This study aims to examine the impact of firm-level political risk on ESG controversies.
This study uses a sample of 12,517 firm-year observations from G7 countries between 2002 and 2021. Panel ordinary least squares regression is used to estimate the impact of firm-level political risk on ESG controversies.
This study finds a significant impact of firm-level political risk on ESG controversies, indicating that an increase in firm-level political risk leads to an increase in ESG controversies. The findings support the moral disengagement perspective and the Cressey Fraud Triangle, which explains that negative perceptions due to an adverse political environment lead to unethical firm behaviour. The study further identifies that ESG-linked CEO compensation moderates the underlying relationship. The impact of firm-level political risk is stronger for firms that lack environmental initiatives, high managerial power and low board monitoring.
This study is limited to G7 countries. However, this study provides future research avenues in political risk and ESG controversies, especially in different economic conditions, such as developing countries.
This study offers policy and practical implications for policymakers and top management teams in managing political risks for firms, thus mitigating ESG controversies.
The study advocates for establishing strategies to reduce ESG controversies during high political uncertainty. To the best of the authors’ knowledge, this study is the first to examine the effect of firm-level political risk on ESG controversies.
1. Introduction
Many firms have experienced ESG controversies linked to political risk and exposure in recent decades. Firms have been subjected to intense media coverage due to ESG incidents (Zaman et al., 2020; Jucá et al., 2024; Barkemeyer et al., 2019). For example, in 2016, Samsung’s bribery scandal was enabled by the connection between the management team and the political figures. Samsung’s alleged corruption resulted in loss of trust among its institutional investors, deterioration of public confidence and raised concerns about political influence. In 2018, the New York Times [1] and The Guardian [2] reported a fine of US$5bn charged to Facebook due to the Cambridge Analytica controversy, a data breach associated with political campaigns. British Petroleum had paid approximately US$64bn by 2018 to cover environmental clean-up, compensation and penalties related to the 2010 Deepwater Horizon oil spill. The prolonged legal proceedings from the oil spill resulted in ongoing commitments for British Petroleum to pay for clean-up fees, fines and other related costs, thereby increasing the uncertainty of the company’s future performance (He and Li, 2024). In 2024, German labour courts found that Tesla had violated union election rules related to its plant in Berlin [3]. According to German regulations, there should be a two-year gap between union election campaigns. Tesla had the previous campaign started on 28 February 2022, which means, according to the regulation, Tesla must wait until 29 February 2024 to start the campaign. However, Tesla has called a new election before the completion of the two-year gap, hence violating the regulations on union elections in Germany. Further, in 2024, BBC reported a fine of more than US$220m charged to SAP for their bribery actions from 2014 to 2022 [4]. SAP has engaged in bribing government officials in its subsidiaries operating in Africa, Indonesia and Azerbaijan from 2014 until 2022. Hence, the consequences of these incidents were severe. According to the EY Global Integrity Report 2024 [5], firms in the UK and USA have been fined almost US$1tn since 2010 by regulators and in private litigation. Thus, these real-world incidents and consequences were concerning and required urgent rectification.
Consequently, regulators announce lawsuits and impose penalties on these scandals to discourage similar incidents in the future. Nonetheless, firms are exposed to the volatile political uncertainty, which is unpredictable and challenging (Baker et al., 2016). For instance, governments’ regulatory changes profoundly impact firms (Chatjuthamard et al., 2021). Prior research has explored the geopolitical-level and country-level political influences on firms. For instance, scholars have identified the impact of the macro-level political influences on aspects such as ESG practices (Erzurumlu et al., 2025) and stock returns (Pástor and Veronesi, 2012). However, the findings of these studies are based on the aggregate portfolio level of political risk, which could hinder heterogeneous firm-specific perceived political risk impacts. The danger of relying on perceived macro-level political risk is that it could lead to misleading policy implications and incomplete strategy formulations in firms. In contrast, relying on firm-level political risk would enable the obtaining of the political risk components that are associated with unique factors related to a particular firm (Gupta et al., 2024). Consequently, prior studies have examined the impact of firm-level political risk on several aspects, such as eco innovation (Owolabi et al., 2025), capital investments (Banerjee and Dutta, 2022; Choi et al., 2022; Gyimah et al., 2022; Rahman et al., 2024), liquidity (Das and Yaghoubi, 2022), default risk (Islam et al., 2022), stock price crash risk (Makrychoriti and Pyrgiotakis, 2024) and ESG practices (Chebbi et al., 2024; Safiullah and Kabir, 2024; Wang, 2025). However, whether firm-level political risk leads to ESG scandals remains an empirical question. Hence, despite the real-world evidence on scandals associated with political influences, a lack of comprehension of this nexus could result in adverse consequences for firms, such as inaccurate ESG ratings and scores, misguided investment decisions and loss of social trust.
