over 1995–2017.
- Controls:
  - Level and growth rate of real GDP, consumer price inflation, public debt-to-GDP ratio, budget balance-to-GDP ratio, international reserves as a share of GDP, government effectiveness, bureaucratic quality.
- Data properties and diagnostics:
  - Variables stationary after logarithmic transformation per Im-Pesaran-Shin (2003) panel unit root tests.
  - Durbin–Watson and log-likelihood ratio tests: no significant first-order autocorrelation and presence of cross-sectional correlation.

### IV. Empirical strategy and key results
- Baseline model:
  - Reduced-form panel with country and year fixed effects; dynamic specifications include lagged dependent variable.
  - Key regressors: vulnerability_{i,t} and resilience_{i,t} (ND-GAIN).
  - Robust standard errors clustered at the country level.
- Estimation methods and endogeneity checks:
  - Fixed effects, 2SLS with lagged climate indices as instruments (Kleibergen-Paap and Hansen statistics reported), and System GMM (Arellano and Bover, 1995; Blundell and Bond, 1998) one-step estimator with instrument proliferation addressed.
  - AR(1) and AR(2) p-values reported; high first-order autocorrelation but no significant second-order autocorrelation; Hansen J-test supports internal instruments in GMM.
- Baseline coefficient ranges (selected exact estimates reported):
  - Climate vulnerability coefficient range: 0.579 to 2.526 depending on specification (always positive and statistically significant).
    - Benchmark interpretation reported: a one percentage point increase in climate change vulnerability is associated with an increase of 0.58 percent in long-term government bond spreads.
  - Climate resilience coefficient range: -0.164 to -0.405 depending on specification (always negative and statistically significant).
    - Benchmark interpretation reported: 1 percent improvement in climate change resilience is associated with a decrease of 0.15 percent in long-term government bond spreads.
- Country-group heterogeneity:
  - Advanced economies: climate vulnerability and resilience have no pronounced impact on government bond spreads.
  - Developing countries (benchmark specification):
    - 1 percent increase in climate change vulnerability leads to an increase of 3.11 percent in long-term government bond spreads for emerging market economies.
    - 1 percent improvement in climate change resilience lowers bond spreads by 0.75 percent for emerging market economies.
  - With average long-term government bond spreads in the developing-country sample of about 500 basis points:
    - A one percentage point increase in vulnerability implies an increase in sovereign debt risk premia by 15.55 basis points.
    - A one percentage point improvement in resilience implies a decrease in sovereign debt risk premia by 3.75 basis points.
    - Difference between countries in the 25th and 75th quintile: 233 basis points for vulnerability and 56 basis points for resilience.

