The Relationship between ESG Scores and Firm-Specific Risk of Eurozone Banks

Abstract

This paper investigates the relationship between corporate social responsibility and the idiosyncratic risk of Eurozone banks. Idiosyncratic risk represents firm-specific risks for banks, and the Carhart four-factor model is used for 31 Eurozone banks from 2002 to 2019 to determine the idiosyncratic risk. Thomson Reuters ESG scores are used to determine the ESG scores of these banks during the same period, and the effects of the environmental, social, and governance dimensions are investigated separately. The quantile regression method reveals a relationship between ESG and idiosyncratic risk over different risk levels. A significant negative relationship has been found between the overall ESG scores and the idiosyncratic risk of banks for medium- to high-risk levels. The effect becomes stronger as the riskiness of the banks increases. Similar to the overall ESG score, the governance and environmental dimensions have a negative impact on banks with medium- to high-risk levels. No significant relationship could be identified between the social dimension and the idiosyncratic risk of banks.

1. Introduction

History repeated itself and the need for safe and sound banks for a functioning economy was declared during the Global Financial Crisis (GFC) of 2007–2009. Following the GFC, banks and their stakeholders, including stockholders, depositors, borrowers, and regulators, started emphasizing the importance of sustainability criteria. Environmental, social, and governance (ESG) principles are at the center of sustainable banking’s context. Some international institutions (the European Banking Authority, Bank of England, European Commission) have developed policies and highlighted the role of sustainability, which is operationalized through ESG strategies. Due to the increasing importance of the ESG risk concerns through environmental, management, and governance factors, regulatory authorities, central banks, political institutions, and international financial institutions have developed and integrated ESG-based tools into the banks’ risk management [1,2,3,4]. Bank stability is one of the essential prerequisites of efficient and healthy functioning financial systems. As banking crises present the failure of these institutions creates a damaging effect on the welfare of the countries, as they are the leading financial intermediaries transferring funds among the savers and borrowers. The banking literature documents their role in economic development well [5,6]. Moreover, bank failures create costs for the taxpayers and cause moral hazard problems among the public (mainly the depositors) and the risk-averse banks. As such, it exacerbates intermediation problems by discouraging savings behavior and encouraging risk-taking by banks. From this aspect, the banks’ safety/soundness/risk is equivalent to the sustainability concept and is correlated with diverse literature on the corporate social responsibility (CSR) and ESG [7,8,9,10,11]. Due to their distinct financial structure (in which the third parties, mainly depositors, provide 80/90% of the funds) and opaqueness in financial transactions, the ESG’s role in the banking firm increases. This study attempts to analyze and diagnose the role of ESG in bank risk management for the leading Eurozone countries’ banks by operationalizing the ESG scores and various bank risk measures from 2002 to 2019.

Though the existing literature documents many studies on environmental, social, and governance principles, it is still unclear why banks focus extensively on ESG engagement and invest vast amounts. Traditionally, banks are not accepted as ESG-related companies compared to industries that directly impact the environment, such as construction, mining, chemical, and petroleum. However, KPMG’s research showed that banks disclose ESG information within their annual reports more than any other sector [12]. Furthermore, recent studies argue that environmental and social aspects could be a valuable catalyst for financial institutions to regain the reputation and trust of stakeholders lost during the 2007–2009 GFC [13]. This attribute makes ESG a critical communication tool for banks to minimize information asymmetries with their stakeholders.

Extensive research has been conducted to understand the effect of ESG on the short-term financial performance of companies. Nevertheless, contradictory findings could not reach a definitive conclusion about the impact of the ESG performance on the financial performance of firms assessed by accounting or market-based measures. A meta-analysis conducted on 85 studies over 20 years concluded that socially responsible investment with ethical concerns regarding stock market portfolio management does not create any advantages or disadvantages according to the stock market’s performance [14]. Another literature analysis of over 167 studies investigating accounting performance measures found a positive but considerably small relationship between corporate social performance and financial performance [15]. Therefore, further research is needed to understand the motives behind costly ESG projects financed by the companies. The majority of the studies perceived ESG to be a financial investment and tried to reveal the short-term financial effects for the companies. However, ESG is an essential component of stakeholder engagement which improves the reputation of banks and is used as a risk management tool [16]. ESG assures stakeholder satisfaction and minimizes the firm-specific risk rather than the short-term profit maximization. Improving relationships with stakeholders over ESG reduces the risk of conflicts with major stakeholders. Previous literature has argued that the socially responsible behavior of firms could be seen as a strategic tool and perceived as a “win-win” scenario that strengthens companies’ market position and reputation while considering the interest of society [17]. Concerning banks, ESG not only protects banks from having a negative stakeholder crisis, but it may also stimulate a positive impact. ESG is also an essential communication tool with stakeholders because it legitimates and promotes socially responsible actions. Therefore, it positively contributes to the bank’s reputation and the trust of stakeholders. An increased reputation and the trust of stakeholders directly affect the firm-specific/idiosyncratic risks for banks [18].

Our paper contributes to the banking and CSR/ESG literature from two unique aspects. First, even though the relationship between ESG and firm-specific risks has been investigated extensively by the previous literature, only a few research studies focus on banks [9,10,11,19,20]. Nonetheless, these research studies mostly used accounting-based risk measures as a proxy for firm-specific risks. However, we argue that market-based risk measures are more relevant, as they show the effect of ESG scores on the idiosyncratic risk through stakeholders’ behavior. Therefore, the first contribution of this paper is the use of a market-based risk measure that is not used in the previous bank CSR/ESG studies, namely the idiosyncratic risk volatility. Secondly, in contrast to previous studies, which have used conditional mean methodologies (OLS) and ignored the effect of ESG on different risk levels/groups, we employ a quantile regression. Notably, we use a newly developed approach that is useful for panel data with individual effects and endogenous explanatory variables, as in our data [21]. We think this methodological approach will provide more substantive findings, as the effect of ESG can be different for various risk levels.

