European Non-Performing Exposures (NPEs) and Climate-Related Risks: Country Dimensions

Abstract

The EU faces two economic challenges: managing non-performing exposures (NPEs) and climate change. This paper analyzes the relationship between the NPEs of domestic banking groups and climate risks, including macroeconomic variables such as the GDP growth rate, unemployment rate (UnEmp), and the voice and accountability percentile (VCA) and the interaction variable between the GHG and the Rule of Law Percentile (GhGRLP). The estimation uses ordinary least squares with time-fixed and individual effects. Physical and transition risks significantly affect NPEs, showing that both adverse climate events and the transition to a low-carbon economy worsen the financial situation of European banking institutions. The analysis also revealed that increased levels of VCA lead to a rise in NPEs, while an increase in GhGRLP reduces NPEs. In contrast, financial institutions tend to recognize and report NPEs more accurately in contexts with greater transparency and accountability. In comparison, UnEmp negatively affects NPEs, suggesting that economic support measures during high unemployment can reduce NPEs in the subsequent period. In conclusion, climate risk management represents a crucial challenge for the financial stability of banking institutions. Policymakers and financial institutions must continue to develop and implement climate change mitigation and adaptation strategies to preserve financial system stability amid growing climate uncertainties.

1. Introduction

In recent decades, the European Union (EU) has been confronted with various economic challenges, among which the management of non-performing exposures (NPEs) in the banking sector has been particularly prominent. However, the definition of an NPE is currently binding only for supervisory reporting, although institutions are strongly encouraged to use it for reporting within the internal risk control system and for financial disclosure to the public. Non-performing exposures should always be categorized as the whole exposure, including when non-performance relates to only a part of the exposure, such as unpaid interest. For off-balance sheet exposures, such as loan commitments or financial guarantees, the whole exposure is the entire uncancellable nominal amount (Committee on Banking Supervision 2017). The Bank for International Settlements (BIS) has defined non-performing exposures as exposures that are “defaulted” under the Basel regulations or are credit-impaired; that is, have experienced a devaluation due to deterioration of their creditworthiness. In addition, exposures that are past due more than 90 days or where it is evident that full repayment is unlikely without the bank’s realization of collateral are considered non-performing exposures. These deteriorated assets can result from various economic factors, such as recessions, real estate market crashes, or financial difficulties that borrowers face.

In addition to traditional economic factors, climate risks have emerged as a new and significant threat to the European financial system (Acharya et al. 2023; Committee on Banking Supervision 2021; Breitenstein et al. 2021; Brunetti et al. 2021; European Central Bank 2020, 2021b; Ranger et al. 2022; World Economic Forum 2022). The sixth Intergovernmental Panel on Climate Change report affirmed that climate change is global, fast, and escalating. The pace at which climate change has occurred is unprecedented. Human activities, particularly greenhouse gas (GHG) emissions, are causing an increase in the Earth’s average surface temperature, leading to greater climate variability and more frequent and intense extreme climate events. The European Green Deal represents a comprehensive set of strategic initiatives to guide the European Union toward a green transition, ultimately striving to achieve climate neutrality by 2050. This initiative underscores the necessity for a multidisciplinary approach, where various policy domains contribute to the overarching climate objective. The package encompasses initiatives in climate, environment, energy, transport, industry, agriculture, and sustainable finance, recognizing the strong interconnectedness among these sectors. Climate change and climate-related risks are concrete threats to economies and financial systems. Climate-related risks are usually categorized into physical and transition risks, each with unique features (European Central Bank 2020).