### V. Appendix—Selected numeric results (Appendix Tables A1–A4)
- Appendix Table A1 (Alternative Measures, Specs 1–6; robust s.e. in parentheses):
  - Climate vulnerability:
    - Spec 1: 0.661* (0.360)
    - Spec 2: 0.579* (0.347)
    - Spec 3: 2.730*** (0.950)
    - Spec 4: 3.150*** (0.961)
  - Climate resilience:
    - Spec 1: -0.164*** (0.063)
    - Spec 2: -0.151** (0.060)
    - Spec 3: -0.610*** (0.160)
    - Spec 4: -0.740*** (0.175)
  - Real GDP and other controls reported across specs (examples):
    - Real GDP Spec 1: 2.722* (1.553); Spec 4: 17.100** (0.309)
    - Real GDP growth Spec 1: -0.183*** (0.07)
    - Inflation Spec 1: 0.357*** (0.127)
    - Debt Spec 1: 0.057*** (0.010)
  - Sample sizes and fit:
    - Number of countries: 53 (Specs 1–3), 44 (Specs 4–6)
    - Observations: 823 (Specs 1–3), 518 (Specs 4–6)
    - R-squared: 0.73 (Specs 1–3), 0.83 (Specs 4–6)
- Appendix Table A2 (Excluding Outliers, Specs 1–6):
  - Climate vulnerability:
    - Spec 1: 2.844*** (0.85)
    - Spec 2: 2.949*** (0.837)
    - Spec 3: 0.751** (0.428)
    - Spec 4: 0.885** (0.472)
  - Climate resilience:
    - Spec 1: -0.546*** (0.183)
    - Spec 2: -0.530*** (0.183)
    - Spec 3: -0.203*** (0.072)
    - Spec 4: -0.223*** (0.083)
  - Sample details:
    - Observations vary by spec: 804, 817, 687, 671, 686, 587
    - % excluded: 19%, 18%, 31%, 18%, 17%, 29%
    - R-squared examples: 0.48, 0.46, 0.49 (Specs 1–3); 0.71, 0.72, 0.72 (Specs 4–6)
- Appendix Table A3 (2SLS Estimations, Specs 1–6):
  - Climate vulnerability:
    - Spec 1: 3.024*** (0.894)
    - Spec 2: 2.727*** (0.815)
    - Spec 3: 0.806*** (0.429)
    - Spec 4: 0.686*** (0.41)
  - Climate resilience:
    - Spec 1: -0.438*** (0.150)
    - Spec 2: -0.302** (0.122)
    - Spec 3: -0.171*** (0.06)
    - Spec 4: -0.161*** (0.058)
  - Diagnostics:
    - Kleibergen-Paap statistic (p-value): 0.000.000.000.000.000.00
    - Hansen statistic (p-value): 0.170.030.100.620.470.78
    - R-squared: 0.47, 0.49, 0.50 (Specs 1–3); 0.72, 0.73, 0.73 (Specs 4–6)
- Appendix Table A4 (Dynamic Estimations S-GMM, Specs 1–6):
  - Lagged dependent variable:
    - Spec 1: 0.6049*** (0.034)
    - Spec 2: 0.6134*** (0.034)
    - Spec 4: 0.6350*** (0.067)
  - Vulnerability (examples):
    - Spec 1: 0.1840*** (0.059)
    - Spec 2: 0.1945*** (0.045)
    - Spec 3: 0.1324* (0.080)
    - Spec 4: 0.1635** (0.077)
  - Resilience (examples):
    - Spec 1: 0.0063 (0.071)
    - Spec 3: -0.0072 (0.078)
  - AR1 (p-value) examples: 0.002, 0.002, 0.002 (Specs 1–3)
  - AR2 (p-value) examples: 0.399, 0.380, 0.442 (Specs 1–3)
  - Hansen statistic (p-value) examples: 0.369, 0.416, 0.992 (Specs 1–3)

### VI. Robustness checks and sensitivity analyses
- Alternative dependent variables: 10-year bond yields (sample of 53 countries) and EMBIG spreads (44 emerging market economies) yield broadly similar findings.
- Outlier treatment: truncating sample at 5th and 95th percentiles yields similar results.
- Instrumental variables: 2SLS with lagged climate indices supports detrimental effect of climate vulnerability and beneficial effect of resilience.
- Dynamic specification: System GMM broadly consistent with baseline; coefficient on climate change resilience less robust in GMM, and S-GMM is demanding with limited, unbalanced panels.

### VII. Conclusion — implications and policy recommendations
- Empirical conclusions:
  - Climate vulnerability significantly raises the cost of government borrowing; climate resilience significantly lowers it.
  - Effects materially larger and more statistically significant in developing countries with weaker adaptation and mitigation capacity.
  - Results robust across multiple measures, specifications, and estimation methods, subject to estimator limitations noted.
- Policy implications for developing countries:
  - Enhance structural resilience through mitigation and adaptation investments.
  - Strengthen financial resilience with fiscal buffers and insurance schemes.
  - Improve economic diversification and policy management to cope with climate-related public finance risks.

*Source: wpiea2020079-print-pdf - References (IMF working paper content covering 1995–2017 analysis).*

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

### References

### I. Introduction — framing and scope
- Climate change is characterized as a systemic risk to the global economy with the global average surface temperature rising by 1.1 degrees Celsius since 1880 and projected increases in global annual mean temperatures by as much as 4 degrees Celsius over the next century (IPCC, 2007; Stern, 2007; IPCC, 2014).
- The paper examines how vulnerability and resilience to climate change affect the cost of sovereign borrowing in 98 advanced and developing countries over the period 1995–2017, extending conventional determinants of government bond yields and spreads using ND-GAIN climate vulnerability and resilience indices.
- Key conceptual distinction:
  - Vulnerability: a country’s exposure, sensitivity, and capacity to adapt to the impacts of climate change (36 indicators contributing to vulnerability).
  - Resilience (ND-GAIN “readiness”): a country’s capacity to apply economic investments and convert them to adaptation actions (9 indicators constituting resilience).
- Main empirical result summary:
  - Climate change vulnerability has a statistically and economically significant positive impact on government bond yields and spreads.
  - Climate change resilience has a statistically and economically significant dampening (negative) effect on the cost of government borrowing.
  - Effects are much greater and more significant in developing countries with weaker adaptive capacity.
- Policy takeaway: enhancing resilience (mitigation and adaptation), financial resilience (fiscal buffers, insurance), economic diversification, and policy management can reduce the climate-related premium on sovereign borrowing.

