Doing good for avoiding peril: can CSR help firms shelter from risks?

In recent years, corporate social responsibility (CSR) has emerged as a critical mechanism for organizations seeking to mitigate various types of business risks associated with both financial instability and reputational threats and to enhance long-term sustainability. CSR performance could be considered as a shield against adverse financial firm risk effects (Godfrey et al., 2009; Shahab et al., 2019), as a risk management approach to protect firms from the risk of adverse political, regulatory, social and financial sanctions (Godfrey, 2005; Mishra and Modi, 2013). The combined benefits of CSR performance may exceed the related costs; therefore, CSR performance could help companies to some degree manage their financial disposition (Godfrey, 2005; Salehi et al., 2018) to lessen the firm risk and reduce corporate risk-taking (Wang et al., 2023). Investigating the consequences of CSR performance is still an ongoing concern for academic researchers in the finance and accounting field due to its major importance to various stakeholders (Boubaker et al., 2020; Ben Saad and Belkacem, 2022).

Prior studies on the consequences of CSR performance suggest that it can reduce the risk of financial distress (Jo and Na, 2012; Sun and Cui, 2014; Al-Hadi et al., 2019). For example, El Ghoul et al. (2011) highlighted that CSR performance negatively affects the total cost of capital, where risk is a key element in determining it. Similarly, Sassen et al. (2016) got similar results when analyzing how CSR performance impacts a firm’s risk. However, the mechanisms that influence this relationship need to be further investigated (Nirino et al., 2022). Prior studies have examined the effect of CSR performance on firm risk independently or explored the general role of governance structures (Orlitzky and Benjamin, 2001; Jo and Harjoto, 2011). However, addressing how board monitoring and chief executive officer (CEO) power interact with CSR performance to influence risk outcomes on a global scale remains a gap in the literature regarding the nuanced governance–CSR-risk relationship.

The efficacy of CSR performance in achieving risk-mitigating outcomes depends significantly on a firm’s internal governance structure. Specifically, two governance mechanisms – board monitoring and CEO power – play pivotal roles in shaping how CSR performance is leveraged for corporate risk management strategies. As a cornerstone of corporate governance, board monitoring is a critical supervisory force to ensure accountability and strategically align managerial decisions with shareholder interests (Adams et al., 2010). Strong board oversight not only supports CSR performance but also enhances its risk-mitigating capacity by prioritizing ethical decision-making and reinforcing compliance with regulatory standards (Hillman and Dalziel, 2003). On the other hand, powerful and over-confident executives tend to use CSR as leverage for their private gains leading to firm risk (Kuo et al., 2021), whereas they may also assume their stewardship role and champion CSR performance as a proactive risk management tool, leveraging its influence to integrate CSR performance into core business strategies (Tang et al., 2015). Our aim, therefore, is to empirically investigate the complex interplay between CSR performance and corporate governance mechanisms in influencing risk mitigation, specifically focusing on the roles of board monitoring and CEO power. This relationship attracts the attention of boards of directors on how to monitor CSR performance that minimizes the firm risk. Thus, we expect the board monitoring and the CEO power to have a moderating role in the relationship between CSR performance and firm risk.

Consequently, this study seeks to contribute to existing literature in several ways. First, previous studies investigated the direct relationship between the financial risk of firms and their CSR performance showing inconclusive results. For example, some found that CSR performance reduces financial risk (Hsu and Chen, 2015; Chollet and Sandwidi, 2018), some others found that it exacerbates firm risk (Nguyen and Nguyen, 2015). On the other hand, Dumitrescu et al. (2020) found that aggregate CSR proxy has no significant effect on firm risk but found a positive effect for the governance pillar (GOP) and a negative effect for the social pillar. Thus, there is a need to update the existing literature with a broad sample to provide robust recent evidence for CSR-firm risk links on an international scale drawing on both accounting-based and market-based risk measures. Second, researchers have found that CEO characteristics such as CEO narcissism are associated with CSR (Agnihotri and Bhattacharya, 2024). Powerful and over-confident executives tend to use CSR as leverage for their private gains (Kuo et al., 2021). Hence, exploring if CEO power strengthens or weakens the link between CSR and firm risk might suggest theoretical and managerial implications for firms. Third, exploring the role of board monitoring in the risk-reducing effect of CSR might help firms design their upper echelon accordingly. Investigating the role of board monitoring and CEO power in moderating the relationship between CSR performance and firm risk remains as a gap in the literature. By empirically investigating the moderating dynamics, we contribute to the fields of corporate governance and CSR by integrating governance theories, such as stakeholder, agency and stewardship, to offer a deeper understanding of the structural determinants of CSR performance and extend theories on the integration of CSR within broader organizational risk management frameworks. Thus, this study aims to provide valuable insights for academics and practitioners seeking to enhance firm stability through CSR performance, demonstrating how these governance mechanisms can strategically optimize the risk-mitigating potential of CSR performance. Fourth, this study adopts an encompassing CSR performance proxy that includes ten indicators under its three pillars, namely, environmental, social and governance (ESG). This offers a deeper understanding of the relationship between each pillar and the financial risk. Fifth, prior studies in this area tend to use a small sample size (Benlemlih and Girerd‐Potin, 2017), or data from a single country (Ayton et al., 2022; Nirino et al., 2022), using a cross-industry very large worldwide sample of 55,475 firm-year observations for 13 years (2006–2019), this study extends and complements the findings of the prior studies, applying the generalized methods of moments (GMM) technique to manage potential endogeneity issues.