Our results suggest that improvements in the ESG scores lower the bank risk significantly for medium- to high-risk category banks. They also show that the ESG’s contribution to bank stability, and therefore sustainability, increases with the bank’s riskiness. Concerning the ESG pillars’ (environmental, social, and governance), results showed that the governance and environmental dimensions have the highest negative impact on the idiosyncratic risk for medium- to high-risk level banks. Similar to the overall ESG results, as the riskiness of a bank increases, the negative effect increases for the governance and environmental dimensions. No significant relationship could be identified between social dimension and firm-specific risk. The rest of this paper is organized as follows: The Section 2 reviews the relevant literature on the theoretical perspective of ESG and idiosyncratic risk volatility and develops hypotheses. The Section 3 explains the methodology, model, and data used for the study. Finally, the Section 4 presents the findings, and the study concludes with the Section 5.

2. Literature Review and Hypothesis Development

2.1. Theoretical Perspective of ESG

Researchers can use various theories to establish the relationship between the ESG performance and the bank value/risk. For example, “legitimacy theory” states that companies’ activities should be in parallel with society’s beliefs, norms, values, and expectations [22]. Moreover, legitimation strategies should be implemented by companies to avoid legitimacy crises, such as serious accidents, pollution leaks, or financial scandals [23]. As such, social capital could be an essential tool to legitimate the actions and profits of the companies [24]. Another theory that links ESG strategies to the companies’ performance is the “Stakeholder theory” [25,26]. This theory argues that companies should consider the interest of all stakeholders, rather than stockholders, since this strategy contributes to long-term value maximization. The stakeholder group includes shareholders, employees, consumers, public organizations, and the government, representing all social groups within the community who have a direct or indirect relation with the company [26]. According to the stakeholder theory, considering the interest of all stakeholders ensures long-term value gain for the company [25]. It has been argued that ESG should be viewed with a multi-theoretical perspective, as ESG is a complex phenomenon and cannot be explained by a single theory. Actually, the “legitimacy” and “stakeholder” theories are interrelated by acting complementarily, rather than competing with each other, as the legitimacy of companies could be ensured by considering the interests of all stakeholders [27]. Socially responsible behavior ensures that companies’ actions align with society’s expectations. Moreover, ESG contributes to the reputations of companies by creating a moral capital that generates a flow of resources in many forms, such as financial, human, and technological [28]. Increasing interest in the concept of banks has perceived ESG as a tool to increase reputation, trust, and credibility [29]. Since banks have many stakeholder groups, such as depositors, borrowers, stockholders, and the government/public, and are also among the most heavily regulated firms, these theories provide the theoretical basis for ESG studies in banking.

2.2. Idiosyncratic Bank Risk and ESG

This study uses idiosyncratic volatility, which represents the gap between a market portfolio and individual stock fluctuations, to measure the idiosyncratic bank risk. A company’s stock volatility is determined by the systematic risk and the unsystematic/idiosyncratic/firm-specific risk. The systematic risk depends on the market portfolio, while the idiosyncratic risk represents the portion of the market portfolio that cannot be explained by the firm’s actions. Numerous studies have found that the idiosyncratic risk of the firms represents the majority of the total stock price variance compared to the systematic risk [30,31]. Idiosyncratic risk mainly depends on firm-specific factors. Nevertheless, it is argued that idiosyncratic volatility is not important, as diversification in efficient markets can eliminate this. However, it is evident that markets are not perfectly efficient due to transaction costs, agency problems, and informational problems (asymmetric information). Therefore, market inefficiencies increase the importance of idiosyncratic risk [32].

An analysis of the ESG scores of firms showed that positive ESG reduces idiosyncratic risks, while negative ESG has a risk-increasing effect [33]. Previous research on controversial industries, including alcohol, tobacco, gambling, and others, found that CSR, intended as ESG scores, is not a window-dressing activity, as it significantly reduces the idiosyncratic risk [34]. An analysis on the idiosyncratic risk reduction effect of CSR on different market states concluded that CSR is a significant risk management tool both in up- and down-trending market states [35]. Similarly, another research study show that the corporate social performance has a negative relationship with the idiosyncratic volatility for firms with better communication with their stakeholders [18]. An increased attention of the stakeholders in the ESG performances of banks is evident, and the adoption of ESG practices has been shown to secure the reputation of banks by minimizing the possibility of sanctions [36]. ESG could be a risk-mitigation tool, especially during periods of financial distress by signaling prudent banking activities, enhancing reputation, and ensuring good relations with the community [29]. These findings suggest that ESG should be considered an effective risk-reducing tool, as it minimizes the idiosyncratic risk through communication with the stakeholders.

Concerning the banking industry, previous research has asserted the significance of the idiosyncratic risk for banks [37,38,39,40]. On the other hand, the risk-related literature generally undermines the idiosyncratic risk, as it can be eliminated by diversification. Nonetheless, the failure of one bank can affect the whole banking industry through the contagion effect [40]. Moreover, deposits, insurance, and too-big-to-fail guarantee schemes encourage banks to increase risk and underestimate risk diversification in a lax regulatory environment. Therefore, monitoring the idiosyncratic risk is more critical for banks than other firms. Previous literature has stated that the idiosyncratic risk of banks is related to the business model, risk culture, and bank-specific factors [41]. Furthermore, the above studies generally report a negative relationship between a bank’s size and its idiosyncratic risk, as a larger size allows banks to better diversify. Therefore, we think the diversifiable character of the idiosyncratic risk makes it more critical for the banks than the ESG concern, since banks can follow the diversification process independently and use ESG principles (dimensions) as a risk management tool.

This study will be the first study to analyze the idiosyncratic risk of banks and ESG policies. The significance of the ESG for banks is related to their business structure as it captures multiple groups of stakeholders. The first two stakeholder groups, depositors and borrowers, are the products of the financial intermediary role of the banks. The third one is the regulators. Due to their policy role, deposit insurance, too-big-to-fail guarantee schemes, and the liability structure, banks are closely regulated by different regulatory authorities. They also offer investment products to investors who represent the fourth group. The fifth one is the shareholders, who are the owners. As such, the banks’ ESG policies should directly affect the groups mentioned above through their ESG dimensions. Nonetheless, these would have indirect implications for the other public groups as well. For example, taxpayers who do not have a direct relationship with a bank can be affected by a bank’s failure, or a villager can be negatively affected by a bank-given loan that destroys the environment. As such, we believe that ESG policies should be a significant concern of the banks. The report published by the Canadian Credit Union Association asserts that the senior managers of eight credit unions perceived socially responsible behavior as an important risk management tool for their institutions [42].