### II. Literature overview — two threads synthesized
- Determinants of sovereign bond yields and spreads:
  - Government borrowing costs depend on macroeconomic fundamentals and institutional factors (numerous references: Engen and Hubbard, 2004; Kinoshita, 2006; Ardagna and others, 2007; Laubach, 2009; Hilscher and Nosbusch, 2010; Gómez-Puig and others, 2014; etc.).
- Macroeconomic impact of climate change:
  - Higher temperatures and weather anomalies reduce economic growth, especially in developing/hotter countries (Gallup, Sachs, and Mellinger, 1999; Nordhaus, 2006; Dell, Jones, and Olken, 2012; Burke, Hsiang, and Miguel, 2015; Acevedo and others, 2018; Burke and Tanutama, 2019; Kahn and others, 2019).
  - Natural disasters, human capital accumulation, and trade balance effects documented (Loyaza and others, 2012; Noy, 2009; Raddatz, 2009; Skidmore and Toya, 2002; Rasmussen, 2004; Cuaresma, 2010; Gassebner and others, 2010).
- Scarcity of research on how climate-related risks are priced in financial markets, with some evidence that physical risk and temperature shifts affect asset valuations and bond pricing (Bansal, Kiku, and Ochoa, 2016; IMF, 2020; Bernstein, Gustafson, and Lewis, 2019; Painter, 2019; Kling and others, 2018).

### III. Data overview — sources and construction
- Sample and period:
  - Panel of annual observations covering 98 advanced and developing countries over 1995–2017.
- Dependent variables:
  - 10-year foreign-currency-denominated government bond yields and spreads vis-à-vis the U.S. benchmark from Bloomberg.
  - Alternative measure: EMBIG sovereign bond spreads on external U.S. dollar-denominated debt from J.P. Morgan Emerging Market Bond Index Global (EMBIG) to broaden coverage of emerging market economies and developing countries.
- Main explanatory variables:
  - ND-GAIN indices of vulnerability and resilience (composite indices based on 45 indicators: 36 vulnerability variables; 9 resilience variables). ND-GAIN database covers 184 countries over 1995–2017.
- Controls:
  - Level and growth rate of real GDP, consumer price inflation, public debt-to-GDP ratio, budget balance-to-GDP ratio, international reserves as a share of GDP, government effectiveness, and bureaucratic quality.
- Data properties and diagnostics:
  - Variables used in the analysis are stationary after logarithmic transformation per Im-Pesaran-Shin (2003) panel unit root tests (results available upon request).
  - Durbin–Watson and log-likelihood ratio tests indicate no significant first-order autocorrelation and presence of cross-sectional correlation in the dataset.