The detrimental consequences of ESG controversies, such as regulatory fines, reputational damage and investor loss, ignite the need for further empirical examination of the determinants of ESG controversies. “Fraud, in all of its forms, remains a persistent challenge”, states the PwC Global Economic Crime Survey 2024. Media reports on ESG scandals lead to reputational damage, potential litigation costs and regulatory fines for a firm (Walsh et al., 2008). A firm could face public criticism and stigmatisation as negative ESG information is revealed and disseminated to a broad audience (Wiesenfeld et al., 2008). As a result, stakeholders may become reluctant or even hostile towards continuing their business relationship. This situation heightens the uncertainty surrounding the firm’s operations and future performance (Karpoff et al., 2008; Lin et al., 2016). At the same time, ESG controversies create uncertainty about a firm’s prospects, prompting firms to make strategic changes. Furthermore, ESG controversies reduce a firm’s overall investment efficiency, placing a firm at a significant capital risk (Xue et al., 2023).
Against this political and ethical backdrop, we identify the research gap on the firm-level political risk and ESG controversies nexus. We believe that addressing this gap is of timely importance for firms, investors, regulators and the public.
We hypothesise the causal link between political risk and ESG controversies from two theoretical perspectives. Firstly, the moral disengagement theory argues that perceptions about the political environment and risks could significantly impact the motive for unethical organisational behaviour (Bandura et al., 1996; Valle et al., 2019). In the same vein, Schiemann and Tietmeyer (2022) highlight that more ESG controversies can occur when firms experience higher uncertainties. In times of high political uncertainty, firms reduce their irreversible capital investments to divert the surplus funds into more operating and reversible investments and activities (Choi et al., 2022; Banerjee and Dutta, 2022). This is because when investments are irreversible, firms may reduce and delay their capital investments during high political uncertainty until they experience a more stable political environment (Bernanke, 1983). Such capital investment reductions will significantly negatively impact a firm’s financial performance (Santoso, 2019). A financial performance shortfall could increase a firm’s appetite for controversial practices or behaviours (DasGupta, 2022; Chari et al., 2019; Kahneman and Tversky, 1979).
Secondly, the Cressey fraud triangle outlines an explanatory framework with the three conditions: pressure, opportunity and rationalisation that lead to fraudulent behaviour (Cressey, 1953). When an individual, a group or a firm is under pressure, that could be converted into a reason to commit fraud. If they get a chance to commit fraud and find a justification for the adverse behaviour, there is a possibility of fraudulent behaviour [6]. In times of high political uncertainty, a firm’s corporate exposure to political risk creates additional pressure for managers that could threaten their job security. Given that the managers are responsible for a matrix of goals, facing political risk adds another layer of risks that could likely motivate them to pretend that the business activities are unaffected by the exposure to political risk (Hoang et al., 2023). Firms could make their earnings announcements intentionally non-transparent when exposed to extensive political risks. Hence, firms under high political risk attempt to manipulate the earnings information and convey positive signals to the market, resulting in agency costs, mistrust and litigations (Hoang et al., 2023). Although the literature provides evidence on the impacts of political risks on firms and the causes of ESG controversies in isolation (Schiemann and Tietmeyer, 2022), the impact of firm-level political risk on ESG controversies remains an empirical question. We investigate whether firm-level political risk leads to a rise in a firm’s ESG controversies in a cross-country setting.
From the behavioural and institutional theoretical perspectives, the mechanistic pathway for the relationship between political risk and ESG controversies can be justified as follows. In line with behavioural theory, a firm’s subjective perception of political risk influences its decisions regarding how to respond to political risk, potentially resulting in either engagement or disengagement (Giambona et al., 2017). The institutional theory emphasises the role of institutional contexts in shaping corporate governance and ethical decision-making, with political institutional systems influencing firm behaviour (González and García-Meca, 2013; Chan and Ananthram, 2018; Selznick, 1948). Various institutional frameworks, such as National Business Systems (NBS), impact the relationship between political risk and ESG controversies, as governance inefficiencies and weak legal structures in different political institutional contexts can contribute to unethical practices (Nielsen and Massa, 2012; Heim and Mergaliyeva, 2024; Jamali et al., 2020).
