Climate-related risks are increasingly recognised as an important threat to long-term fiscal sustainability. This column studies the extent to which credit rating agencies integrate these risks into their sovereign rating assessments. It finds that measures of physical risk lead to lower credit ratings, but the impacts are small in magnitude. Ambitious CO2 reduction targets and actual emission reductions have been reflected in higher ratings, but only since the 2015 Paris Agreement. These findings highlight the need for caution in using credit ratings for regulatory and macroeconomic policy, as they seem to only partially account for environmental considerations.
Climate change can exert significant pressure on countries’ fiscal positions. This occurs through well-documented channels, including rising costs associated with more frequent and severe natural disasters, essential investment in adaptation, and financing the transition to a green economy (Mallucci 2022, Klusak et al. 2023, Volz et al. 2020, Zenios 2022). Sovereign credit ratings aim to measure a country’s creditworthiness and should take these pressures into account. But do they?
Credit rating agencies (CRAs) have recognised that climate change can significantly affect sovereign ratings. For example, pressed by central banks and regulators, 1 credit rating agencies have begun disclosing how they incorporate climate risks into their rating methodologies. 2 Major agencies have also acquired stakes in firms specialising in climate risk data and analytics, as they strive to build capacity and expertise in evaluating climate-related financial risks. 3 Moreover, historical data show that natural disasters often lead to rating downgrades, especially for shock-prone, low-income countries. 4
At the same time, several obstacles hinder the full and systematic integration of climate risks into credit rating agencies’ rating frameworks. These include data limitations, the high uncertainty about the economic impact of climate change, doubts about governments’ commitments to net-zero emissions targets, and the relatively short time horizon of credit ratings compared with the long-term nature of climate change (Kraemer 2021).
Thus, the extent to which sovereign ratings incorporate climate risks is an empirical question. In our study (Cappiello et al. 2025), we examine whether credit rating agencies include physical and transition climate risks in their models and also whether they have assigned greater weight to climate-related factors since the Paris Agreement was adopted in 2015. Our analysis draws on data covering an extended period and a large sample of countries, including advanced, emerging and low-income economies (Figure 1).
Figure 1 Countries included in the analysis
Source: Cappiello et al. (2025).
Note: The dataset covers 124 countries in total, of which 33 are advanced economies, 62 are emerging market economies, and 29 are low-income countries.
We focus on the sovereign ratings issued by the four major credit rating agencies: S&P Global Ratings, Moody’s, FitchRatings, and Morningstar DBRS. Because rating disagreement among these agencies is limited, with approximately 90% of ratings differing by no more than one notch since 1991, we use the average rating across all these agencies as our dependent variable.
Our findings suggest that credit rating agencies take physical risk exposure into account when assessing the probability of sovereign default, which is consistent with anecdotal evidence that natural disasters can lead to downgrades, particularly for low-income countries. Conversely, countries that are more resilient to extreme weather events – typically, advanced economies – tend to receive higher ratings.
However, transition risk factors, such as carbon emissions, primary energy consumption, and CO2 reduction targets, are not reflected in ratings. Moreover, even when climate variables are statistically significant, they still only have a marginal impact on credit ratings (Figure 2).
Figure 2 Economic impact of temperature anomalies on sovereign credit ratings (credit rating notches)
Source: Cappiello et al. (2025).
Notes: The figure shows the absolute value of the effect of a (1% winsorised) one standard deviation shock of macroeconomic and climate risk variables on sovereign credit ratings for advanced economies (AEs) and emerging market economies (EMEs). The impact elasticities are derived from a panel regression covering 124 countries over the period 1999-2021. The dependent variable is the average of the sovereign ratings issued by S&P Global Ratings, Moody’s, FitchRatings, and Morningstar DBRS.
Have credit rating agencies changed their assessment of climate risks since the Paris Agreement?
The Paris Agreement marked a significant turning point in public awareness of the consequences of climate change and the urgent need for policy action. This heightened awareness has resulted in more stringent climate policies worldwide that increasingly warrant their inclusion in credit ratings. In addition, after the Paris Agreement was adopted in 2015, all major credit rating agencies signed the UN Principles for Responsible Investment in May 2016, committing to systematically evaluating the relevance of environmental factors in credit assessments and to review how these factors are integrated into credit analysis. Thus, it is reasonable to conjecture that since these events, credit rating agencies have reassessed climate change risks and incorporated transition costs into their models for evaluating sovereign creditworthiness. To test this hypothesis, we define a natural experiment using the Paris Agreement as an exogenous event that may have shifted credit rating agencies’ assessments of climate-related risks.