### IV. Empirical strategy and key results
- Baseline model (reduced-form panel with country and year fixed effects):
  - Dependent variable y_{i,t} = government bond spreads or yields; includes lagged dependent variable in dynamic specifications.
  - Key regressors: vulnerability_{i,t} and resilience_{i,t} (ND-GAIN).
  - Controls: macroeconomic and institutional variables (real GDP, real GDP growth, inflation, debt-to-GDP, budget balance, international reserves, government effectiveness, bureaucratic quality).
  - Robust standard errors clustered at the country level.
- Estimation methods and endogeneity checks:
  - Standard fixed effects.
  - 2SLS with lagged climate indices as instruments (validated by Kleibergen-Paap and Hansen statistics).
  - System GMM (Arellano and Bover, 1995; Blundell and Bond, 1998) dynamic specification using one-step estimator; instrument proliferation addressed following Roodman (2009). AR(1) and AR(2) reported as p-values; high first-order autocorrelation but no significant second-order autocorrelation; Hansen J-test supports internal instruments.
- Baseline coefficient findings (selected exact estimates reported in the source):
  - Coefficient on climate change vulnerability ranges between 0.579 and 2.526 depending on specification (always positive and statistically significant).
    - Benchmark interpretation: a one percentage point increase in climate change vulnerability is associated with an increase of 0.58 percent in long-term government bond spreads.
  - Coefficient on climate change resilience ranges between -0.164 and -0.405 depending on specification (always negative and statistically significant).
    - Benchmark interpretation: 1 percent improvement in climate change resilience is associated with a decrease of 0.15 percent in long-term government bond spreads.
- Country-group heterogeneity (advanced vs developing):
  - In advanced economies: climate vulnerability and resilience have no pronounced impact on government bond spreads.
  - In developing countries (benchmark specification):
    - 1 percent increase in climate change vulnerability leads to an increase of 3.11 percent in long-term government bond spreads for emerging market economies.
    - 1 percent improvement in climate change resilience lowers bond spreads by 0.75 percent for emerging market economies.
  - With average long-term government bond spreads in the developing-country sample of about 500 basis points, the paper computes:
    - A one percentage point increase in vulnerability (or resilience) implies an increase (or decrease) in sovereign debt risk premia by 15.55 basis points (or 3.75 basis points).
    - Difference between countries in the 25th and 75th quintile amounts to 233 basis points for vulnerability and 56 basis points for resilience.
- Robustness checks and sensitivity analyses:
  - Replacing bond spreads with 10-year bond yields (sample of 53 countries) and EMBIG spreads (44 emerging market economies) yields broadly similar findings (Appendix Table A1).
  - Truncating sample at 5th and 95th percentiles to exclude outliers yields similar results (Appendix Table A2).
  - 2SLS with lagged climate indices supports detrimental effect of climate vulnerability and beneficial effect of resilience (Appendix Table A3).
  - System GMM dynamic specification broadly consistent with baseline; coefficient on climate change resilience less robust in GMM (Appendix Table A4). The system GMM estimator is noted as demanding with limited, unbalanced panels.

### V. Conclusion — implications and recommendations
- Empirical conclusions:
  - Climate vulnerability significantly raises the cost of government borrowing; climate resilience significantly lowers it.
  - Effects are materially larger and more statistically significant in developing countries with weaker adaptation and mitigation capacity.
  - Findings are robust across multiple measures, specifications, and estimation methods, subject to the limitations noted for specific estimators.
- Policy implications emphasized for developing countries:
  - Enhance structural resilience through mitigation and adaptation investments.
  - Strengthen financial resilience with fiscal buffers and insurance schemes.
  - Improve economic diversification and policy management to cope with climate-related public finance risks.

*Source: wpiea2020079-print-pdf - References (IMF working paper content covering 1995–2017 analysis).*