The role of the ESG arising from the idiosyncratic risk concern is vital for the bank’s stakeholders described above. As the literature asserts, bank stakeholders, such as depositors, borrowers, investors, regulators, and managers, are directly affected by the idiosyncratic risk; therefore, they care about it [37,38,39,40]. As beneficiary groups, depositors, borrowers, and investors care about the idiosyncratic risk for sustainable banking services. In addition, by nature, shareholders are the owners; hence, the idiosyncratic risk is crucial for them as it affects profitability and the share price. For regulators, the safety and stability of the banks in the banking system make the idiosyncratic risk a significant risk concern. These imply that banks need to inform stakeholders about their actions more than other sectors [43]. As such, ESG could impact key firm-specific risks for banks.

In the light of the legitimacy and stakeholder theories, and the above arguments, we conclude that ESG is necessary to have a stable relationship with stakeholders and protect companies from random shocks from the idiosyncratic risk sources. ESG can help banks minimize their key firm-specific risks related to stakeholders. Increased communication with key stakeholders through the promotion of socially responsible actions minimizes the idiosyncratic risks. Nonetheless, ESG is not considered a risk factor by traditional risk models, such as CAPM or Fama–French, and is inadequately included within firm-specific risk [44]. Therefore, we decided to carry out this research and contribute to the ESG and banking literature. Accordingly, as stated in Hypothesis 1 (H1), this research predicts a negative relationship between CSR and idiosyncratic volatility by minimizing the idiosyncratic risks of banks.

Hypothesis 1 (H1).

ESG and idiosyncratic risk have a negative relationship.

2.3. ESG Dimensions and Idiosyncratic Risk

Though the general ESG measure provides guidance, identifying the effect of specific ESG dimensions on the idiosyncratic risk is more important. Generally, this is done by adapting the banks’ ESG (environmental, social, and governance) scores to the research models. Following this idea, we also use ESG scores separately in our analyses. Therefore, our study will reveal the effectiveness of these dimensions and create better guidance for banks. Previous researchers found that environmental, social, and governance dimensions could interact differently with firm-specific risks since stakeholders are not homogenous and are affected differently by ESG dimensions [8,11,45].

Environmental responsibility leads to energy- and resource-saving as it aims to minimize the carbon footprint of banks. Environmental responsibility could increase the operational efficiency of banks as energy and resource consumption is monitored. Additionally, increasing stakeholders’ awareness of environmental manners creates a risk reduction effect for environmentally responsible companies. Promoting environmentally friendly actions establishes a communication channel with stakeholders that minimizes the information asymmetries [46]. Moreover, environmental disasters potentially affect bank risks, such as operational, liquidity, and credit risks. As such, some regulatory bodies and institutions warn banks to care about the environmental risk [2,3,4,47]. Due to the increased awareness of community, the European Central Bank announced that recent bank stress tests have included environmental risks. The stakeholder theory argues that the actions of institutions must align with the expectations of the whole society. Research on banks that has solely focused on the environmental dimension found an inverse relationship between the environmental performance and firm-specific risk [9,10,11]. It has been argued that the main reasoning behind this is that environmental engagement enhances the reputation of banks and legitimizes the banks’ actions by improving their social images [19]. Enhanced reputation and protection from adverse consequences legitimize the actions of banks and increase the loyalty of stakeholders. As such, H2 predicts a negative relationship between the environmental dimension and idiosyncratic risk by satisfying the concerns of environmentally friendly stakeholders.

Hypothesis 2 (H2).

The environmental dimension of ESG and idiosyncratic risk have a negative relationship.

The social aspect of ESG has a direct impact on reputation, and banks can use it as a communication tool with various stakeholders. The social element assures product responsibility, positive community commitment, and good employee relations. Socially responsible activities could minimize the idiosyncratic risk for banks by considering the interests of various stakeholders. Higher social performance signals an improved social capital and increase the reputation for stakeholders [45]. Reputation is crucial for healthy functioning banks due to trust relationships with the depositors, investors, and borrowers. Employee strikes, boycotts, and lawsuits could damage the reputation of banks within the community. Therefore, employee relations form an essential part of the social dimension, and previous research has concluded that good employment practices and policies minimize firm-specific risks [48]. A previous study on international companies found that the social dimension has a risk-reducing effect on the financial risk of companies [8]. In line with this finding, another research study found a risk-reducing effect of the social dimension for the banks [9]. Nevertheless, some research found an ambiguous influence of the social dimension [11]. Positive social performance will legitimize the actions of banks and increase stakeholders’ trust and loyalty, which contributes to stakeholder relationship management. As such, H3 predicts a negative relationship between the social dimension and idiosyncratic risks.

Hypothesis 3 (H3).

The social dimension of ESG and idiosyncratic risk have a negative relationship.

Governance is another dimension of ESG that is related to idiosyncratic risk. Governance ensures effective management and accurate decision-making and affects banks from multidimensional factors. Therefore, good governance directly affects the above-mentioned stakeholder groups. The stakeholders who transact with the bank need to be assured that the institution is governed properly. Previous research has shown that the ownership structure of banks affects the riskiness of banks; a high ownership concentration increases the incentive for risk-taking, while non-shareholding managers tend to decrease it [49]. Good governance ensures risk management and increases the trust that stakeholders feel towards the bank. As the stakeholder theory predicts, promoting the governance quality of the bank creates better communication channels with stakeholders and minimizes the information asymmetries. Therefore, good governance positively affects banks’ reputations, contributes to bank–stakeholder relationships, and reduces idiosyncratic risk. In line with the above views, it is been argued that the governance dimension could have a stronger negative relationship with firm risks, as they are more relevant and visible to the investors compared to other dimensions [8]. Concerning banking, previous research has found a risk-reducing effect of corporate governance for banks in common law countries [44] and European countries [9]. On the other hand, there are some mixed results for a sample of worldwide banks [11]. As we have asserted many positive implications of governance for idiosyncratic risk, we expect a negative relationship between the governance dimension and idiosyncratic risk in Hypothesis 4.

Hypothesis 4 (H4).