We found that CSR performance (for composite CSR proxy and its ESG pillars) significantly reduces firm risk for both accounting-based and market-based risk proxies. In addition, the moderation analyses showed that board monitoring and environmental and social performance are substitutes for alleviating firm risk. Furthermore, while CSR performance in firms with powerful CEOs, particularly environmental and governance performance, exacerbates firm risk, social performance does not. The robustness tests confirmed that our findings are robust to industry-adjusted CSR indicators, alternative estimators and alternative firm risk indicators.

This research has important managerial implications. Management needs to know that stakeholders react positively to CSR performance. Moreover, environmental and social performances are a substitute for governance performance, advising firms whose financial resources are limited, to selectively invest in either the governance dimension or environmental and social CSR dimensions to alleviate firm risk. However, for firms with sufficient financial slack resources for discretionary expenditures, it should be noted that they are all influential per se in curbing firm risk. The CSR performance, which is discretionary, exacerbates agency conflicts and hence firm risk under powerful CEOs implying that managerial over-confidence and opportunism reverse the risk-reducing effect of CSR. Finally, the findings may guide shareholders in shaping their portfolios considering firm riskiness, CSR performance and board contingencies.

The paper is structured as follows. In Section 2, we introduce the theoretical framing upon which our hypotheses are developed. Then, in Section 3, we discuss the research methodology, data collection procedure, variable construction and the models used for hypothesis testing. Next, our findings are presented and discussed in Section 4. Sections 5 and 6 conclude the paper, by discussing the results suggesting the implications of this research for researchers and managers and proposing avenues for further research.

Stakeholder theory posits that CSR performance strengthens relationships between firms and key stakeholder groups, such as customers, employees, regulators and the community, thereby, stabilizing their operational environment and reducing risks associated with reputational damage and regulatory backlash. The theory has been adopted by many researchers in the field of CSR and corporate governance (Farooq et al., 2024; Khan et al., 2024) as it extended the board’s responsibilities from a traditional approach focusing on shareholders toward the interests of stakeholders. By proactively addressing stakeholder needs, companies foster trust and loyalty, which reduces potential risks of boycotts, legal disputes and reputational crises (Freeman, 1994; Hillman and Keim, 2001). These positive effects help buffer firms against adverse impacts from regulatory changes or social backlash, contributing to overall risk reduction (Hillman and Keim, 2001). CSR performance can mitigate the risk of overreactions to bad news involving the company and can reduce the volatility of the share prices (Nirino et al., 2021). Strong CSR performance builds trust, which can insulate firms from sudden adverse shocks by fostering stakeholder loyalty and cooperation during crises (Sciarelli et al., 2024).

It has been evidenced in the literature that CSR performance may play a role in mitigating different types of risks (Sun and Cui, 2014; Hsu and Chen, 2015). For example, Al-Hadi et al. (2019) show that CSR performance significantly reduces the financial distress of the firm, Albuquerque et al. (2019) conclude that CSR performance decreases systematic risk and Boubaker et al. (2020) show a negative relationship between CSR performance and financial distress. Mishra and Modi (2013) find that firm CSR performance is negatively and significantly related to idiosyncratic risk. Jo and Na (2012) found that firms with high CSR ratings exhibited lower systematic risk and improved financial stability. CSR performance can affect a firm’s risk in three ways (Attig et al., 2013), by increasing the possibility of long-term survival, signaling the effective utilization of resources and reducing stakeholder controversies costs. Moreover, CSR performance acts as a hedge against financial constraints as the firm may be able to rely on its linkages with various stakeholders to mitigate financial risk (Al-Hadi et al., 2019).