### Appendix Table A1. Climate Change and Sovereign Risk—Alternative Measures

### Appendix Tables A1–A4. Climate Change and Sovereign Risk

### Appendix Table A1. Climate Change and Sovereign Risk—Alternative Measures (Specifications 1–6)
- Dependent variables:
  - Specifications 1–3: Bond yields
  - Specifications 4–6: EMBIG spread
- Country group:
  - Specifications 1–3: All
  - Specifications 4–6: Developing
- Key coefficients and standard errors (in parentheses):
  - Climate vulnerability:
    - Spec 1: 0.661* (0.360)
    - Spec 2: 0.579* (0.347)
    - Spec 3: 2.730*** (0.950)
    - Spec 4: 3.150*** (0.961)
  - Climate resilience:
    - Spec 1: -0.164*** (0.063)
    - Spec 2: -0.151** (0.060)
    - Spec 3: -0.610*** (0.160)
    - Spec 4: -0.740*** (0.175)
  - Real GDP:
    - Spec 1: 2.722* (1.553)
    - Spec 2: 1.189 (1.225)
    - Spec 3: 2.756* (1.566)
    - Spec 4: 17.100** (0.309)
    - Spec 5: 10.485** (0.282)
    - Spec 6: 19.650** (0.327)
  - Real GDP growth:
    - Spec 1: -0.183*** (0.07)
    - Spec 2: -0.192*** (0.069)
    - Spec 3: -0.200*** (0.069)
    - Spec 4: -0.034*** (0.008)
    - Spec 5: -0.034*** (0.008)
    - Spec 6: -0.034*** (0.009)
  - Inflation:
    - Spec 1: 0.357*** (0.127)
    - Spec 2: 0.358*** (0.127)
    - Spec 3: 0.352*** (0.125)
    - Spec 4: 0.004** (0.002)
    - Spec 5: 0.004** (0.002)
    - Spec 6: 0.004** (0.002)
  - Debt:
    - Spec 1: 0.057*** (0.010)
    - Spec 2: 0.057*** (0.010)
    - Spec 3: 0.057*** (0.010)
    - Spec 4: 0.018*** (0.002)
    - Spec 5: 0.017*** (0.002)
    - Spec 6: 0.018*** (0.002)
  - Budget balance:
    - Spec 1: 0.0930 (0.044)
    - Spec 2: 0.0820 (0.043)
    - Spec 3: 0.0970 (0.044)
    - Spec 4: 0.0130 (0.010)
    - Spec 5: 0.0130 (0.010)
    - Spec 6: 0.013 (0.010)
  - International reserves:
    - Spec 1: -0.003 (0.011)
    - Spec 2: -0.006 (0.011)
    - Spec 3: -0.007 (0.011)
    - Spec 4: -0.066** (0.010)
    - Spec 5: -0.015** (0.009)
    - Spec 6: -0.017** (0.009)
  - Government effectiveness:
    - Spec 1: -1.363* (0.725)
    - Spec 2: -1.093 (0.687)
    - Spec 3: -0.910 (0.69)
    - Spec 4: -0.671*** (0.128)
    - Spec 5: -0.649*** (0.141)
    - Spec 6: -0.642*** (0.143)
  - Bureaucratic quality:
    - Spec 1: -0.486 (0.724)
    - Spec 2: -0.148 (0.721)
    - Spec 3: -0.271 (0.724)
    - Spec 4: -0.376** (0.172)
    - Spec 5: -0.368** (0.182)
    - Spec 6: -0.351** (0.181)
- Sample and fit:
  - Number of countries: 53 (Specs 1–3), 44 (Specs 4–6)
  - Number of observations: 823 (Specs 1–3), 518 (Specs 4–6)
  - Country FE: Yes (all)
  - Year FE: Yes (all)
  - R-squared:
    - Specs 1–3: 0.73
    - Specs 4–6: 0.83
- Notes:
  - Robust standard errors in brackets. A constant is included but not shown. *** p<0.01, ** p<0.05, * p<0.1

### Appendix Table A2. Climate Change and Sovereign Risk—Excluding Outliers (Specifications 1–6)
- Dependent variable: Bond spreads (all specs)
- Key coefficients and standard errors (in parentheses):
  - Climate vulnerability:
    - Spec 1: 2.844*** (0.85)
    - Spec 2: 2.949*** (0.837)
    - Spec 3: 0.751** (0.428)
    - Spec 4: 0.885** (0.472)
  - Climate resilience:
    - Spec 1: -0.546*** (0.183)
    - Spec 2: -0.530*** (0.183)
    - Spec 3: -0.203*** (0.072)
    - Spec 4: -0.223*** (0.083)
  - Real GDP:
    - Spec 1: 3.684** (1.764)
    - Spec 2: 2.463* (1.300)
    - Spec 3: 5.595*** (2.084)
  - Real GDP growth:
    - Spec 1: -0.190** (0.074)
    - Spec 2: -0.213*** (0.073)
    - Spec 3: -0.227*** (0.075)
  - Inflation:
    - Spec 1: 0.354*** (0.128)
    - Spec 2: 0.355*** (0.129)
    - Spec 3: 0.342*** (0.125)
  - Debt:
    - Spec 1: 0.066*** (0.012)
    - Spec 2: 0.071*** (0.013)
    - Spec 3: 0.081*** (0.014)
  - Budget balance:
    - Spec 1: 0.0920 (0.05)
    - Spec 2: 0.0660 (0.054)
    - Spec 3: 0.080 (0.058)
  - International reserves:
    - Spec 1: -0.038* (0.023)
    - Spec 2: -0.007 (0.015)
    - Spec 3: -0.041* (0.024)
  - Government effectiveness:
    - Spec 1: -1.689** (0.801)
    - Spec 2: -1.491* (0.797)
    - Spec 3: -1.634* (0.863)
  - Bureaucratic quality:
    - Spec 1: -0.719 (0.872)
    - Spec 2: -0.472 (0.871)
    - Spec 3: -0.722 (0.979)
- Sample and fit:
  - Number of observations vary by spec: 804, 817, 687, 671, 686, 587
  - % excluded: 19%, 18%, 31%, 18%, 17%, 29%
  - Country FE: Yes (all)
  - Year FE: Yes (all)
  - R-squared:
    - Specs 1–3: 0.48, 0.46, 0.49
    - Specs 4–6: 0.71, 0.72, 0.72
- Notes:
  - Robust standard errors in brackets. A constant is included but not shown. *** p<0.01, ** p<0.05, * p<0.1