The first type is typically described as risks originating from the tangible consequences of climate change. These risks may be acute, arising from extreme weather events or severe environmental deterioration like floods, cyclones, storms, wildfires, and landslides. They can also be chronic and associated with gradual climate shifts such as rising temperatures and sea levels, water scarcity, loss of biodiversity, and resource depletion. Physical risks are often distributed unequally across different geographical areas, with some regions being more vulnerable than others. The second category of climate-related risks pertains to transition risks, which arise from adjusting toward a low-carbon and more sustainable economy. This transition involves policy and technological changes and shifts in consumer preferences toward environmental consciousness. Both types of climate-related risks have significant economic implications. Physical risks can damage properties, assets, and production capacity, and cause disruptions in supply chains. Transition risks, conversely, can lead to the devaluation of certain assets, reputational harm, and the obsolescence of certain industries or activities. The combination of NPEs and climate risks creates a complex challenge for the European banking sector. Extreme climate events can increase the likelihood of defaults among borrowers, especially those operating in climate-vulnerable sectors such as agriculture, energy, and tourism. All this can lead to a rise in NPEs on banks’ balance sheets. Climate policies to reduce carbon emissions can negatively impact high-carbon sectors, leading to further financial difficulties for the involved companies and, consequently, increasing the risk of non-performing exposures. Banks must, therefore, integrate climate risk assessments into their risk management (European Central Bank 2021b) processes to mitigate the combined impact of NPEs and climate change.

The scope of the paper is to analyze the relationship between the non-performing exposures of domestic banking groups and stand-alone banks in European countries with transition and physical risks. The novelty of this research is the consideration of NPEs, not only non-performing loans (NPLs); it also includes the contemporary country governance index with climate variables. NPEs include a broader range of credit exposures that can become problematic, not just loans. This can also include other assets such as bonds, guarantees, and off-balance sheet commitments. In addition to delinquent loans (as with NPLs), NPEs may consist of exposures that have been restructured or refinanced due to the borrower’s financial distress, even if they are not yet 90 days delinquent. NPEs cover a broader range of problem credit exposures, providing a more comprehensive view of the credit risk on a bank’s balance sheet. In addition, the analysis includes macroeconomic variables such as the GDP growth rate, unemployment rate, and the voice and accountability percentile, as will be illustrated in the third section. Moreover, the novelty of the analysis is the introduction of the interaction variable between greenhouse gas emissions and the rule of law percentile to verify whether a mandatory law on reducing emissions in the atmosphere can influence NPEs. The interaction variable may reflect a country’s legal and institutional system’s capacity to enforce environmental regulations, suggesting that NPEs are affected not only by GHG emissions per se but also by the capacity to manage and regulate such emissions. We investigate the influence of climate-related risks on European countries’ banking stability using a dataset of 13 countries between 2010 and 2023 with quarterly data. The results show a positive link between a higher level of transition and physical risk and a higher level of NPEs; moreover, effective governance can foster economic development and access to responsible credit. Countries with robust governance can promote policies that improve the business environment and reduce information asymmetries in the credit market, improving the overall quality of the loan portfolio and control over the NPE ratio.

The rest of the paper is organized as follows: Section 2 provides an overview of the recent literature; Section 3 describes the data, methodology, and results; and, finally, Section 4 presents the conclusions.

2. Literature Review

Recent papers on non-performing exposures (NPEs) have provided significant insights into this field. The guidelines from the EBA (KPMG 2019) serve as a key reference for defining and identifying NPEs. Farnè and Vouldis (2024) examined non-performing loans (NPLs) across various business models and found that retail-oriented European banks have lower levels of non-performing loans. Similarly, Pancotto et al. (2024) explored the evolution and determinants of NPLs in Italian banking, revealing that better-capitalized banks tend to exhibit lower bad loans. Zhang et al. (2024) investigated the impact of climate-related risks on NPLs in Chinese commercial banks, finding a positive correlation between higher levels of these risks and higher NPLs.

Transition risks

The literature on climate risks has also grown significantly. Chabot and Bertrand (2023) proposed a theoretical framework to examine the influence of climate-related risks on European financial intermediaries, highlighting the significant role of Scope 3 greenhouse gas emissions. Chronic and acute climate risks exacerbate financial vulnerabilities, affecting individual banks and the broader financial system. Di Febo and Angelini (2023) provided a preliminary analysis of the transition risk in the European financial system, emphasizing the importance of environmental factors in loans to NACE sectors. D’Orazio et al. (2022) focused on carbon neutrality in German banking, identifying energy, manufacturing, and transportation as the most problematic industries, with large private banks being the most exposed. Zhang (2021) examined the impact of environmental performance on collateral requirements and lending decisions in European companies, showing that environmentally friendly companies are more likely to obtain credit and face fewer collateral demands. Using a combination of Carbon Taxes and Green Supporting Factors, Dunz et al. (2021) examined the effects of carbon taxes and green support factors on greening the economy and the banking industry, examining how risk is transmitted from the credit market to the economy through loan contracts and how reinforcing feedback could result in cascading effects. Faiella and Lavecchia (2020) analyzed the carbon impact of Italian business bank loans, comparing methods to identify the most emission-intensive sectors and their indirect effect on the banking system.