The governance dimension of ESG and idiosyncratic risk have a negative relationship.

3. Data and Methodology

3.1. Data

The ESG data is obtained from the Thomson Reuters Eikon ESG Database. Thomson Reuters is a leading agency that provides financial data and is used intensively by investors. The database uses algorithmic and human processes together with over 400 ESG metrics while determining the score of companies. In addition, the database includes negative media stories, which are captured as ESG controversies and deducted from the overall ESG scores. This eliminates the bias of relying solely on company-provided sources, which is the method of some previous studies that used content analysis to determine the ESG scores of companies. The database uses separate performance indicators and provides scores for environmental, social, and governance pillars. The main categories of environmental pillars are resource use, emissions, and innovations. The main categories of social pillars are workforce, human rights, community, and product responsibility. The governance categories include management, shareholders, and CSR strategy. Unfortunately, the ESG scores are the limiting data for the research, and it is available from 2002 to 2019. The sample includes 31 Eurozone banks from the leading Eurozone countries from 2002 to 2019, and country distribution is illustrated in

Table 1. Distribution of banks by country.

Country	Number	Percentage
Austria	2	6.45
Belgium	1	3.23
France	4	12.90
Germany	1	3.23
Greece	3	9.68
Ireland	3	9.68
Italy	9	29.03
The Netherlands	1	3.23
Portugal	1	3.23
Spain	6	19.35
Total	31	100

below. In this study, we concentrate on Eurozone countries that share similar economic and regulatory environments, as well as common monetary policy under the regulatory council of the European Central Bank. Any bank with missing data for a particular year has been removed from the analysis, and the final sample size is 471 firm-year observations.

Financial and accounting data for banks were obtained from the Thomson Reuters Eikon database. The data includes daily stock prices, dividend yield, provision for loan loss, operating profit margin, total loans, return on equity (ROE), capital adequacy, liquidity, and market-to-book ratio. The Institutional Brokers’ Estimate System (I/B/E/S) estimate is used to determine 12-month forward earnings per share. The country-specific variable, inflation, is obtained from the World Bank. The detailed variable descriptions are presented in Appendix A 

Table A1. Variable descriptions.

Variable	Description	Source
ESG	Weighted average of environment, social, and governance score. Represents overall CSR score.	Eikon
Environment	Environment score measures overall environment performance of banks.	Eikon
Social	Social score measures overall social performance of banks.	Eikon
Governance	Corporate governance measures overall governance performance of banks.	Eikon
Dividend yield	Dividend yield is the percentage of dividend paid compared to stock price	Eikon
Provision for Loan Loss	Shows forecast of future loan losses	Eikon
Operating Profit Margin	Operating profit margin shows efficiency of banks by dividing operating income by net sales	Eikon
LN(Total Loans)	Logarithmic transformation of total loans representing size of banks	Eikon
Return on Equity	Return on equity is profitability ratio showing net income over equity capital	Eikon
Inflation	Yearly inflation value for the relevant country	World Bank
IBES 12-Month Forward EPS	Institutional Brokers’ Estimate System forecast for 12-month forward earnings per share of banks	Eikon
Capital Adequacy	Capital adequacy ratio shows percentage of capital to risk-weighted assets	Eikon
Liquidity	Liquidity represents ratio of banks’ liquid assets to obligations of banks	Eikon
Market-to-Book Ratio	Market-to-book ratio represents market value of banks over book value	Eikon

. Summary statistics for the collected data are presented in

Table 2. Descriptive Statistics.

Variable	Mean	Median	Maximum	Minimum	Std. Dev.
Dependent Variable
Idiosyncratic Volatility	0.533878	0.363764	4.175236	−1.529179	1.076037
Independent Variable
ESG	0.572691	0.6078	0.9501	0.0791	0.213987
Environment	0.538681	0.63955	0.9744	0	0.336896
Social	0.590842	0.6307	0.9732	0.0657	0.224753
Governance	0.561912	0.58	0.9599	0.0597	0.240326
Dividend Yield	0.033309	0.02605	0.6845	0	0.054551
Provision For Loan Loss	1,633,554	723,870.5	18,549,000	−925,000	2,415,079
Operating Profit Margin	0.046993	0.10085	0.5226	−2.0111	0.250602
Total Loans (ln)	18.57284	18.4779	20.68213	16.04809	1.095184
Return on Equity	−0.097159	0.06515	0.9814	−42.9847	2.125264
Inflation	0.015652	0.01539	0.048971	−0.044781	0.013377
IBES 12 Month Forward EPS	2722.307	1.067	323,418.6	−44,314.02	23,956.89
Capital Adequacy	0.135079	0.135	0.2206	−0.061	0.032291
Liquidity	0.970924	0.5906	11.5126	0.0814	1.27492
Market to Book Ratio	1.058286	0.87	5.86	−2.58	0.824165

below.

3.2. Idiosyncratic Risk Measure

The idiosyncratic risk is measured by the standard deviation of residuals from daily stock returns of the Carhart four-factor model. The Carhart four-factor model, stated below, includes the momentum factor as an addition to the Fama–French three-factor model [50]. This model is used widely by the previous literature to determine the idiosyncratic risk of companies [18,33,45].

$$(R_{it}-R_{ft})={\alpha }_i+{\beta }_{im} \left(R_{Mt}-R_{ft}\right)+{\beta }_{is}SM{B}_t+{\beta }_{ih}HM{L}_t+{\beta }_{iu}UM{D}_t+{\epsilon }_t$$

(1)

In the model above, $(R_{it}-R_{ft})$ represents the excess return for a bank i on a day t. Risk free rate ($R_{ft})$ stands for a 1-month T-bill rate and $\left(R_{Mt}-R_{ft}\right)$ is the excess return of the market portfolio for Europe. The other factors of the model are the difference between small and big stocks $SM{B}_t$; the difference between high and low book-to-market ratio stocks $HM{L}_t$; and the momentum factor $UM{D}_t$. Data for market return and other factors are obtained from the Kenneth French data library’s European database, and the daily excess returns for 31 banks are retrieved from the Eikon database. To derive the idiosyncratic risk volatility of each bank, which is represented by the standard deviation of residuals (${\epsilon }_t$) we run Equation (1). Following the previous research, logarithmic transformation is applied to the idiosyncratic volatility to ensure homoscedasticity, as shown in Equation (2) [18,35,51]. The idiosyncratic risk for bank i in year t is represented as $I{V}_{it }$ in Equation (2).