G7 provides an interesting setting for this study. G7 comprises the world’s most developed and influential political and economic giants (Bloom, 2009). These economies account for 10% of the population, 40% of global gross domestic product (GDP) and 25% of carbon emissions (Sun et al., 2022) compared to the world. Furthermore, all the countries in the G7 share similar dominant economic powers and democratic characteristics (Nini et al., 2012; World Bank, 2021). Therefore, the G7 is a strong and influential group ideal as a political, economic and environmental research sample. The group has identified their vital need to commit to climate change and economic and political uncertainties. Although the group actively works on global political, environmental and economic challenges (European Council, 2021), these countries constantly face issues related to climate change, tariffs and regulations. Hence, G7 countries create an ideal setting to empirically examine associations related to political risk and ESG controversies (Rungmaitree et al., 2022).
Using 12,517 firm-year observations from 2002 to 2021, we identify a significant positive impact of firm-level political risk on ESG controversies after controlling for macroeconomic and firm-level factors. Our cross-sectional analyses reveal that the positive impact of political risk on ESG controversies is common among US and non-US G7 firms. The effect of political risk on ESG controversies is strong when the firms lack environmental initiatives. The underlying effect is prominent in the presence of high managerial power and the absence of effective board monitoring. Our findings are justified for robustness by using an alternative measure of ESG controversy score, controlling for firm fixed effects and using the leading values of the dependent variable. We use the entropy matching procedure to alleviate possible endogeneity issues from self-selection bias. Our results remain robust after considering the impact of the global financial crisis (GFC) and the Paris Agreement as exogenous shocks.
Our study makes four empirical and strategic contributions. Firstly, this paper extends the literature of the ESG domain as follows. Due to investors’ concerns about ESG controversies, there have been crucial academic discussions on the effectiveness of ESG practices and the credibility of ESG disclosures. The existing literature on ESG controversies documents associations of ESG controversies with firm value (Aouadi and Marsat, 2018), governance (Agnese et al., 2023) and investment (Xue et al., 2023). Hence, our study extends the ESG controversies literature as this is the first study to explore the association of ESG controversies with firm-level political risk and identify firm-level political risk as a driver of ESG controversies.
Secondly, we extend the literature on firm-level political risk. We use the Hassan et al. (2019) measure that captures political risks at the firm level. Previous related studies examine how political risk influences areas such as CSR (Chatjuthamard et al., 2021; Hasan and Jiang, 2023; Wang, 2025), investments (Banerjee and Dutta, 2022; Choi et al., 2022), tax avoidance (Hossain et al., 2022) and earnings opacity (Hoang et al., 2023). We advance this growing literature by exploring a novel relationship between firm-level political risk and ESG controversies of a firm, highlighting the importance of firm-level political uncertainty on various stakeholders of firms. Studies closely related to this study are Erzurumlu et al. (2025) and Chebbi et al. (2024). The former has investigated geopolitical and country-level political risk, and the latter has examined firm-level political risk. However, both explore the impact of political risk on ESG practices, unlike ours. Whereas we emphasise the direction of ESG controversies by extending the strand of literature. The studies focused on macro-level political risk could lead to incomplete or misleading policy implications, as the levels and extents to which firms perceive macro-level political risks are distinct (Hassan et al., 2019). Thus, generalising the findings of studies focused on macro-level political risk could be questionable. In contrast, by focusing on firm-level political risk, this study creates a sharper intellectual tension and provides novel and value-adding evidence to the broader political risk–ESG nexus.
Thirdly, we provide cross-country evidence from the G7 group, which represents one of the most influential groups of nations. The G7 countries have increased the renewable energy share in the energy supply grid and led the energy transition to achieve the net-zero carbon emission target by 2050 (Khan and Su, 2022; Usman, 2022). As political stability is proven to influence carbon emissions (Benlemlih et al., 2022), the findings of our study offer significant insights into the G7 strategy reforms for the reduction of political risk and the mitigation of controversies. Hence, our study contributes towards the G7’s willingness to address critical environmental issues and enhance stability.