Physical risks

Regarding physical risks, the European Central Bank (2021a) mapped physical and transition risks, stressing the need for granularity in assessing their impacts across sectors and firms. Battiston et al. (2021) provided a roadmap for evaluating the effect of extreme climate events on bank asset values. Oguntuase (2020) explored the relationship between climate-related, credit, and financial stability risks, demonstrating how these risks can increase credit risk and destabilize the financial system. Westcott et al. (2019) analyzed the differential exposure of investors and lenders to physical risks, noting that the physical effects of climate change can spread through various channels. The physical effects of climate change are direct and indirect, spreading through sectors, markets, and business value chains.

Our study fits into the existing literature with several distinctive contributions:

European Regional Analysis: while many studies are global or focused on individual countries, our examination of 13 European countries offers an important and comparative regional perspective.

Use of Innovative Proxies: the use of the NPEs instead of the NPLs gives a more comprehensive and detailed view of a bank’s overall credit risk. Using the SPEI index for physical risk and greenhouse gas (GHG) emissions for transition risk represents an innovative approach to quantify these risks and their relationship with NPEs. Governance and its impact on financial stability have been explored in studies such as those by Kaufmann et al. (2010). Using the Voice and Accountability Percentile as an explanatory variable for NPEs represents an innovation in the literature, offering a new angle on the relationship between governance and credit quality.

Integrating Different Variables: the combination of macroeconomic, financial, and governance variables in a single study allows for a holistic view of the determinants of NPEs.

Contemporary Empirical Approach: we use recent data to provide up-to-date and relevant analysis for policymakers and the financial sector.

In summary, our study enriches the existing literature by providing new empirical evidence on the relationship between NPEs and climate, economic, financial, and governance variables in a European context, thus contributing to understanding the factors influencing financial stability in the region.

3. Data and Methodology

The study covers the period between 2010 to 2023 with quarterly data, and several databases are used. The non-performing exposures (NPEs) and loans to calculate the growth rate are sourced from ECB Data. The Eurostat Database provides the climate data relative to transition and physical risks. The World Bank DataBank is the source of the macroeconomic variables.

3.1. Sample

Our analysis is explorative, with the scope of analyzing the impact of climate risks on the non-performing exposures of the banking system in European countries. Moreover, the government’s quality and the interaction variables affirm the crucial importance of single countries in the decarbonization process, which should bring the European Union to a neutral carbon state and the necessary consideration of these risks in the credit risk management process. There are 13 countries, as reported in

Table 1. The European countries in the database.

Country
Austria	AU
Belgium	BEL
Germany	DE
Spain	ESP
Finland	FIN
France	FRA
Greece	GRE
Ireland	IRL
Italy	ITA
Luxemburg	LUX
Netherland	NET
Portugal	PORT
Sweden	SWE

.

Table 2. The variables of the analysis.

Dependent Variable
NPEs	Gross forborne non-performing loans and advances [% of total gross non-performing loans and advances]
Independent Variables
PhyR	Physical Risk: Standardised Precipitation Evapotranspiration Index (SPEI)
TranR	Transition Risk: Logatrith natural of the Greenhouse Gas Emissions
LGR	Loans Growth Rate
GDP	Gross Domestic Product Growth Rate
UnEmp	Unemployment, total (% of total labor force)
VCA	Voice and Accountability Percentile
GhgRLP	Interaction Variable: Ln Greenhouse Gas Emissions × Rule of Law Percentile
Instrumental Variable
CR3	Banking concentration
Robustness Check
Covid	Dummy variable 1 = 2020, 2021, 2022

reports the variables used in the analysis.