$$I{V}_{i t }=\ln (\frac{1-R_{i t}^2}{R_{i t}^2})$$

(2)

3.3. Method

The quantile regression method follows in order to investigate the relationship between the ESG performance and idiosyncratic risk. The quantile regression method is advantageous compared to conditional mean methods, as it explains the relationship between different risk levels and ESG within the sample population. The variance inflation factor (VIF) for all variables is tested to assess possible multicollinearity problems. No multicollinearity is detected, as none of the variables have a VIF higher than 5 (see Appendix B). Mean regression methods are sensitive to outliers, non-normal distribution, and heteroscedasticity of error terms, which could lead to misleading results. The quantile regression minimizes the sum of the absolute residuals, while mean regression methods minimize the sum of squared residuals. The quantile regression method has no sample selection bias when determining the quantiles compared to a piecewise regression [52]. The quantile regression divides the sample population into different percentiles with a quantile-fitting regression. The quantile approach is defined as:

$$Y_{it}={\beta }_{\theta }{x}_{it}+{\epsilon }_{\theta it}\hspace{1em}\hspace{1em}\hspace{1em} 0 < \theta < 1$$

(3)

where $Y_{it}$ is the dependent variable for bank i at a time t, and $x_{it}$ represents the vector of the explanatory variables at the $\theta th$ percentile for the dependent variable. The model’s error term, in which the conditional quantile distribution is zero, is ${\epsilon }_{\theta it}$. In order to investigate the risk-reducing impact of ESG, the following quantile model is developed:

$$\begin{array}{cc}{Q}_{\theta }\left(I{V}_{it}\mid X_{it}\right)=& {\delta }_i+{\beta }_{1}CS{R}_{it}+{\beta }_2{\mathit{Div}}_{it}+{\beta }_3{\mathit{PLLoss}}_{it}+{\beta }_{4}OP{M}_{it}+{\beta }_{5}Siz{e}_{it}\hfill \\ & +{\beta }_6{\mathit{ROE}}_{it}+{\beta }_{7}IBE{S}_{it}+{\beta }_8{\mathit{MTB}}_{it}+{\beta }_{9}CapAd{q}_{it}\hfill \\ & +{\beta }_{10}LI{Q}_{it}+{\beta }_{11}IN{F}_{it}+{\epsilon }_{it}\hfill \end{array}$$

(4)

where $Q_{\theta }(I{V}_{i t }| X_{i t})$ represents the $\theta th$ quantile regression function, and the 0.05, 0.25, 0.5, 0.75, and 0.95 percentiles are assigned to $\theta$ to investigate the ESG effect of 5 different percentiles. The dependent variable is the idiosyncratic volatility for bank i at a time t $(I{V}_{i t })$, and ${\beta }_{1}CS{R}_{it}$ is assigned for the overall ESG scores of bank i at a time t. The environmental, social, and governance scores are also tested separately by replacing ${\beta }_{1}CS{R}_{it}$ with ${\beta }_{1}EN{V}_{it}$, ${\beta }_{1}SO{C}_{it},$ and ${\beta }_{1}GO{V}_{it},$ respectively. The above model is estimated by employing the Machado and Silva (2019) methodology by using the xtqreg command of STATA software. This method has advantages over other methods as it considers the individual effects and endogeneity and makes calculations simpler.

Idiosyncratic bank risk (IR) is not independent of the bank-specific variables that represent bank characteristics. Therefore, following the previous research on idiosyncratic bank risk, we employ the following bank characteristics as control variables in our model [37,38,40]. Dividend payments (Div) represent the financial health of firms, and paying out dividends positively signals to the shareholders. A provision for loan loss (PLL) represents the credit risk associated with the bank and ensures future cash flow; therefore, it is included as a control variable. The operating profit margin (OPM), which shows a bank’s financial efficiency and management performance, is related to the idiosyncratic risk. Size is an important characteristic that could affect the IR of banks: bigger banks could manage financial risks more efficiently, and total loans are used as a proxy. Instead of total assets, total loans are used as it is assumed that total loans better represent bank-related risks. Profitability is another indicator included within the model, and return on equity (ROE) is used as a proxy. The market-to-book ratio (MTB) shows investment opportunity and is included as a control variable. The 12-month forward earnings per share rate from the Institutional Brokers’ Estimate System (IBES) is used to represent the expected future earnings. Capital adequacy (CA) was incorporated as a proxy to capture the banks’ capital risks. The liquidity ratio (LIQ) is used to capture bank liquidity risks and is included as a control variable. Finally, inflation (INF) is used as a control variable to capture the country-specific risks associated with banks.

4. Empirical Results and Discussion

Table 3. Quantile regression results with ESG overall.

	Q(5)	Q(25)	Q(50)	Q(75)	Q(95)
Dependent Variable = IV	IV	IV	IV	IV	IV
ESG	0.00305	−0.492	−0.879 *	−1.340 *	−1.985 *
	(0)	(−1.13)	(−2.37)	(−2.40)	(−1.99)
Dividend Yield	1.099	0.211	−0.483	−1.311	−2.467
	(−1.03)	(−0.33)	(−0.88)	(−1.60)	(−1.68)
Provision For Loan Loss	−5.796	−13.97	−20.37	−27.99	−38.64
	(−0.28)	(−1.13)	(−1.94)	(−1.77)	(−1.37)
Operating Profit Margin	−0.773	−1.097	−1.351 **	−1.653 *	−2.075
	(−0.80)	(−1.90)	(−2.76)	(−2.23)	(−1.57)
Total Loans	−0.343	−0.519 **	−0.657 ***	−0.821 ***	−1.051 **
	(−1.17)	(−2.94)	(−4.37)	(−3.62)	(−2.59)
ROE	−0.036	−0.0275	−0.0209	−0.013	−0.002
	(−1.25)	(−1.59)	(−1.43)	(−0.59)	(−0.05)
Inflation	−6.749	−8.056 *	−9.079 **	−10.3	−12
	(−0.99)	(−1.97)	(−2.61)	(−1.96)	(−1.28)
IBES 12-Month Forward EPS	−4.18 × 10−6	−4.83 × 10−6 ***	−5.33 × 10−6 ***	−5.94 × 10−6 **	−7 × 10−6
	(−1.50)	(−2.89)	(−3.76)	(−2.76)	(−1.77)
Capital Adequacy	3.636	2.778	2.107	1.308	0.19
	−1.26	−1.61	−1.44	−0.59	−0.05
Liquidity	−0.0553	−0.116 *	−0.164 ***	−0.221 ***	−0.301 *
	(−0.65)	(−2.26)	(−3.74)	(−3.34)	(−2.54)
Market-to-Book	0.128	0.0665	0.0186	−0.0384	−0.118
	−0.94	−0.81	−0.27	(−0.36)	(−0.63)
N	471	471	471	471	471

t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001.