Prior studies used different measures and indicators of firm risk, including the cost of finance, volatility in returns, credit rate downgrading, stock price crash and default risk. If CSR performance is associated with a good business reputation, increased employee satisfaction and higher customer loyalty, it can reduce the possibility of financial distress by positively influencing future cash flows. This can lower the cost of financing by raising credit ratings (Attig et al., 2013). For example, Sun and Cui (2014) show that CSR performance improves creditworthiness and Kim et al. (2014) show that firms with highly CSR-oriented policies have lower stock price crash risk. Although credit rating reflects firms’ financial distress, it may not strictly measure true financial distress. Using distance to default risk, Dumitrescu et al. (2020) show no impact of CSR initiatives on financial distress. This study extends the above stream of research by using two alternative proxies for firm risk, the volatility of profitability and the volatility of firm value. By doing so, we assess whether CSR performance attenuates accounting and market risk.

The ability of firms to meet the expectations regarding CSR allows them to create a positive image and greater stakeholders’ loyalty, which reduces a firm’s risk (Godfrey et al., 2009). For example, El Ghoul et al. (2011) concluded that irresponsible companies, in terms of CSR, are seen as riskier by investors. Companies with lower CSR performance have lower financial performance and value for shareholders (Nirino et al., 2021). Companies with greater CSR performance may reduce the possibility of crises related to social and environmental aspects, keeping performance and cash flows stable and reducing the firm’s risk. Thus, we expect that firms with CSR performance enjoy lower financial risks and formulate the following hypothesis:

Agency theory suggests that boards act as monitors of the management to ensure alignment with shareholder interests. By monitoring CSR performance, boards can limit managerial discretion and prevent environmentally or socially irresponsible behaviors, which could otherwise increase legal, regulatory and reputational risks. Board monitoring strengthens the credibility and accountability of CSR performance, which, in turn, helps mitigate potential risks from noncompliance or stakeholder backlash (Jensen and Meckling, 1976; Jo and Harjoto, 2011). Research shows that shareholders’ interests are better served when the protection of stakeholders is not left to CEOs’ discretion, which, in turn, reduces the potential for financial risk (Cespa and Cestone, 2007). The board’s decisions play a primary role in achieving high CSR performance. CSR practices are positively associated with internal corporate governance and monitoring mechanisms, including the board (Jo and Harjoto, 2011).

Stakeholder theory further supports the moderating role of board monitoring in the relationship between CSR performance and firm risk. Active boards are better equipped to assess stakeholder demands, ensuring that CSR performance effectively addresses environmental concerns and social expectations. Research indicates that active boards show a greater ability to reduce exposure to potential financial losses due to unmet stakeholder expectations. Prior literature shows that corporate governance shapes CSR performance strategies. For example, Nirino et al. (2022) found that firm risk decreases more when CSR strategies are used with appropriate corporate governance mechanisms.

However, our approach is to analyze the interaction of environmental and social pillars of ESG with board monitoring [¹]. With this investigation, we aim to highlight if environmental and social performance and board monitoring substitute for or complement each other in reducing firm risk. While the substitution effect implies one factor reduces the marginal effect of another, the complementary effect means one factor reinforces the marginal effect of another factor in reducing risk (Oh et al., 2018). Boubaker et al. (2020) find that US firms with high CSR performance mitigate financial distress risk only when they have strong internal corporate governance mechanisms. Hence, CSR and governance mechanisms become more effective through cooperation and mutual enhancement in reducing risk. This supports the complementary hypothesis; that CSR performance reduces firm risk more profoundly in firms with stronger boards. Following this argument, we formulate below hypothesis:

Agency theory suggests that board monitoring typically serves aligning management actions with shareholder interests (Jensen and Meckling, 1976); however intensive monitoring can lead to a short-term focus, prioritizing financial outcomes over CSR performance. Boards heavily focused on monitoring may pressure managers to prioritize immediate financial performance, potentially substituting CSR risk management with conventional financial risk controls. This could inadvertently diminish the protective effect that CSR performance has on firm risk. The negative relation between CSR performance and risk is pronounced for firms with less effective corporate governance. Thus, we expect that strong board monitoring limits the effect of CSR on risk. This is consistent with Kim et al. (2014) and Orazalin and Baydauletov (2020), who found that CSR performance is particularly important when governance mechanisms such as monitoring by the boards are weak implying the substitutive relationship. The results are consistent with our notion that both CSR and board monitoring need not be strong to reduce firm risk; one is sufficient to be strong. This leads to the substitution hypothesis such that CSR and board monitoring substitute each other in reducing firm risk:

According to agency theory, the CEO duality makes CEOs more powerful decreasing the positive impact of CSR performance on firm outcomes. Following stakeholder theory, CEO duality results in weak monitoring of the board, reducing the impact of CSR performance on firm risk. However, the stewardship theory (Davis and Schoorman, 1997) states that powerful CEOs enhance the decision-making process and help CEOs conduct proactive CSR performance, due to executives’ intrinsic motivation and self-fulfillment.