In detail, the dependent variable is the ${NPEs}_{it}$. This represents the gross forborne non-performing loans (percentage of total gross non-performing loans and advances) where i is the country and t the time (Figure 1). It should be remembered that NPEs include the exposures that have become non-performing due to the application of forbearance measures; exposures that were non-performing before the extension of forbearance measures; forborne exposures that have been reclassified from the performing category, including exposures reclassified according to paragraph 260 of Part 2 of Annex V to Commission Implementing Regulation (EU) No 680/2014. While NPLs and NPEs are closely related, NPEs offer a more comprehensive view of a bank’s non-performing assets, including loans and other credit exposures. The broader scope of NPEs makes them a more useful measure for regulatory purposes and risk management.

The trend from 2010 to 2023 highlights the cyclical nature of NPEs, closely linked to broader economic conditions and the effectiveness of regulatory interventions. However, differences between countries remain evident, with nations such as Germany, France, and Sweden maintaining low NPE rates. In contrast, countries such as Italy, Greece, and Ireland have faced significant challenges in reducing their high NPE levels.

The climate-related variables are the ${TranR}_{it}$, which is the proxy of the transition risk and is represented by the logarithm natural of the greenhouse emissions of a single European country (Figure 2), and the ${PhyR}_{it},$ which is the proxy of physical risk, represented by the Standardised Precipitation Evapotranspiration Index (SPEI) (Figure 3), representing dry spells. The SPEI index measures the balance between precipitation and atmospheric water demand (evapotranspiration). Unlike other drought indices, the SPEI considers precipitation and temperature, making it more sensitive to climate change and temperature variations. Chronic physical risk refers to the long-term effects of climate change, such as changes in precipitation and temperature patterns that can lead to persistent extreme weather events such as drought. These risks can have significant impacts on sectors such as agriculture, energy, and water resources, as well as impact a region’s economic and financial stability.

The SPEI is widely used to monitor and quantify drought, one of the most relevant extreme weather events due to climate change. It may indicate prolonged periods of water deficiency that characterize chronic physical risk. An article in Nature (Vicente-Serrano et al. 2010) discusses how the SPEI can be used to monitor drought and its implications on a global scale, highlighting its ability to detect long-term climatic and hydrological variations. The Intergovernmental Panel on Climate Change (IPCC) has recognized the importance of indices such as the SPEI for monitoring and assessing climate risks, including chronic physical hazards such as prolonged drought.

The countries most exposed to transition risk are Germany, France, Italy, and Greece.

The SPEI considers precipitation and potential evapotranspiration, making it especially useful for understanding how much rain falls and how much moisture is lost through evaporation and plant transpiration (Figure 2). Positive values indicate wetter-than-average conditions, while negative values indicate drier-than-average conditions. Only Ireland and Sweden have an index between 1 and 1.49, presenting moderately humid conditions.

${LGR}_{it}$ represents the growth rate of loans for each country, ${GDP}_{it}$ indicates the economic growth rate, and ${UnEmp}_{it}$ is the percentage of the total unemployment for each country.

Regarding the control variables, we selected ${VCA}_{it}$. This represents a country’s governance dimension and stands for the Voice and Accountability percentile. It is a measure used in governance and economic studies to evaluate the extent to which a country’s citizens can participate in selecting their government, as well as the degree of freedom of expression, freedom of association, and free media.

“Voice and accountability” is a measure from the Worldwide Governance Indicators (WGIs) and reflects the extent to which a country’s citizens can participate in selecting their government, as well as freedom of expression, freedom of association, and free media. These aspects can significantly impact a bank’s risk assessment framework in terms of political stability and predictability, regulatory environment, market confidence, policy development and implementation, etc.

High scores in voice and accountability often correlate with political stability and predictability, reducing the political risk banks face. Political risk can impact the stability of a country’s financial system and the ability of borrowers to repay loans, influencing the bank’s credit risk assessments (Spornberger 2021). Governments with high levels of accountability and citizen participation tend to implement more transparent and predictable policies. This circumstance helps banks assess the regulatory environment and adjust their risk models accordingly (KPMG 2023). Democracies with high accountability generally have stronger and more reliable regulatory frameworks. This can reduce banks’ compliance and operational risk, as they can better anticipate and prepare for regulatory changes (European Central Bank 2021b). Enhanced voice and accountability often mean stronger legal systems and protections, which can help banks enforce contracts and recover assets, reducing credit risk.