Finally, we make a theoretical contribution to the Cressey fraud triangle by examining the impact of firm-level political risk on ESG controversies, referring to the pressure component identified in the Cressey fraud triangle. Not many prior studies are built upon the Cressey hypothesis. For instance, Homer (2019) has systematically reviewed the fraud triangle literature, while Schuchter and Levi (2013) found that a firm’s corporate culture significantly influences the fraud triangle elements. Cheliatsidou et al. (2021) have proposed a fraud triangle model framework for an international setting. Chen et al. (2024) have identified how climate uncertainty could lead to corporate fraud, referring to the Cressey hypothesis. Nevertheless, research into ESG controversies inspired by the Cressey hypothesis is scant. Hence, our findings contribute to the literature built upon the Cressey fraud triangle and address a relationship of current interest.
The organisation of the paper is as follows. We present a review of existing literature and develop our hypothesis in Section 2. Sections 3 and 4 present the methodology and the empirical results, respectively. Finally, we conclude the paper in Section 5.
2. Background and hypothesis development
2.1 Firm-level political risk
Generally, political risk is defined as the risks caused to a firm’s operations and decisions due to changes and instability in the political environment. The type and magnitude of political risk can vary from one country to another, while political risk can arise from several avenues (Mariadoss, 2019). Political risk can stem from government policy changes or legitimate actions of governments to control prices, currencies, taxes and tariffs (Mariadoss, 2019). Further, political risk arises from events outside of government controls, such as labour strikes, revolutions and terrorism (Mariadoss, 2019). The uncertainty stemming from the macro-level political environment has unavoidable consequences in terms of economic and financial statuses of countries, such as financial crises, high stock returns volatility (Mei and Guo, 2004) and a reduction in foreign direct investments (Luo, 2009). Apart from these economic consequences, political stability can have asymmetric effects on a country’s carbon emissions. For example, prior literature identifies political stability as an indicator in examining carbon emissions (Ayhan et al., 2023). Furthermore, geopolitical tensions can have direct adverse impacts on cross-border trade[7]. It can be argued that the impact of political risk can be drawn at the macro level. However, political risk measured at the macro level has a fundamental shortfall as it assumes that the effects of political risk on firms are homogeneous. Since not all political changes affect the entire economy equally, the political risk is not equally distributed across businesses (Hoang et al., 2023).
Further, the macro-level political risk affects firms’ decision-making and business outcomes differently, and the levels and extents to which firms perceive macro-level political risks are distinct (Hassan et al., 2019). The variation in aggregate political risk over time and across sectors accounts for 0.81% and 7.5% of the total variation, and the rest of the 91.69% occurs at the firm level. (Hassan et al., 2019).
The political risk at the firm level is considerably more helpful in explaining firms’ business operations when firms are under political uncertainties (Hoang et al., 2023). Prior literature (Owolabi et al., 2025) evidence that firms navigate the political uncertainties to advance sustainable initiatives such as eco-innovation. In a similar vein, firms increase their ESG performance to benefit from the “insurance-like” effect of ESG during high political uncertainties (Chatjuthamard et al., 2021). Conversely, Wang (2025) argues that firm-level political risk negatively impacts a firm’s CSR activities, especially when firms face financial difficulties. Further, it is reported that political uncertainties increase firms’ stock premiums and the cost of capital (Brandon and Yook, 2012; Gungoraydinoglu et al., 2017). Furthermore, high political risk reduces corporate capital investments (Choi et al., 2022). The extant literature further emphasises that political risk could hinder corporate investments as, according to irreversible investment theory, firms decide to delay irreversible investments during uncertainties. (Bernanke, 1983; Bloom et al., 2007; Bloom, 2009; Brandon and Yook, 2012; Choi et al., 2022). Consequently, firms would prefer equity issuance to finance the investments during such time periods (Rahman et al., 2024).