To determine whether major credit rating agencies have updated their models to reflect this commitment to including environmental factors, we rank countries’ exposures to climate risks using the Climate Change Risk Country Scoring Model developed by Ferrazzi et al. (2021). We categorise countries into two groups – countries with high exposure to climate risks (the treatment group) and those with low exposure (the control group) – using the median score as a divider. We apply a difference-in-differences methodology for estimation.
Our findings show that since the Paris Agreement, credit rating agencies have assigned lower ratings to countries with higher exposure to physical risk relative to the control group (Figure 3, panel a). This suggests that credit rating agencies recognise that increasingly frequent natural disasters can have a significant impact on sovereign balance sheets, particularly in low-income countries, and that these risks should be adequately reflected in credit risk models. Additionally, we observe a shift in how CRAs evaluate transition risk, with higher ratings awarded to countries that commit to more ambitious CO2 emission reduction targets and achieve lower CO2 emission intensity post-Paris Agreement. This indicates that CRAs have started to ‘reward’ countries, including smaller ones, that diversify away from reliance on fossil fuels and adopt cleaner energy sources.
The role of stranded assets, transition-critical materials, and fiscal capacity
Finally, we investigate three country-specific factors that may amplify or mitigate the effects of climate risk exposures: reliance on fossil fuel revenues, high sovereign debt levels, and reserves of commodities crucial for the green transition. We find that these factors significantly influence sovereign ratings (Figure 3, panel b). Sovereigns more dependent on fossil fuel revenues and exposed to both physical and transition risks have tended to receive lower ratings since the Paris Agreement, likely owing to the ‘stranding’ of assets, i.e. fossil fuel reserves potentially losing value before the end of their expected economic life owing to decarbonisation efforts. Additionally, countries with high sovereign debt generally receive lower credit ratings post-2015, indicating that high debt levels amplify climate risk exposures. The significant coefficients suggest that constrained fiscal capacity limits a country’s options for mitigating the impacts of rising physical risk and for funding the green transition. Conversely, countries that are major exporters of transition-critical materials (TCMs) – such as copper, graphite, nickel, manganese, lithium, cobalt, and rare earths – tend to receive higher ratings despite climate risks.
Figure 3 Estimated effects of climate risks on sovereign credit ratings post-Paris Agreement
a) Post-Paris Agreement
b) Oil exporters, exporters of TCM, and high-debt countries
Source: Cappiello et al. (2025).
Notes: Panel a) shows coefficient estimates from six difference-in-differences models for physical risk (PR) and transition risk (TR). ‘Exposure to physical risk’ is interacted with a post-Paris Agreement (PPA) dummy variable, set to one after 2015. ‘Temperature anomalies’ and ‘Frequency of natural disasters’ are interacted with the PR and PPA dummy variables. ‘Emission intensity’ is interacted with the TR and PPA dummy variables. The ‘CO2 reduction target’ interacts with the TR dummy variable, while the ‘Primary energy consumption-to-GDP ratio’ interacts with the TR and PPA dummy variables. Panel b) shows coefficient estimates from five difference-in-differences models, augmented with information on fossil fuel reliance, transition-critical material (TCM) exports, and indebtedness levels. The first two coefficients represent triple interactions among climate risk exposure dummy variables (PR and TR), the PPA dummy variable, and fossil fuel reliance, defined as a time-invariant dummy variable equal to one if the country is among the top 20 fossil fuel exporters. The third coefficient represents the triple interaction among an exposure dummy variable (to PR or TR), the PPA dummy variable, and TCM exports as a percentage of the country’s total exports. The fourth and fifth coefficients represent double interactions between the PPA dummy variable and a high-debt dummy variable, set to one if the country’s debt-to-GDP ratio in 2015 exceeds the median for advanced economies and emerging market economies prior to 2015. The sample period for all estimates is from 1999 to 2021.
Conclusions
Our results provide insights for market participants and policymakers, as credit ratings are widely used to assess the default probability of sovereign debt and are integral to various economic and regulatory policies. If sovereign ratings do not systematically reflect climate change risks, there is a risk of future asset repricing, which could transmit to different parts of the financial system, potentially affecting banks, insurers, and other financial institutions that hold sovereign bonds. This could lead to financial instability if climate-related shocks cause sudden rating downgrades and asset value losses.
Moreover, underestimation of climate risks in credit ratings could mislead market participants into taking on risks which they are not fully aware of, and which are therefore not adequately reflected in risk premia. Conversely, sovereigns that implement effective adaptation and mitigation measures may not receive adequate rewards in the form of lower borrowing costs. When using sovereign debt as collateral in monetary policy operations, central banks may hold assets that are more vulnerable to climate-related shocks than they think, if the credit ratings used do not reflect climate risks. Therefore, our findings highlight the need for caution when using credit ratings for regulatory and macroeconomic policy, as these ratings appear to underestimate environmental factors.
Source : VOXeu