### Appendix Table A3. Climate Change and Sovereign Risk—2SLS Estimations (Specifications 1–6)
- Dependent variable: Bond spreads (all specs)
- Key coefficients and standard errors (in parentheses):
  - Climate vulnerability:
    - Spec 1: 3.024*** (0.894)
    - Spec 2: 2.727*** (0.815)
    - Spec 3: 0.806*** (0.429)
    - Spec 4: 0.686*** (0.41)
  - Climate resilience:
    - Spec 1: -0.438*** (0.150)
    - Spec 2: -0.302** (0.122)
    - Spec 3: -0.171*** (0.06)
    - Spec 4: -0.161*** (0.058)
  - Real GDP:
    - Spec 1: 3.6495** (1.725)
    - Spec 2: 1.774* (1.265)
    - Spec 3: 3.6331** (1.722)
  - Real GDP growth:
    - Spec 1: -0.192*** (0.067)
    - Spec 2: -0.200*** (0.066)
    - Spec 3: -0.210*** (0.066)
  - Inflation:
    - Spec 1: 0.352*** (0.121)
    - Spec 2: 0.354*** (0.121)
    - Spec 3: 0.347*** (0.119)
  - Debt:
    - Spec 1: 0.064*** (0.011)
    - Spec 2: 0.064*** (0.01)
    - Spec 3: 0.064*** (0.011)
  - Budget balance:
    - Spec 1: 0.0900 (0.043)
    - Spec 2: 0.0850 (0.042)
    - Spec 3: 0.102 (0.043)
  - International reserves:
    - Spec 1: -0.005 (0.011)
    - Spec 2: -0.007 (0.011)
    - Spec 3: -0.009 (0.011)
  - Government effectiveness:
    - Spec 1: -1.598** (0.723)
    - Spec 2: -1.265** (0.684)
    - Spec 3: -1.082** (0.686)
  - Bureaucratic quality:
    - Spec 1: -0.181 (0.715)
    - Spec 2: 0.172 (0.716)
    - Spec 3: 0.049 (0.717)
- Sample and diagnostics:
  - Number of countries: 54, 54, 54, 53, 53, 53 (as listed)
  - Number of observations: 938 (Specs 1–3), 801 (Specs 4–6)
  - Country FE: Yes (all)
  - Year FE: Yes (all)
  - Kleibergen-Paap statistic (p-value): 0.000.000.000.000.000.00
  - Hansen statistic (p-value): 0.170.030.100.620.470.78
  - R-squared:
    - Specs 1–3: 0.47, 0.49, 0.50
    - Specs 4–6: 0.72, 0.73, 0.73
- Notes:
  - Robust standard errors in brackets. A constant is included but not shown. The Kleibergen-Paap test null is that the structural equation is underidentified. Stock-Yogo critical values were applied. The Hansen test is a test of overidentifying restrictions. *** p<0.01, ** p<0.05, * p<0.1