Further, firms find it challenging to forecast future cash flows as high firm-level political risk increases the information asymmetry in the market (Gyimah et al., 2022). Concurrently, increased political risk can accelerate a firm’s default risk (Nini et al., 2012) and increase a firm’s stock price crash risk (Makrychoriti and Pyrgiotakis, 2024). Political risk can have significant impacts on organisational behaviour. For instance, Hoang et al. (2023) identify that high firm-level political risk increases corporate earnings opacity. According to the agency perspective, when managers feel threatened due to political risks, they get motivated to behave in a self-serving manner to protect their benefits. When firms are exposed to high political risk, manipulative behaviours could occur to create positive signals to the market. Hence, according to the stakeholder perspective, the additional pressure the managers experience from political risks would motivate them towards speculative or manipulative behaviours. Concurrently, Hossain et al. (2022) document that firm-level political risk raises the propensity to avoid taxes by managers. However, it is difficult for firms to fully diversify their exposure to political risk due to the exogeneity of political risks to firms (Hoang et al., 2023).
Against this backdrop, it is evident that political risk puts pressure on several business aspects. According to the fraud triangle theory, pressure is one of the three fraud determinants. Hence, it is important to investigate whether firm-level political risk could be a component of pressure on the occurrence of ESG controversies in a firm in the context of growing demand for ESG-related investments.
2.2 ESG controversies
ESG controversy is a scandal, dispute or unethical behaviour occurring in a firm in the form of action or incident (Brighi et al., 2022). Such inappropriate environmental scandals and social behaviours of firms attract the attention of investors and the media (Jucá et al., 2024). These damaging actions adversely impact stakeholders and, more importantly, a firm’s corporate image. Previous research has established that controversies and scandals negatively impact a firm’s reputation (Walsh et al., 2009). For instance, based on analysing 132 cases of US-based corporate fraud, Karpoff and Lott (1993) revealed that firms face significant reputational losses compared to legal sanctions after being accused of corporate fraud. Similarly, ESG controversies account for corporate ESG news stories such as unethical social behaviour, product and policy scandals and environmental harm, which can put firms under media scrutiny and investor attention (Cai et al., 2011). Greater media scrutiny of ESG controversies results in higher stakeholder scepticism, triggers perceptions of corporate hypocrisy and leads to lower credibility (Du et al., 2010; Godfrey et al., 2008). In addition, ESG controversies could challenge organisational legitimacy, which could negatively affect the long-term prosperity of firms (Palazzo and Scherer, 2006).
Moreover, ESG controversies adversely affect the firm value (Aouadi and Marsat, 2018). A firm’s value declines following the news of a misrepresentation or fraud committed by the firm. Thus, following controversies, the highest penalties for firms are imposed by the market, not regulators (Karpoff et al., 2008). Concurrently, ESG controversies adversely impact a firm’s sustainable investment (Xue et al., 2023) and analyst forecast accuracy (Schiemann and Tietmeyer, 2022). Furthermore, ESG controversies could negatively impact a firm’s market share and positively impact total and idiosyncratic risk (Brighi et al., 2022). For instance, De Franco (2019) identified that ESG controversies could deteriorate stock returns in European and US markets. Concurrently, based on a study conducted in European listed firms, Nirino et al. (2020) argue that corporate controversies negatively influence the financial performance of firms. In a similar vein, based on a study of a sample of European non-financial listed firms, Elamer and Boulhaga (2024) reveal that the adverse impact of ESG controversies on corporate firm performance could be mitigated through good corporate governance. Contributing to the controversies-financial performance literature nexus, Jucá et al. (2024) further confirm the negative impact of ESG controversies on financial performance in environmentally sensitive industries and in developed countries. On the one hand, one major impediment to ESG performance is the controversy, which is like “doing good but not well” (Barnea and Rubin, 2010; Dorfleitner et al., 2020). On the other hand, studies find that firms would increase their disclosure of ESG information following an ESG controversy. These firms are expected to reveal more details about the controversies and their plans to resolve them to regain legitimacy (Schiemann and Tietmeyer, 2022). This phenomenon has the danger of firms engaging in greenwashing with the aim of impression management. For example, Fathoni et al. (2025) find that ESG controversies drive a firm’s intention on ESG washing. Besides, it is critical to understand the determinants of ESG controversies, given the increased investor attention on ESG issues and ESG disclosures to evaluate the impacts of ESG controversies (Amel‐Zadeh and Serafeim, 2018; Du et al., 2011; Maignan and Ferrell, 2004). Only a handful of studies have examined the determinants. For example, in relation to internal governance, Muhammad and Tadele (2025) argue that firms with larger boards experience more ESG controversies due to the potential challenges in consensus decision-making. Nevertheless, it is identified that a higher number of female directors on boards reduces the probability of ESG controversies. When external governance is considered, the takeover market’s effect (the potential for a firm to be acquired) acts as an external discipline of firms to prevent ESG controversies from occurring (Treepongkaruna et al., 2024).