Climate governance refers to the policies, regulations, and frameworks that govern actions to mitigate and adapt to climate change. The relationship between voice and accountability and climate governance is crucial: high voice and accountability facilitate greater public participation in developing and implementing climate policies. This ensures that climate governance measures are more inclusive equitable, and reflect societal needs, leading to more robust and widely accepted policies. Governments with strong accountability are more likely to implement transparent climate policies and be held accountable for their environmental commitments. This can enhance the credibility and effectiveness of climate governance.

Another fundamental variable is the interaction variable composed of the transition risk variable and the Rule of Law variable ${GhgRLP}_{i,t}$. The variable can be interpreted as an indicator of a country’s ability to implement and enforce greenhouse gas reduction regulations. A high rule of law implies that norms and regulations, including environmental ones, are effectively enforced. This may mean that firms must meet stricter environmental standards, potentially reducing greenhouse gas emissions and better managing environmental risks. The interaction variable may reflect a country’s legal and institutional system’s capacity to enforce environmental regulations, suggesting that NPEs are affected not only by GHG emissions per se but also by the capacity to manage and regulate such emissions. For example, (Klomp and De Haan 2014) examine how the quality of institutions (rule of law, control of corruption) moderates the effect of banking regulations on banking risk. This approach allows for a more comprehensive understanding of the factors influencing NPEs, considering both environmental and institutional aspects.

Finally, we consider a COVID-19 dummy variable. This assumes a value of one only for 2020, 2021, and 2022. The pandemic affected the whole world, increasing losses (Acharya and Steffen 2020; Baldwin and Weder 2020; Demirgüç-Kunt et al. 2021). Several papers suggest that ESG (environmental focus) reduces the impact of COVID-19 on bank performance (Yuen et al. 2022). So, we include the effect of COVID-19 in our baseline model.

The summary statistics are reported in

Table 3. The descriptive statistics.

Variable	Mean	Std. Dev.	Min	Max
NPEs	10.028	12.715	11.239	74.721
LGR	0.2097	2.3811	−11.32744	35.7450
TranR	10.4654	1.2105	7.55298	12.5360
GDP	0.382	0.780	−0.93	1.82
PhyR	−0.3670	1.0045	−2.473475	2.1355
VCA	90.2003	8.3219	68.07512	99.5305
UnEmp	9.1955	5.2797	3.1	27.686
CR3	71.8183	16.0219	33.17041	98.2282
GhgRLP	906.971	125.14	641.126	1167.903

.

3.2. The Model

The impact of climate risks on financial banking stability is analyzed using panel data. Kolstad and Moore (2020), a recent review of literature on statistical methods in climate and economics, highlighted that panel data is particularly useful for understanding how an entire system responds to climate change (Colacito et al. 2019; Burke et al. 2015; Dell et al. 2012; Deschênes and Greenstone 2007). Specifically, panel data enables the examination of conceptual elements over time by analyzing multiple years of observations from the same financial institutions and, within a given year, across the entire observed population. The model is the following:

$$\begin{gathered} {NPEs}_{it}={\beta }_0+{\beta }_1{PhyR}_{i,t-1}+{\beta }_2{TranR}_{i,t-1}+{\beta }_3{VCA}_{i,t-1}+{\beta }_4{LGR}_{i,t-1}+{\beta }_5{GDP}_{i,t-1}+{\beta }_6{UnEmp}_{i,t-1}+ \\ {\beta }_7{GhgRLP}_{i,t}+{\beta }_8{Covid}_t+{\theta }_i+{\delta }_t+{\epsilon }_{it} \end{gathered}$$

(1)

The estimation uses the ordinary least square with time fixed and individual effect.