### Appendix Table A4. Climate Change and Sovereign Risk—Dynamic Estimations (Specifications 1–6)
- Dependent variables:
  - Specs 1–3: Bond spreads (Estimator: S-GMM)
  - Specs 4–6: EMBIG (Estimator: S-GMM)
- Country group:
  - Specs 1–3: All
  - Specs 4–6: Developing
- Key coefficients and standard errors (in parentheses where provided):
  - Lagged Dependent Variable:
    - Spec 1: 0.6049*** (0.034)
    - Spec 2: 0.6134*** (0.034)
    - Spec 3: 0.6063*** (0.028)
    - Spec 4: 0.6350*** (0.067)
    - Spec 5: 0.5032*** (0.095)
    - Spec 6: 0.5130*** (0.093)
  - Vulnerability:
    - Spec 1: 0.1840*** (0.059)
    - Spec 2: 0.1945*** (0.045)
    - Spec 3: 0.1324* (0.080)
    - Spec 4: 0.1635** (0.077)
  - Resilience:
    - Spec 1: 0.0063 (0.071)
    - Spec 2: 0.0052 (0.039)
    - Spec 3: -0.0072 (0.078)
    - Spec 4: 0.0598 (0.060)
  - Real GDP:
    - Spec 1: -0.1001 (0.231)
    - Spec 2: -0.1033 (0.220)
    - Spec 3: -0.1615 (0.154)
    - Spec 4: -0.1337 (0.160)
    - Spec 5: 0.0982 (0.213)
    - Spec 6: -0.0090 (0.164)
  - Real GDP growth:
    - Spec 1: -0.1812** (0.073)
    - Spec 2: -0.0507 (0.080)
    - Spec 3: -0.1075 (0.081)
    - Spec 4: -0.4032*** (0.127)
    - Spec 5: -0.5515*** (0.126)
    - Spec 6: -0.4139*** (0.087)
  - Inflation:
    - Spec 1: 0.1372** (0.069)
    - Spec 2: 0.1246 (0.082)
    - Spec 3: 0.1213** (0.059)
    - Spec 4: -0.0183 (0.036)
    - Spec 5: -0.0082 (0.031)
    - Spec 6: 0.0118 (0.028)
  - Debt:
    - Spec 1: -0.0099 (0.012)
    - Spec 2: -0.0050 (0.008)
    - Spec 3: -0.0077 (0.007)
    - Spec 4: 0.0565 (0.051)
    - Spec 5: 0.1383* (0.071)
    - Spec 6: 0.1066* (0.063)
  - Budget balance:
    - Spec 1: -0.1149 (0.115)
    - Spec 2: -0.2420 (0.181)
    - Spec 3: -0.1268 (0.117)
    - Spec 4: -0.2363 (0.181)
    - Spec 5: -0.0873 (0.139)
    - Spec 6: 0.0185 (0.129)
  - International reserves:
    - Spec 1: 0.0090 (0.082)
    - Spec 2: -0.0113 (0.099)
    - Spec 3: 0.0088 (0.046)
    - Specs 4–6: 0.0000 (0.000) (as listed)
  - Government effectiveness:
    - Spec 1: -2.4385 (0.804)
    - Spec 2: -1.7067 (1.787)
    - Spec 3: -1.9778 (1.051)
    - Spec 4: -1.9058 (1.068)
    - Spec 5: -2.6397 (1.884)
    - Spec 6: -2.3126 (1.214)
  - Bureaucratic quality:
    - Spec 1: -2.4316** (1.197)
    - Spec 2: -2.2399* (1.198)
    - Spec 3: -1.8838** (0.889)
    - Spec 4: -2.6051** (1.094)
    - Spec 5: -5.4923*** (1.661)
    - Spec 6: -5.4290*** (1.933)
- Sample and diagnostics:
  - Number of countries: 52 (Specs 1–2), 52 (Spec 3 listed), 44 (Specs 4–6)
  - Number of observations: 810 (Specs 1–3), 495 (Specs 4–6)
  - Country FE: Yes (all)
  - Year FE: Yes (all)
  - AR1 (p-value):
    - Specs 1–3: 0.002, 0.002, 0.002
    - Specs 4–6: 0.083, 0.024, 0.063
  - AR2 (p-value):
    - Specs 1–3: 0.399, 0.380, 0.442
    - Specs 4–6: 0.297, 0.394, 0.466
  - Hansen statistic (p-value):
    - Specs 1–3: 0.369, 0.416, 0.992
    - Specs 4–6: 0.755, 0.891, 0.804
- Notes:
  - Robust standard errors in brackets. A constant is included but not shown. The Hansen test is a test of overidentifying restrictions. AR1 and AR2 test for first and second autocorrelation, respectively. *** p<0.01, ** p<0.05, * p<0.1

*Appendix Tables A1–A4 as provided in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020079-print-pdf.pdf_