2.3 Firm-level political risk and ESG controversies
We refer to the moral disengagement theory and the Cressey fraud triangle theory to establish the link between firm-level political risk and ESG controversies. Firstly, the moral disengagement theory (Bandura et al., 1996) identifies eight interrelated cognitive mechanisms that enable a person or a group of people to ignore moral standards and behave immorally without feeling guilty (Moore, 2015). The moral disengagement concept explains the potential impact of firm-level political risk on ESG controversies. When firms are exposed to high political risk, moral disengagement arising from the perceptions of the adverse political environment can result in unethical firm behaviour (Valle et al., 2019). Similarly, in the context of organisational corruption, there are instances where firms tend to engage in unethical conduct without apparent distress (Bandura et al., 1996; Moore, 2007). Moral disengagement may urge firms to take unethical actions in the business’s interest. Furthermore, according to Moore (2007), moral disengagement may ease managers’ unethical decision-making by preventing psychological discomfort experienced by managers at the prospect of making a business-friendly but morally challenged decision. Empirical evidence supports this view. During high political risk, firms reduce their irreversible capital investments to divert the surplus funds into more operating and reversible investments and activities (Choi et al., 2022; Banerjee and Dutta, 2022). This is because when the investments are irreversible and the environment is volatile, firms reduce and delay their capital investments until they gain some certainty about the political environment, leading to a financial performance shortfall (Bernanke, 1983; Santoso, 2019). Consequently, a financial performance shortfall increases a firm’s appetite for controversial or unethical practices or behaviours (Chari et al., 2019; DasGupta, 2022; Kahneman and Tversky, 1979).
Secondly, the fraud triangle theory by Cressey (1953) outlines three elements that influence corporate fraudulent activities: pressure, opportunity and rationalisation (Chen et al., 2024). When a firm is under pressure, it could be converted into a reason to commit fraud, get a chance and find a justification for the adverse behaviour. Then, there is a possibility of fraudulent behaviour. Empirically, scholars have used the fraud triangle framework in fraud and business ethics literature across different subjects, industries and countries (Ghafoor et al., 2018; Homer, 2019). Concurrently, prior literature has identified financial distress (Dellaportas, 2013) and regulatory impacts (Kedia and Rajgopal, 2011) as a few of the pressures that could lead to committing fraud. In a similar vein, prior literature has identified poor corporate governance (Hasnan et al., 2012) and poor audit mechanisms (Hasnan et al., 2012; Power, 2012) as leading opportunity factors to the probability of committing fraud, while the culture of fraud and ethical judgement can be factors that provide rationalisation to commit fraud (Kula et al., 2011; Kwarteng and Aveh, 2018; Namazi and Rajabdorri, 2019). Furthermore, empirical studies provide mixed evidence on applying the three elements to the tendency to commit fraud. For instance, based on the evidence from Statement on Auditing Standard (SAS) No. 99, Suyanto (2009) finds that pressure and opportunity are significantly related to the motive for committing fraud, while rationalisation is less related. In contrast, Ghafoor et al. (2018) identified the effects of all three factors on the probability of fraud. Nevertheless, Villoria et al. (2013) and Dass et al. (2016) observe that political corruption increases social and institutional mistrust and firms’ willingness to violate regulations. Hence, we predict that political pressure makes firms facing high political risk more prone to ESG controversies. Empirical evidence supports this justification. During periods of significant political instability, a company’s exposure to political risk can increase the pressure on managers, potentially jeopardising their job security. Managers tasked with achieving multiple objectives may be inclined to downplay the impact of political risk on business operations. This situation can lead firms to obscure their earnings reports when facing substantial political risks. Consequently, companies under high political risk may manipulate earnings information to present a more favourable image to the market, resulting in agency costs, mistrust and potential legal issues (Hoang et al., 2023).