Agency theory explains the possible effect of the CEOs’ power in determining firms’ risk attitude and CSR performance. CEOs become powerful when they perform the dual role of CEO and chairperson. When CEOs become more powerful, they may not act in the best interests of stakeholders and shareholders, engaging in actions to maximize their interests (Harper and Sun, 2019). Moreover, Jensen (1993) points out that when the CEO also holds the position of the chairman of the board, internal control systems fail, as the board cannot effectively perform its key functions. In a CEO duality setting, the CEO is assigned multiple tasks with possibly incongruent objectives prioritizing some goals at the expense of others (Hill and Jones, 1992; Jo and Harjoto, 2011). Competing goals may lead to focus prioritization concerns on the part of the CEO. In addition, Fama and Jensen (1983) argue that the concentration of decision management and decision control in one individual reduces a board’s effectiveness in monitoring top management.

CEO duality represents the multiplicity of contracted tasks (Eisenhardt, 1989; Holmstrom and Milgrom, 1991) affecting CSR and firm risk relationships. CSR performance requires multiple tasks of different natures endorsing specific outcomes in favor of a group of stakeholders that may negatively affect other stakeholders and, thus, affect firm risk. Powerful CEOs may trade-off, mainly in the short run, the interests of some stakeholders that might put their jobs at risk, such as not favoring environmental policies. In addition, a powerful CEO may mitigate the board’s monitoring effectiveness and reduce the board’s involvement in CSR performance. Empirical research findings support this view. For example, Li et al. (2016) and Sheikh (2019) found that CEO power is negatively correlated with CSR performance. Rashid et al. (2020) found that CEO power is negatively associated with the level of CSR, reducing the positive impact of CSR disclosure on a firm’s financial performance. Thus, we expect that CSR performance does not reduce firm risk in firms with CEO duality and formulate the following hypothesis:

Stewardship theory has the opposite view to the agency theory assuming that individuals will naturally align their interests with the firm (Davis and Schoorman, 1997). Stewardship theory emphasizes the positive effect of CEO duality on better firm performance due to the agility in decision-making that the combination of CEO and chairman roles provides (Freeman and Hasnaoui, 2011). Adopting stewardship theory, studies have found a positive relationship between the CEO duality and firm performance (Wang et al., 2014). The stewardship theory assumes that CEOs are stewards and will act in the stakeholders’ optimum interest. It explains that CEO duality may enhance the firms’ performance by maximizing shareholder interests (Mubeen et al., 2021). CEOs are involved in CSR performance as part of the discharge of their accountability to all stakeholders. Stewardship theory suggests that CEOs with substantial influence and authority are more likely to champion CSR performance as part of a broader strategic approach to firm stability and stakeholder trust, thus, reinforcing CSR’s risk-reducing effects.

Empirical research findings support this view. CEO duality positively and significantly impacts CSR performance (El Saleh and Jurdi, 2024). For example, Jo and Harjoto (2011) state that a more powerful CEO leads to better CSR performance. Jiraporn and Chintrakarn (2013) showed that high CEO power is associated with better CSR performance. Li et al. (2018) found that CEO power positively moderates the association between CSR and financial performance. CEO duality can enhance CEOs’ focus on social issues (Zhao et al., 2016), providing leadership to rapidly respond to environmental changes. CEOs with dual roles engage in opportunistic activities to improve their reputation by empowering CSR activities, given the interest of various stakeholders in these activities (El Saleh and Jurdi, 2024). Adopting stewardship theory, Torres and Augusto (2021) found that powerful CEOs are likely to align with firms’ orientation toward social issues. Powerful CEOs, who act as stewards of the firm, may prioritize the organization’s long-term stability and welfare over short-term financial goals. Thus, we expect that CSR performance reduces firm risk more profoundly in firms with CEO duality and formulate the following hypothesis:

Figure 1 depicts the hypothesized relationships among variables.

In this section, we describe the sample formation and distribution of the study, define the research variables and explain the empirical methodology and models.