Using lagged variables is justified by causal inference, dynamic relationships, and reduced endogeneity. Lagged variables help establish a temporal order, which is useful for causal inference. This setup implies that the independent variables at time $t-1$ potentially influence the dependent variable at time $t$. The lag captures the dynamic nature of the relationship, recognizing that the effects of explanatory variables might not be immediate and using them to help mitigate potential endogeneity issues, where current values of independent variables might be correlated with the error term. Using lagged variables to address the endogeneity problem is a well-documented and supported practice in the econometric literature. Articles that justify and discuss this approach include the following: Campbell and Shiller (1988) use lagged OLS regressions to examine the relationship between the dividend-price ratio and future dividend expectations and discount factors. Fama and French (1988) use lagged variables to explore the relationship between expected stock returns and past dividend returns. Mankiw et al. (1985) use lagged variables in an OLS framework to examine stock market volatility. Finally, Cornelli et al. (2023) also use lagged variables to study the new form of digital lending by fintech and big tech companies. All authors used the lagged variables to mitigate the endogeneity issues.

Country fixed effect ${\theta }_i$ is necessary for the control of unobserved heterogeneity across countries that might influence NPEs, such as institutional quality, legal frameworks, and cultural factors. The time-fixed effect ${\delta }_t$ serves to control for time-specific shocks or trends that affect all countries, such as regulatory changes.

We also tested the endogeneity problem through instrumental variables. We used the variable CR3 as the instrumental variable.

${CR3}_{it}$ is an economic indicator used to measure the degree of concentration of a market. It represents the combined market share of the three largest companies in a specific sector (Čihák et al. 2012). It is the assets of the three largest commercial banks as a share of total commercial banking. Total assets include total earning assets, cash and due from banks, foreclosed real estate, fixed assets, goodwill, other intangibles, current tax assets, deferred tax assets, discontinued operations, and other assets. A low CR3 implies that the market is more competitive, with numerous small and medium-sized actors sharing the market. Companies can be under pressure to innovate and improve efficiency in such markets. A high CR3 suggests that dominant companies could have significant advantages, such as economies of scale, which can create barriers to entry for new businesses. It can also indicate that dominant companies have substantial market power, influencing the prices and conditions of the market. In line with (Tran et al. 2022), we used CR3 as an instrument variable because the market concentration may increase bank loan growth. However, the p-value of the Wu–Hausman Test is greater than 0.1, so there is no endogeneity problem.

The regression results (

Table 4. Results of Model 1.

	(1)
VARIABLES	Robust
L.PhyR	0.711 **
	(0.332)
L.TranR	15.245 ***
	(5.777)
L.VCA	0.936 ***
	(0.195)
L.LGR	0.018
	(0.154)
L.GDP	0.505
	(0.535)
L.UnEmp	−0.646 **
	(0.276)
GhgRLP	−0.038 **
	(0.016)
Covid	1.306
	(3.269)
Constant	−163.104 ***
	(56.520)
R-squared	0.843
Quarterly FE	YES
Country FE	YES
Wu–Hausman Test	1.76
p-value	0.1850

Robust standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.1.

) show how the transition and physical risk have a positive and significant coefficient, indicating that higher levels of greenhouse gas emissions and a dry spell as climate events that worsen non-performing exposures (Zhang et al. 2022, 2024).

The results are in line with the literature. Lagged physical risks have a significant positive relationship with NPEs. A unit increase in risk in the previous period is associated with a 0.711 increase in NPEs. This shows that higher physical risk leads to higher levels of non-performing exposure. This finding is consistent with research indicating that climate-related physical risk can adversely affect financial stability by increasing default rates (Battiston et al. 2017; Dietz et al. 2016). The coefficient of the transition risk is significantly positive and large, indicating that higher transition risks (associated with moving towards a low-carbon economy) in the previous period significantly increase NPEs. This substantial effect suggests that changes in policies or market dynamics toward climate adaptation have a notable impact on financial stability. This result aligns with findings by Delis et al. (2024), who report that banks’ exposure to transition risks can lead to higher credit risk. The positive relationship between non-performing exposures (NPEs) and the Voice and Accountability (VCA) percentile could indeed be unexpected, given that better governance and accountability are generally thought to enhance financial stability.