, below, presents the results of Equation (4), in which the ESG score is the main independent variable. The results indicate a negative relationship between ESG and idiosyncratic risks for the quantiles 0.50, 0.75, and 0.95. These findings suggest that ESG has a negative relationship with medium/median- and high-risk banks. A closer analysis of the quantile base also provides some clues regarding the effect of ESG on the different risk levels. As can be seen from the ESG coefficients, which indicate the economic significance of the ESG impact on the IR, high-risk banks earn relatively more benefits by increasing their ESG scores. In other words, better ESG for these banks makes them more stable. For example, the negative effect of ESG is nearly two times higher when a medium-risk category bank (quantile 0.50) is compared with the highest risk category bank (quantile 0.95) (i.e., the coefficients are −0.879 and −1.985, respectively). These results support H1 for medium- and high-risk banks, where a negative relationship is expected between the ESG scores and idiosyncratic risks of banks. Though ESG is not statistically significant for the lowest risk group, we think this is acceptable since low-risk banks do not need ESG promotions. These findings suggest that ESG ensures communication with the key stakeholders of banks, and, as the risk level of a bank increases, this information flow becomes more important.

Our results have theoretical and empirical implications as well. The results align with the stakeholder and legitimacy theories, which argue that ESG strategies ensure a stable relationship with stakeholders by legitimizing the banks’ actions and providing insurance-like protection against possible adverse shocks, thereby minimizing the IR. This implies that ESG could be an important risk management tool and an ESG disclosure become more important as the risk-mitigation effect increases with the riskiness of the bank. The results also support the findings of previous researchers who used accounting-based risk measures (Z-score, non-performing loans) and found a negative relationship between ESG and the riskiness of banks [9,10,11]. Though classical models do not consider CSR as a risk factor, the findings show that promoting socially responsible actions minimizes the idiosyncratic risk for medium- to high-risk category banks. Hence, the CSR indicators should not be undermined by the researchers in their risk models.

Table 4. Quantile regression results with environmental score.

	Q(5)	Q(25)	Q(50)	Q(75)	Q(95)
Dependent Variable = IV	IV	IV	IV	IV	IV
Environment	−0.59	−0.708 *	−0.786 ***	−0.900 *	−1.033
	(−1.24)	(−2.50)	(−3.33)	(−2.56)	(−1.72)
Dividend Yield	1.163	0.158	−0.507	−1.476	−2.612
	−1.07	−0.25	(−0.93)	(−1.84)	(−1.90)
Provision For Loan Loss	−12.86	−15.43	−17.14	−19.63	−22.54
	(−0.71)	(−1.43)	(−1.91)	(−1.47)	(−0.99)
Operating Profit Margin	−1.162	−1.236 *	−1.285 **	−1.356 *	−1.44
	(−1.37)	(−2.45)	(−3.06)	(−2.17)	(−1.35)
Total Loans	−0.133	−0.355	−0.501 **	−0.715 **	−0.965 *
	(−0.40)	(−1.81)	(−3.05)	(−2.93)	(−2.32)
ROE	−0.0468	−0.0334 *	−0.0244	−0.0115	0.00377
	(−1.77)	(−2.12)	(−1.86)	(−0.59)	−0.11
Inflation	−7.818	−8.738 *	−9.349 **	−10.24 *	−11.28
	(−1.20)	(−2.26)	(−2.90)	(−2.13)	(−1.38)
IBES 12-Month Forward EPS	−3.25 × 10−6	−3.49 × 10−6 *	−3.65 × 10−6 **	−3.88 × 10−6	−4.15 × 10−6
	(−1.18)	(−2.14)	(−2.68)	(−1.91)	(−1.20)
Capital Adequacy	5.548	4.224 *	3.346 *	2.068	0.569
	−1.87	−2.39	−2.27	−0.94	−0.15
Liquidity	−0.0519	−0.107 *	−0.144 ***	−0.198 **	−0.260 *
	(−0.62)	(−2.15)	(−3.44)	(−3.18)	(−2.46)
Market-to-Book	0.0773	0.0255	−0.00875	−0.0586	−0.117
	−0.56	−0.31	(−0.13)	(−0.58)	(−0.67)
N	471	471	471	471	471

t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001.

below presents the relationship between the environmental dimension of ESG and the idiosyncratic volatility of banks. The environmental dimension negatively correlates with the idiosyncratic risk for low- to high-risk category banks in the 0.25, 0.50, and 0.75 quantiles. Our findings support the argument that ever since the stakeholders’ attention to environmental issues, such as global warming and the carbon footprints of institutions, has increased, the pressure on the banking sector has also increased. A significant negative relationship is evident between the environmental dimension’s performance and the idiosyncratic bank risk. Therefore, our second hypothesis, H2, is supported, except for the lowest (quantile 0.05) and highest (quantile 0.95) risk category banks. The economic significance of the negative effect becomes slightly more robust as the banks’ risk categories increase. Our results show that communicating with stakeholders about the environmentally friendly actions of banks minimizes the idiosyncratic risk volatility by legitimizing the banks’ activities. This result aligns with our expectations, as stakeholders, including the regulatory bodies for European banks, demand increased disclosure over environmental issues. The results support the argument of the previous literature, which found a risk-reducing relationship between the environmental performance of banks and accounting-based risk measures due to improved reputation [11,12,13,14,15,16,17,18,19].