The sample is drawn from the Thomson Reuters Eikon database which is a reliable source for collecting mainly ESG data, firm financials and board characteristics. The data screening process is a crucial step before further analysis (Hair et al., 2019). The data screening includes preparation of the raw data, winsorization, missing data analysis, outlier detection and imputation steps. The raw data is retrieved from the databases, then the raw data set is subject to a cleaning process by removing the string characters or words from the numerical variables, variables are organized and finally, the data set is prepared for the analyses.

The data sample covers 13 years, 10 major business sectors affiliated with 65 countries and 55,475 observations. Our investigation is based on the period 2006–2019 since the ESG data were rare before 2006, so we excluded earlier periods. We capped the data in 2019 as it was the most recent year for which the data were available when we commenced this research. The long period and a wide range of sectors affiliated with numerous countries within the sample reinforce the generalizability of our findings. The initial sample size was 59,182 observations. We excluded 3,707 observations belonging to 2005 and earlier years to focus on the target sample, so the sample starts from 2006 onwards. The final sample size was 55,475 observations. Detailed sample development is reported in Table 1 (Panel A) and the sample’s country distribution is reported in Table A2 ( Appendix).

Some of the variables were heavily skewed around mean values with a high standard deviation. Therefore, StdROA, StdTobinq, Bsize, ROA, Leverage, Currentr and RDintensity [²] were winsorized at one per cent of both tails. The values at the end tails of the one per cent levels are replaced with their corresponding winsorized values.

In the following phase, the missing data analysis is performed. Some of the research variables had missing values with relatively less than 5%. Namely, Resource had 0.52% firm-year missing records; Emissions, Workforce, Humanrights, Community and Productesp had 0.03%; ENPs and SOPs had 0.02%; Bsize had 0.28%; Fsize and ROA had 0.21%; Leverage had 0.23%; and FFP had 0.81% missing observations in the sample. The ratio of the missing values is significantly less than 5% which is inconsequential (Schafer, 1999). In the final step of the data screening, the missing values are imputed using the Markov Chain Monte Carlo approach with linear regression. The sample distribution based on the year and the sectors are presented in Table 1. According to the results of the sector distribution, 22.63% of the firm-year observations are from financial, 16.13% are from industrial, 14.47% are from consumer cyclical, 10.28% are from basic material, 8.81% are from technology, 7.12% are from healthcare, 6.9% are from consumer noncyclical, 6.84% are from energy, 4.2% are from utilities and 2.62% are from telecommunications service sectors. Regarding the year, the firm-year observations range between 2.96% in 2006 and 13.88% in 2019 with an increasing trend from 2006 to 2019. In total, there are 55,475 firm-year observations in the sample.

The study uses two alternative proxies for firm risk; one is the rolling standard deviation of return on assets (StdROA) and the other one is the rolling standard deviation of Tobin’s Q over three years (StdTobinq) (Vithessonthi, 2016; Hoang et al., 2021). While the former measures the volatility of profitability, the latter measures the volatility of firm value. In the baseline analysis, we adopted StdROA as the dependent variable, in the robustness test, we adopted StdTobinq.

As proxies of CSR performance, we used aggregate ESG scores (ESGs), its three pillars’ scores, namely, environmental (ENPs), social (SOPs) and governance (GOPs) (Govindan et al., 2021) and 10 individual indicators under these three pillars (Please see Table A1 in  Appendix for the definitions of all ESG proxies).

We use two indicators for moderators: one for board monitoring which is the management score (Management) dimension of GOPs of ESG and the other one for CEO power (CEOpow) which is CEO duality (Uyar et al., 2021). Rather than using board diversity, independence or any other board indicator, using Management proxy enables us to assess board monitoring functions more comprehensively as they are based on a broad range of corporate governance indicators including board structure, committees and compensation, among others.

To control firm characteristics, we use several commonly used proxies which are potential variables likely to affect firm risk. Among them are CEO duality (CEOpow) [³], board size (Bsize), firm size (Fsize), return on assets (ROA), Leverage, Current ratio (Currentr), Research and development intensity (RDintensity) and free float percentage (FFP) (Al-Hadi et al., 2019; Boubaker et al., 2020).

All the data were retrieved from the Thomson Reuters Eikon database following prior studies (Elmassri et al., 2023; Kuzey et al., 2023; Uyar et al., 2024). All the variables are defined and listed in Table A1.

The formulation of the proposed models is based on panel regression analysis. It was used since there is a time-variant association between the dependent and the independent variables, as well as the research sample, is in a longitudinal structure in which the firm is the panel variable, and the year is the time variable.