Regarding regulatory and supervisory stringency, in countries with higher VCA, as demonstrated in Europe with an average value of 90.2 (Table 3), stricter regulatory frameworks and more stringent supervisory practices might compel financial institutions to recognize and report non-performing loans more promptly and accurately, leading to higher NPE ratios. In contrast, in countries with lower VCA, banks might delay recognizing NPEs due to weaker regulatory pressures. Countries with higher VCA often undergo significant economic transitions, including implementing reforms and moving to more transparent and accountable financial systems. During these transitions, financial institutions may experience short-term increases in NPEs due to adjustments in their operations and risk management practices.

The lagged unemployment rate is significant, with a negative relationship (UnEmp). The negative relationship between non-performing exposures (NPEs) and the lagged unemployment rate suggests that higher unemployment in the previous period is associated with lower levels of NPEs in the current period. Several factors and mechanisms explain this. During high unemployment, governments and central banks often implement countercyclical policies to stabilize the economy. These may include monetary easing, fiscal stimulus, and social welfare programs that provide financial support to individuals and businesses. The impact of these policies can lag, meaning that the support provided during periods of high unemployment helps to stabilize or even reduce NPEs in subsequent periods. Considering our variable NPEs, which also include exposures that have become non-performing due to the application of forbearance measures, the debt restructuring and forbearance factors’ are more important. Financial institutions may respond to rising unemployment by offering debt restructuring or forbearance programs to struggling borrowers. These measures can prevent loans from becoming non-performing in the short term, resulting in lower NPEs in the following period.

An interesting aspect is the result of the interaction variable. It is statistically significant with a negative relationship. This allows us to capture the combined effects of greenhouse gas emissions and the quality of governance (rule of law) on non-performing exposures. The variable is important because the impact of greenhouse gas emissions on NPEs can vary depending on a country’s ability to implement and enforce environmental regulations. A negative coefficient suggests that a high rule of law reduces the negative impact of GHG emissions on NPEs, supporting the hypothesis that better environmental governance mitigates environmental risks.

The high R-squared value and the inclusion of fixed effects strengthen the robustness of the model, providing valuable insights into the determinants of non-performing exposures in the European context.

3.3. Robustness Check

The previous regression shows how transition and physical climate risk worse the non-performing exposures of the European banking system. Robustness analysis is performed to verify that the results obtained from the main model are reliable and consistent. We divided Model 1 into two regressions: one that considered the physical risk and the other the transition risk (for details,

Table 5. Robustness Check, Model 2 and Model 3.

	(2)	(3)
VARIABLES	Robust	Robust
L.PhyR	0.612 *
	(0.330)
L.TranR		10.103 *
		(5.588)
L.VCA	0.920 **	1.021 ***
	(0.125)	(0.188)
L.LGR	0.051	−0.006
	(0.163)	(0.152)
L.GDP	0.603	0.294
	(0.544)	(0.499)
L.UnEmp	−0.608 **	−0.526 **
	(0.278)	(0.253)
GhgRLP	−0.031 *	−0.036 **
	(0.016)	(0.015)
Covid	2.572	0.508
	(3.157)	(3.454)
Constant	−20.195	−123.380 **
	(21.555)	(53.996)
R-squared	0.839	0.824
Quarterly FE	YES	YES
Country FE	YES	YES

Robust standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.1.

), thus assuming the risks in pureness.

$$\begin{array}{l}{NPEs}_{it}={\beta }_0+{\beta }_1{PhyR}_{i,t-1}+{\beta }_2{VCA}_{i,t-1}+{\beta }_3{LGR}_{i,t-1}+{\beta }_4{GDP}_{i,t-1}+{\beta }_5{UnEmp}_{i,t-1}+{\beta }_6{GhgRLP}_{i,t-1}\\ +{\beta }_7{Covid}_t+{\theta }_i+{\delta }_t+{\epsilon }_{it}\end{array}$$

(2)

$$\begin{array}{l}{NPEs}_{it}={\beta }_0+{\beta }_1{TranR}_{i,t-1}+{\beta }_2{VCA}_{i,t-1}+{\beta }_3{LGR}_{i,t-1}+{\beta }_4{GDP}_{i,t-1}+{\beta }_5{UnEmp}_{i,t-1}+{\beta }_6{GhgRLP}_{i,t-1}\\ +{\beta }_7{Covid}_t+{\theta }_i+{\delta }_t+{\epsilon }_{it}\end{array}$$