Results for the effect of the social dimension on banks’ idiosyncratic risks are presented in

Table 5. Quantile regression results with social score.

	Q(5)	Q(25)	Q(50)	Q(75)	Q(95)
Dependent Variable = IV	IV	IV	IV	IV	IV
Social	0.276	−0.0783	−0.345	−0.693	−1.105
	−0.46	(−0.21)	(−1.18)	(−1.66)	(−1.55)
Dividend Yield	0.996	0.139	−0.508	−1.348	−2.347
	−0.9	−0.2	(−0.93)	(−1.74)	(−1.78)
Provision For Loan Loss	−4.344	−12.46	−18.58	−26.54	−36
	(−0.20)	(−0.95)	(−1.79)	(−1.79)	(−1.42)
Operating Profit Margin	−0.724	−1.05	−1.297 **	−1.616 *	−1.997
	(−0.73)	(−1.73)	(−2.68)	(−2.34)	(−1.69)
Total Loans	−0.384	−0.604 ***	−0.769 ***	−0.985 ***	−1.240 ***
	(−1.33)	(−3.41)	(−5.44)	(−4.89)	(−3.62)
ROE	−0.0324	−0.0256	−0.0204	−0.0137	−0.0058
	(−1.07)	(−1.37)	(−1.38)	(−0.65)	(−0.16)
Inflation	−6.96	−8.253	−9.228 **	−10.50 *	−12
	(−0.99)	(−1.93)	(−2.72)	(−2.16)	(−1.45)
IBES 12-Month Forward EPS	−3.8 × 10−6	−4.52 × 10−6 **	−5.05 × 10−6 ***	−5.75 × 10−6 **	−6.59 × 10−6 *
	(−1.40)	(−2.71)	(−3.83)	(−3.04)	(−2.04)
Capital Adequacy	3.475	2.561	1.872	0.976	−0.0889
	−1.17	−1.41	−1.3	−0.47	(−0.03)
Liquidity	−0.063	−0.121 *	−0.164 ***	−0.221 ***	−0.288 **
	(−0.71)	(−2.22)	(−3.78)	(−3.56)	(−2.73)
Market-to-Book	0.136	0.0789	0.0358	−0.0204	−0.0871
	−0.98	−0.93	−0.53	(−0.21)	(−0.53)
N	471	471	471	471	471

t statistics in parentheses * p < 0.05, ** p < 0.01, *** p <0.001.

below. H3 could not be supported for all quantile levels as no significant relationship is detected between the banks’ social dimensions and the idiosyncratic bank risks. These results show that stakeholders focus more on environmental issues and good governance, rather than social aspects, such as a positive commitment to society or good employee relations in banks. However, previous research has found a risk-reducing relationship for some elements of the social dimension, such as human rights and labor protections [11]. Still, the overall social score has no significant relationship with bank riskiness.

Table 6. Quantile regression results with governance score.

	Q(5)	Q(25)	Q(50)	Q(75)	Q(95)
Dependent Variable = IV	IV	IV	IV	IV	IV
Governance	−0.00772	−0.357	−0.638 *	−0.959 **	−1.404 *
	(−0.02)	(−1.14)	(−2.49)	(−2.62)	(−2.17)
Dividend Yield	0.868	0.105	−0.51	−1.211	−2.184
	−0.81	−0.16	(−0.96)	(−1.59)	(−1.62)
Provision For Loan Loss	−3.51	−12.25	−19.29	−27.33	−38.47
	(−0.17)	(−0.96)	(−1.87)	(−1.84)	(−1.47)
Operating Profit Margin	−0.61	−1.005	−1.322 **	−1.685 *	−2.188
	(−0.62)	(−1.67)	(−2.71)	(−2.41)	(−1.77)
Total Loans	−0.321	−0.559 ***	−0.750 ***	−0.969 ***	−1.273 ***
	(−1.21)	(−3.44)	(−5.64)	(−5.12)	(−3.79)
ROE	−0.0339	−0.0229	−0.0141	−0.0041	0.00982
	(−1.16)	(−1.29)	(−0.98)	(−0.20)	−0.27
Inflation	−7.745	−8.174 *	−8.520 *	−8.915	−9.463
	(−1.14)	(−1.97)	(−2.54)	(−1.85)	(−1.11)
IBES 12-Month Forward EPS	−4.23 × 10−6	−5.09 × 10−6 **	−5.79 × 10−6 ***	−6.59 × 10−6 **	−7.69 × 10−6 *
	(−1.45)	(−2.86)	(−4.01)	(−3.18)	(−2.10)
Capital Adequacy	3.04	2.061	1.273	0.374	−0.873
	−1.05	−1.16	−0.89	−0.18	(−0.24)
Liquidity	−0.0526	−0.112 *	−0.161 ***	−0.216 ***	−0.292 **
	(−0.59)	(−2.07)	(−3.63)	(−3.41)	(−2.60)
Market-to-Book	0.132	0.0791	0.0369	−0.0113	−0.0781
	−0.97	−0.95	−0.55	(−0.12)	(−0.45)
N	471	471	471	471	471

t statistics in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001.

below presents the results of Equation (4), in which governance is the main independent variable. Governance negatively relates to the idiosyncratic risk for medium- to high-risk category banks. Similar to the overall ESG score, as the riskiness of banks increases, the negative effect of the governance dimension also increases. The risk-reducing impact is more than double for the highest risk category (quantile 0.95) compared to medium-risk category (quantile 0.50) banks, where the coefficients are −1.404 and −0.638, accordingly. The signaling effect of governance on stakeholders, regarding management quality, transparency, and accountability, becomes essential as the risk category of banks increases. These findings show that the environment and governance dimensions are similarly crucial and negatively relate to the IR of banks for medium- and high-risk banks. These results indicate that focusing on the governance and environment dimensions encourages a healthy relationship with the stakeholders, which leads to a risk-mitigation effect of the idiosyncratic risks of banks. In contrast, the social dimension has no significant effect. However, unlike other dimensions, the governance dimension has a strong negative relationship with IR for the highest-risk category banks, and this aligns with the expectation that the governance dimension has a higher negative impact on firm-specific risks, as it is more relevant and visible to the investors [8]. This finding supports the idea that good governance strengthens the banks’ risk management.