Three postestimation tests are performed to decide to select the most appropriate regression analysis among fixed-effects (FE) panel, random-effects (RE) panel and ordinary (pooled) regression analyses. First, the F-test is used to decide between the FE panel and ordinary regression analysis. The results show that (F test: M1:7.21, p < 0.001; M2: 7.18, p < 0.001; M3: 7.21, p < 0.001; M4:7.20, p < 0.001) FE panel regression should be used instead of ordinary regression analysis. Second, the results of the Breusch–Pagan Lagrange Multiplier (LM) test show that (LM test: M1: 28081.07, p < 0.001; M2: 28042.36, p < 0.001; M3: 28091.88, p < 0.001; M4: 28277.96, p < 0.001) RE panel regression should be selected instead of ordinary regression analysis. Finally, the results of Hausman’s test show that (Hausman’s test: M1: 240.47, p < 0.001; M2: 246.46, p < 0.001; M3: 250.44, p < 0.001; M4: 183.87, p < 0.001) FE panel regression analysis should be used instead of the RE panel regression analysis. The labels M1–M4 indicate the model number. Based on these three postestimation analysis results; the FE panel regression should be selected as the most appropriate regression analysis which has significant benefits such as minimizing the multicollinearity risk (Baltagi, 2021) and controlling the omitted variable bias (Wooldridge, 2010). The functional formulation of the proposed models is shown in equation one below:

The dependent variable, y_it, is StdROA. The independent testing and control variables are denoted by X_it: the independent testing variables are ESGs, ENPs, SOPs and GOPs, while the independent control variables are Bsize, Bgdiversity, Bindependence, CEOpow, Fsize, ROA, Leverage, Currentr, RDintensity and FFP. Furthermore, the index “i” represents the panel variable (Firm) while the index “t” represents the time variable (Year). Also, the term “ϑ_it + ∈_it” represents the error term with “ϑ_it” as a firm-specific error term while “∈_it” is a regular error term.

For the regression analyses, the Huber Sandwich Estimator (Huber, 1967; White, 1980) as the robust standard errors are incorporated to control the heteroskedasticity issue (Wooldridge, 2013). Thus, robust standard errors are reported in the analysis.

Before running the further regression analysis, the independent variables of the research models are subject to multicollinearity analysis using variance inflation factor (VIF). The results reveal that the values of VIF range between 1.05 and 1.74 in Model 1; range between 1.04 and 1.67 in Model 2, range between 1.04 and 1.65 in Model 3, range between 1.06 and 1.61 in Model 4. There is no risk of multicollinearity since the values of VIF are significantly less than the cut-off value of 10 (Hair et al., 2019).

The baseline proposed models include two moderation analyses. The moderating role of Management and CEOpow on the relationship between the dependent variable and the independent testing variables is investigated. To test the interaction effects, Hayes’s (2017) moderation analysis methodology is used. The formulation of the moderation analysis is provided in equation two:

The dependent variable, y_it, is StdROA. The independent testing variables (x_1it) are ENPs and SOPs when the moderating variable, M_it, is Management, while the independent testing variables (x_1it) are ESGs, ENPs, SOPs and GOPs when the moderation variable (M_it) is CEOpow. Finally, the control variables (x_2it) are Bsize, Bgdiversity, Bindependence, CEOpow, Fsize, ROA, Leverage, Currentr, RDintensity and FFP. While running the moderating effect for Management, we interacted only with ENPs and SOCs but not GOPs since Management is a dimension of GOPs, and, hence, we avoided overlapping.

The research variables’ summary based on descriptive statistics is presented in Table 2. The mean value of StdROA is 0.02 ± 0.03 ranging from 0 to 0.15 while the mean value of StdTobinq is 0.29 ± 0.48 ranging from 0 to 3.21. Moreover, the average of ESGs is 41.35 ± 20.58, ENPs is 31.86 ± 28.78, SOPs is 41.80 ± 23.50 and GOPs is 48.13 ± 22.74. Finally, 37.40% of the firm-year observations indicate the existence of CEOpow.

The bivariate linear correlation among the research variables is examined using Pearson’s correlation analysis. The correlation coefficients are given in Table 3. The results reveal that ESGs, ENPs, SOPs and GOPs have a significant negative linear correlation with StdROA and StdTobinq (p < 0.01).

The proposed research models formulated in the prior sections are subject to FE panel regression analysis with StdROA as the dependent variable. The results are presented in Table 4. According to the obtained results, the coefficients of ESGs, ENPs, SOPs and GOPs are significantly negative (p < 0.01). Hence, the results confirm the risk-reducing role of CSR (i.e. H1).