(3)

The results confirm the baseline model and show how various factors influence Non-Performing Exposures (NPEs) in the analyzed countries. In Model 2, Physical Risk has a positive and significant impact on NPEs, suggesting that an increase in Physical Risk is associated with an increase in NPEs. In parallel, in Model 3, Transition Risk emerges as a significant factor with a positive effect on NPEs, indicating that changes associated with the energy transition may increase NPEs. Another interesting result concerns the Voice and Accountability Percentile (VCA), which is significant and positive in both models. The finding suggests that greater transparency and accountability in institutions could bring more NPEs to light. Furthermore, the interaction between greenhouse gas emissions and the rule of law (GhgRLP) has a significant negative effect, indicating that better environmental governance may contribute to reducing NPEs. The Unemployment Rate shows a significant negative impact in both models, suggesting that banks may become more cautious in lending during periods of high unemployment.

Finally, the impact of COVID-19, represented by a dummy variable, is insignificant in both models, suggesting that the pandemic’s effect on NPEs was irrelevant in the analyzed data. These results highlight the importance of considering various factors, including physical and transition risks, institutional quality, and environmental governance when managing financial risks associated with NPEs.

4. Conclusions

The analysis of the impacts of climate risks on financial banking stability, conducted using panel data, provided significant results in line with existing literature. The first aspect to underline is the use of non-performing exposures (NPEs) compared to non-performing loans (NPLs), as this is complete and representative of the actual situation of the bank’s non-performing assets. While NPLs are primarily based on payment delays, NPEs also consider restructured or refinanced exposures. NPEs provide a more comprehensive and detailed view of a bank’s overall credit risk than NPLs. These distinctions are important for risk management and banking regulation, as they allow for a more precise assessment of the financial health and stability of the banking sector. The results show that physical and transition risks have a positive relationship with NPEs, highlighting that both adverse climate events and the transition to a low-carbon economy aggravate the financial situation of European banking institutions. This suggests that climate policies and market dynamics related to the ecological transition have a relevant impact on financial stability, consistent with previous studies.

The analysis also revealed that a high level of voice and accountability led to increased NPEs. In particular, less competitive markets with a few dominant companies may present greater risks of non-performing exposures, while in contexts with greater transparency and accountability, financial institutions tend to recognize and report NPEs more accurately. In contrast, the unemployment rate (UnEmp) negatively affects NPEs, suggesting that economic support measures implemented during periods of high unemployment can reduce NPEs in the subsequent period.

The interaction variable shows the importance of enforcing climate regulation, highlighting that strong governance and effective enforcement of environmental laws can reduce the credit risk associated with non-performing exposures. It is important for banks and financial institutions as it highlights the crucial role of environmental policies and robust institutions in credit risk management.

The high R-squared value and the inclusion of fixed effects strengthen the robustness of the model, providing valuable insights into the determinants of NPEs in the European context. Robustness analysis, which separates the two types of climate-related risks, confirms that the main results remain unchanged, underlining the reliability of the initial model. Despite the significant impact of the pandemic on the global economy, policy responses and economic resilience have helped mitigate the direct impact of COVID-19 on NPEs.

Future research developments could focus on several key aspects: geographic expansion, transition risk insights, and interaction with other system risks.

Expansion of the analysis to a global level, including data from different regions of the world, could help to verify the generalizability of the results and to explore any regional differences in the mechanisms of transmission of climate risks to NPEs.

A more detailed analysis of the elements that constitute transition risks, such as government policies, changes in energy prices, and the adoption of new technologies, should be carried out to better understand how these factors influence financial stability. Finally, the interaction between climate risks and other systemic risks, such as geopolitical or technological ones, should be explored to understand better how these factors can jointly influence the banking system’s stability.

In conclusion, climate risk management represents a crucial challenge for the financial stability of banking institutions. Policymakers and financial institutions must continue to develop and implement climate change mitigation and adaptation strategies to preserve the economic system’s stability in the context of growing climate uncertainties. These results can guide both banking policies and government regulations towards more sustainable and responsible practices, improving the global financial system’s stability.