Nonetheless, signalling theory offers a contradictory view that when firms are exposed to high political risk, firms would increase their ESG performance as a risk buffer to take advantage of the “insurance-like effect” of the ESG (Godfrey et al., 2008; Chatjuthamard et al., 2021; Chebbi et al., 2024). According to signalling theory, this increased ESG performance helps firms signal to investors and other stakeholders on their enhanced risk management and governance qualities (Godfrey et al., 2008; Erzurumlu et al., 2025; Harjoto and Wang, 2024) and their commitment to stakeholder interests despite the negative economic and political influences.
We further rely on the behavioural theory and institutional theory to explore the mechanistic pathway through which the relationship between political risk and ESG controversies is established. The behavioural theory views that a firm’s subjective perception of political risk affects a firm’s decision on how to react to the political risk (Giambona et al., 2017), whether it could lead to a disengagement or not. That is, if firms feel threatened by an adverse political influence, there is the probability that it could lead to a disengagement resulting in an unethical firm reaction. In contrast, it could lead to firms exercising risk aversion, avoidance or reduction techniques. Consequently, firms that are under high political and economic uncertainties would increase their ESG performance to benefit from the “insurance-like” effect of ESG (Chatjuthamard et al., 2021) and would produce extensive ESG disclosures to signal their commitments towards stakeholders (Harjoto and Wang, 2024) as a strategic means of risk management.
Additionally, institutional theory highlights the crucial role of institutional contexts in shaping corporate governance (González and García-Meca, 2013) and ethical decision-making (Chan and Ananthram, 2018) within firms. Selznick (1948) pioneered the idea that a firm’s external environment, including political institutional systems, can significantly influence its decisions and actions. Institutional forces play a key role in ethical business practices, with different political systems exhibiting distinct institutional logics and ethical challenges (Nielsen and Massa, 2012). Prior research emphasises the importance of understanding firms within their embedded institutional frameworks (Jamali et al., 2020), as various characteristics of institutional contexts, such as government inefficiency and weak rule of law, can contribute to rising unethical practices in firms (Heim and Mergaliyeva, 2024). Different institutional frameworks, such as NBS, can shape the relationship between political risk and ESG controversies in diverse ways (Nielsen and Massa, 2012). Appendix 2 provides a detailed description of the NBS framework. Studies have shown that poor corporate governance is a contributing factor to ESG controversies (Agnese et al., 2023). For example, taking different economies identified in the NBS framework into consideration, firms in emerging economies, socialist economies and Arab oil-based economies tend to exhibit weak or average corporate governance norms, making them more susceptible to unethical business practices under adverse political influences (Witt et al., 2017; Zaman et al., 2020). In contrast, European peripheral economies have moderately strong corporate governance, while highly coordinated economies tend to have governance structures dominated by internal stakeholders (Witt et al., 2017; Zaman et al., 2020). Liberal and coordinated market economies generally adhere to corporate governance norms centred on shareholder wealth maximisation, whereas firms in advanced emerging economies benefit from well-defined governance frameworks. Firms in advanced city economies maintain superior governance strategies, reducing their likelihood of experiencing ESG scandals (Witt et al., 2017; Zaman et al., 2020). Thus, national institutional contexts play a key role in understanding the basis for a nexus between political risk and ESG controversies. Strong governance norms can act as a buffer, reducing the negative effects of political risk on ESG issues, while weak governance mechanisms can amplify this nexus, making firms more vulnerable to unethical business practices. Recognising these dynamics is essential for understanding the broader implications of institutional theory in shaping the mechanistic pathways for the impact of political risk on ESG controversies. The theoretical framework is depicted in Figure 1.
The model presents firm level political risk on the left connected by a right pointing arrow to E S G controversies on the right. Above this relationship, a box labelled moral disengagement theory and Cressey fraud triangle spans the model, with directional arrows indicating its overarching theoretical framing. Below the relationship, a box labelled behavioural theory and institutional theory provides an additional theoretical foundation. Together, the elements show how firm level political risk relates to E S G controversies within these complementary theoretical perspectives.Theoretical framework
Note(s):Figure 1 depicts the theoretical framework, which demonstrates the theories that define the impact of firm-level political risk on ESG controversies and the theories that ground the mechanistic pathway to a relationship between firm-level political risk on ESG controversies
Source: Authors’ own work
The model presents firm level political risk on the left connected by a right pointing arrow to E S G controversies on the right. Above this relationship, a box labelled moral disengagement theory and Cressey fraud triangle spans the model, with directional arrows indicating its overarching theoretical framing. Below the relationship, a box labelled behavioural theory and institutional theory provides an additional theoretical foundation. Together, the elements show how firm level political risk relates to E S G controversies within these complementary theoretical perspectives.Theoretical framework
Note(s):Figure 1 depicts the theoretical framework, which demonstrates the theories that define the impact of firm-level political risk on ESG controversies and the theories that ground the mechanistic pathway to a relationship between firm-level political risk on ESG controversies
Source: Authors’ own work
We propose the following testable hypothesis based on the above theoretical perspectives and empirical evidence:
Firm-level political risk increases firms’ ESG controversies.