The baseline research models incorporated subdimensions of the independent testing variables. These alternative independent testing variables are re-run based on FE panel regression analysis (Table 5) with StdROA as the dependent variable. The results show that the coefficients of Resource, Emissions, Eco-innovation, Workforce, Humanrights, Community, Productresp, Management and CSRstrategy are significantly negative (p < 0.01) while the coefficient of Shareholders is not statistically significant. Thus, H1 is confirmed by nine indicators of ESGs out of 10.

Furthermore, the baseline research models include the moderation role of Management in the relationship of ENPs and SOPs with StdROA. The results in Table 6 reveal that the interaction variables, ENPs × Management and SOPs × Management, have a significant positive relationship with StdROA. Thus, the interaction variables are statistically significant which implies that GOPs is a significant moderator on the relationship of ENP and SOP with StdROA. These results lend support to H2b but not H2a implying there is a substitution effect between board monitoring and environmental and social performance in alleviating firm risk.

The interaction variables are illustrated visually by plotting the simple slopes at −1, 0 and +1 standard deviation of the Management as the moderator variable (Figures 2 and 2). Both Figures 2 and 3 highlight that the slopes are greater at the high level of Management implying that risk is greater.

Also, the baseline proposed models incorporate the moderating role of CEOpow on the relationship of ESGs, ENPs, SOPs and GOPs with StdROA (Table 7). The results indicate that the coefficients of the interaction variables ESGs × CEOpow (p < 0.05), ENPs × CEOpow (p < 0.05) and GOPs × CEOpow (p < 0.01) are significant and positive while the coefficient of the interaction variable SOPs × CEOpow is not significant. Hence, while H3a is accepted, H3b is rejected implying that CEO power weakens the negative link between CSR performance and firm risk.

Similarly, the visual representation of the main interacting variable (ESGs × CEOpow) is illustrated in Figure 4 where the simple slopes are plotted at −1, 0 and +1 standard deviations. The slopes of the lines with CEO power are greater in Figure 4 showing that CSR amplifies firm risk under CEO duality.

Multiple further analyses are applied within the robustness test section to examine the robustness of the results from the baseline analyses. Toward this objective, several additional analyses are performed by incorporating alternative independent testing variables, alternative dependent variables and alternative analysis methods.

First, alternative independent variables (ESGs-ind, ENPs-ind, SOPs-ind and GOPs-ind) are incorporated into the model. These alternative testing variables are industry-adjusted counterparts of ESG pillars presuming that sector-specific conditions may affect CSR performance. The results are shown in Table 8. Accordingly, the coefficients of ESGs-ind, ENPs-ind, SOPs-ind and GOPs-ind are significantly negative. Thus, the robustness check confirms the baseline analysis reported in Table 4 completely.

Second, StdTobinq is selected as an alternative dependent variable to check whether CSR alleviates market risk. The models with StdTobinq as an alternative dependent variable are subject to FE panel regression analysis (Table 9). Based on the obtained results, ESGs (p < 0.01), ENPs (p < 0.01), SOPs (p < 0.01) and GOPs (p < 0.01) have a significant negative relationship with StdTobinq. Hence, the robustness check validates the baseline analysis reported in Table 4.

Third, alternative regression analysis is used for the baseline research models. To control the omitted variable bias as well as the possible risk of endogeneity issues (Arellano and Bond, 1991), a GMM-based dynamic panel regression analysis is used (Table 10). In the analysis, one lag of the StdROA (dependent variable) as an independent variable is included while also it contains unobserved panel-level FE that makes the standard estimators consistent (Arellano and Bond, 1991). The results of the GMM-based dynamic panel regression analysis are consistent when there is a possible risk of endogeneity sources (Wintoki et al., 2012). The results indicate that ESGs (p < 0.01), ENPs (p < 0.01) and SOPs (p < 0.01) have a significant negative relationship with StdROA while GOPs has a nonsignificant relationship with StdROA. Thus, the findings of the robustness check and the baseline analysis converge for ESGs, ENPs and SOPs, they diverge for GOPs.

Overall, the robustness tests prove that the findings are largely robust to industry-adjusted CSR indicators, alternative estimators and alternative firm risk indicators.

Although several prior studies examined whether CSR mitigates financial distress, they focused on a particular setting such as the USA, China and Australia. Besides, they largely ignored contingencies’ role in the relationship between financial distress and CSR performance. Thus, our study extends prior studies by focusing on a worldwide sample and also integrating board monitoring and CEO power into the research model as moderators. We draw the attention of managers and stakeholders to the contingencies’ role in leveraging CSR for deriving better firm outcomes. Finally, the study aims to suggest managerial and policymaking implications.