Control variables show that size is the most critical factor in determining the IR of banks. Size has a negative relationship with the IR across all quantile levels, except for the lowest risk quantile, and the risk-reducing effect of size increases as the riskiness of banks increases. This result aligns with the previous literature, which argues that size is the most crucial factor for determining the IR of banks and that, as the size of the banks increases, the IR tends to decrease [39]. Similar to size, liquidity negatively affects banks’ IR over all quantile levels, except for the lowest risk quantile, but the magnitude is considerably lower than the size. As a market estimation figure is anticipated, the IBES 12-month forward EPS expectations are also significant determinants of the IR of banks for medium- to high risk level category banks (quantile 0.25, 0.50, and 0.75).

5. Robustness

To confirm the results provided in the previous section, we conducted some robustness checks. For this purpose, we use accounting-based risk measures to assure the robustness of our results. The previous literature widely accepts the Z-score and capital adequacy (CA) as firm-specific risk measures [9,11,19]. The analysis results in replacing the idiosyncratic volatility with the Z-score as the dependent variable, provided in Appendix C, and the results for CA, replacing the idiosyncratic volatility, are presented in Appendix D. The positive coefficients indicate that ESG contributes to bank stability and has an inverse relationship with bank riskiness. The results align with our findings, in which ESG has an inverse relationship with the idiosyncratic risk of banks, and the negative relationship increases as the risk level of banks increases. This shows that results support each other for a market-based and an accounting-based risk measure. Additionally, these findings are in line with the previous literature, which found an inverse relationship between the CSR performance and the riskiness of banks [9,10,11].

To check the robustness of the ESG dimensions, we regress each dimension to the Z-score and CA. The Z-score, CA, and environmental dimension analysis results are presented in Appendix E and Appendix F, respectively. The results presented in Appendix E and Appendix F show that the environmental dimension positively correlates with bank stability for all quantile levels. These findings align with the previous literature and support the findings of this study. Appendix G presents the results of the governance dimension for the Z-score, and Appendix H shows the governance dimension for CA. However, we do not find any significant relationship between the governance dimension and these accounting-based risk measures. These findings do not support our initial results, which found a significant inverse relationship between governance and the idiosyncratic volatility of banks. Finally, Appendix I shows the relationship between the social dimension and Z-score, while Appendix J presents the relationship between the social dimension and CA. The results indicate a positive relationship between the social dimension and risk measures of banks over the 0.25, 0.50, 0.75, and 0.95 quantile levels. These findings contradict the initial results of this research, which could not identify a significant relationship between the social dimension and idiosyncratic bank risk. Although the robustness findings of the governance and social dimensions for accounting-based risk do not align with the idiosyncratic risk, previous literature has indicated that the governance and social dimensions have different effects on the various bank risks [11]. This shows that the impact of the ESG dimensions could differ for different risk measures, which could explain the different results for accounting and market-based risk measures. Our findings also suggest that market-based risk measures could be more important as these represent the stakeholders’ perceptions better than the accounting figures of the banks.

6. Conclusions

This research contributed to the literature by analyzing the effect of ESG on idiosyncratic bank risk. For this purpose, first, we use the overall ESG scores, and secondly, the ESG’s dimensions, environmental, social, and governance, separately. The sample consists of 31 European banks between 2002 and 2019. This study revealed that ESG has a negative relationship with the idiosyncratic risk of banks for medium- to high-risk levels. Another important finding of this research is that the effect of ESG changes according to the risk levels of banks. As the riskiness of the banks increases, a stronger relationship is detected. Additionally, this research shows that the findings of a market-based risk measure align with the findings of accounting-based risk measures for the ESG’s effects on firm-specific banks’ risks. These findings have both theoretical and practical implications.

From the theoretical perspective, aligned with the stakeholder and legitimacy theories, this research reveals that ESG has a negative relationship with the idiosyncratic risk of banks for medium- to high-risk levels. ESG acts as a communication tool with key stakeholders, minimizes the information asymmetries, and legitimizes the banks’ actions. It is revealed that ESG has a negative relationship with the idiosyncratic risk over medium- to high-risk level quantiles. As the riskiness of banks increases, the relationship between ESG and risk becomes more important. This shows that ESG contributes more to the stability of banks as the risk level increases. Analyses of the individual ESG dimensions showed that the governance and environmental dimensions have a strong negative impact. This suggests that stakeholders focus more on governance quality, which signals the management quality of banks, and that the environment performance has a significant relationship with the banks’ reputations due to society’s increased attention to environmental issues. There was no significant association between the social dimension and idiosyncratic bank risk, indicating that the stakeholders’ interest in social projects is not as high as in other ESG dimensions. Alternatively, stakeholders view social projects as window-dressing activities and do not prioritize them.

Practically, these results reveal some of the reasoning behind banks’ increased commitment to ESG projects. ESG ensures effective communication and good relations with all stakeholders, including customers, employees, shareholders, government, and regulators. The results revealed that ESG negatively relates to idiosyncratic risk for medium- to high-risk banks. Therefore, ESG could be an essential communication tool. With stakeholders legitimizing the actions of banks, and, as banks’ riskiness increases, this tool becomes more important. The analyses of the ESG dimensions showed that governance quality and the environmental dimension have a similar negative relationship with idiosyncratic risk, but governance quality is more important for the highest risk category. This research showed that focusing on governance quality and ESG’s environmental dimension could help banks minimize the idiosyncratic risks. Our results also suggest that regulators and policymakers can use ESG-type non-financial information disclosure requirements as risk-reducing policy tools and maintain the financial stability of the system. Nonetheless, these findings suggest that the environmental and, particularly, the governance dimensions should be emphasized more by regulators and policy makers. This may also explain regulators’ increased disclosure requirements on ESG information.

Lastly, this research has limitations due to the available data on the European market. Future research could increase the number of banks or focus on other economic regions. Different market-based risk metrics could be used to isolate the effect of the ESG dimensions on the riskiness of banks. Additionally, further research could expand this research by investigating the stakeholders’ perceptions of particular aspects of the environment, social, and governance dimensions, which could reveal the interests of key stakeholders. This could provide further guidance for bank management and regulators on socially responsible and sustainable banking and its relation to firm-specific risks.