On one side, while our study confirms prior studies, on the other side it extends them. In the literature, there is no unanimity about the sample, risk proxies and timing of the studies and also contingencies upon which the study is built. For example, Hsu and Chen (2015), Chollet and Sandwidi (2018) and Al-Hadi et al. (2019) conducted their studies in the USA, Australia and international settings respectively and found that CSR helps reduce credit risk, financial distress and market risk respectively. Although those earlier studies provide useful insights, the importance of CSR is gaining momentum around the world and beyond national boundaries. Our study focuses on a longer period including recent years which is critical to suggest timely implications for firms, stakeholders and policymakers. We also extend those prior studies by confirming CSR’s risk-reducing effect building on accounting and market risk proxies. We believe that accounting-based risk metric helps assess firms’ financial sustainability whereas market risk assesses investors’ risks. Thus, our findings indicate that CSR helps firms navigate in the market by lessening the fluctuations in their financial performance, it also impedes market value fluctuations providing shareholders with safe investment opportunities and hedging against risks.

The moderating effect concerning board monitoring aligns with Kim et al. (2014) indicating that CSR is a strong predictor of stock price crash risk when the firm has weaker corporate governance implying the substitutive relationship between CSR and corporate governance in mitigating firm risk. However, there is also counter-evidence in the literature that corporate governance strengthens CSR’s effect in alleviating firm financial distress by supporting complementary relationships (Boubaker et al., 2020). While the substation effect implies one factor (e.g. CSR) reduces the marginal effect of another factor (e.g. board monitoring), the complementary effect means one factor reinforces the marginal effect of another factor in reducing risk (Oh et al., 2018). This perspective might help future studies resolve intricate relationships between firm performance and risk predictors.

On the other hand, with the existence of powerful CEOs, while environmental and governance performance increases firm risk, social performance does not. This finding is in line with agency theory but contradicts stewardship theory. In line with the agency perspective, CEOs with dual roles are assigned multiple tasks with possibly incongruent objectives which may cause sacrifice of some stakeholders’ interests at the expense of others (Hill and Jones, 1992; Jo and Harjoto, 2011). Combining chair and CEO roles might nullify the check and balance system in corporate monitoring leading to inefficient CSR initiatives. The discretionary nature of CSR expenditures might exacerbate the arbitrary and opportunistic behavior of powerful CEOs resulting in greater firm risk.

The study suggests several theoretical and managerial implications. The findings support the proposition of stakeholder theory such that addressing stakeholders’ expectations alleviates agency conflict by mitigating firm risk. The result implies that CSR prevents fluctuation of firm profit and market value over periods. This is of critical importance for investors who seek sustainable investment as well as avoid uncertainties. Overall, the outcome implies that CSR helps firms avoid the perils by doing good. The moderation analysis implies that environmental and social performance substitutes for weak board monitoring. This could be because environmental and social initiatives help firms alleviate some operational, market and regulatory risks (Karwowski and Raulinajtys‐Grzybek, 2021). Furthermore, the moderation effect of CEO power implies that CSR could be a tool for the exploitation of fiduciary duty and firm resources leading to exacerbating firm risk. Thus, CEO power becomes a source of agency costs through CSR rather than enriching stewardship roles. Hence, firms are suggested to weigh the risks and benefits of combining the roles of chair and CEO in a single individual. Taking together, the moderating effects suggest future studies consider contingencies in analyzing CSR’s influence on firm outcomes.

Our study is limited to the availability of mainly CSR data in the data source. This limitation forced us to work on an unbalanced sample across countries and industries. Next, despite our study suggesting reliable and generalizable outcomes drawing on a wide range of sectors and countries, it does not provide sectoral results and implications. Future studies could check the validity of the outcomes for polluting sectors, consumer-oriented sectors and service firms which may yield different outcomes and offer distinct insights and implications. Furthermore, the connection between CSR and firm risk might be contingent upon some other potential organizational and institutional moderators. For example, the institutional environment’s political stability/instability and economic uncertainties might affect the link between CSR and firm risk since instabilities and uncertainties might alter firms’ priorities and risk degrees. Another potential study could be how market competition moderates between CSR and firm outcomes as it imposes pressure on firms. Finally, a specific emerging market study might reveal insightful results as emerging markets have different characteristics such as high growth rates, limited financing and weak institutions.