2.4 ESG-linked CEO compensation on firm-level political risk and ESG controversies
Firms incorporate ESG metrics into executive compensation to improve firm value and sustainability strategies in the long term (Yin, 2024). Concurrently, firms aim to monitor the risk management mechanisms and gain accountability of top executives on sustainability issues through linking their compensation to ESG metrics (Yin, 2024). According to agency theory, if investors and other stakeholders demand ESG, linking CEO compensation with ESG targets will improve ESG outcomes of firms (Homroy et al., 2023) and reduce agency issues. In a similar vein, Cohen et al. (2023) and Tsang et al. (2021) argue that ESG-linked CEO compensation leads to improved ESG performance. In contrast, most firms cut back ESG targets in CEO compensation following ESG controversies, while firms in consumer-sensitive industries such as apparel, entertainment and luxury products continue with an ESG-linked CEO compensation scheme even after controversies (Kuang et al., 2025).
Furthermore, the evidence from prior literature about the effect of ESG-linked CEO compensation on financial performance is inconclusive (Derrien et al., 2021; Gillan et al., 2021). On the one hand, ESG-linked CEO compensation would positively impact long-term financial performance as ESG targets will incentivise CEOs to address long-term ESG risks, such as climate change risks. On the other hand, ESG-linked CEO compensation would be at the cost of shareholder wealth in the short term as ESG investments could reduce short-term profits (Derrien et al., 2021; Gillan et al., 2021).
Prior literature (Di Giuli and Kostovetsky, 2014) argue that a firm’s political environment and compensation practices can have impacts on its ESG performance. A firm’s decision to integrate ESG incentives with CEO compensation will vary according to the political environment. For instance, Peng and Smith (2023) find that US firms located in democratic states are more likely to integrate ESG incentives into CEO compensation. Concurrently, the ESG-linked CEO compensation strategy would promote and enhance ESG performance, hence in turn reducing ESG controversies (Derchi et al., 2020). Building on the findings of this prior literature, we investigate whether ESG-linked CEO compensation is a moderating channel through which firm-level political risk has an impact on ESG controversies. Hence, we propose the following hypothesis:
The impact of firm-level political risk on ESG controversies is moderated by ESG-linked CEO compensation.
3. Sample and methodology
3.1 Sample description and data sources
Our sample comprises 12,517 firm-year observations in G7 countries from 2002 to 2021. We obtained firm-level political risk data from Hassan et al. (2019)[8]. We source ESG controversy scores from the Refinitiv DataStream database. We also use the Refinitiv Eikon DataStream to source firm-level financial data and the World Bank database [9][10] to source macro-level economic and governance data. We adopt Economic Policy Uncertainty (EPU) data from the EPU index by Baker et al. (2016). The independent and control variables are winsorised at the 1st and 99th percentiles to mitigate the outliers. Figure 2 demonstrates the sample distribution over the years compared with the average firm-level political risk. The sample proportion has gradually increased over the years, notably after 2015. The average firm-level political risk shows a fluctuating trend over time. A more drastic increase occurred in 2020, suggesting a widespread shock, possibly due to social and political pressures related to the COVID-19 outbreak.
The chart plots years from 2002 to 2021 on the horizontal axis. The left vertical axis shows number of firms, and the right vertical axis shows average political risk. The number of firms increases overall from about 150 in 2002 to about 1180 in 2021, with a decline around 2013 to 2014 before rising sharply after 2015. Average political risk fluctuates across the period, starting around 0.8 in 2002, peaking near 1.0 in 2012, declining to about 0.8 in 2015, rising to about 1.25 in 2020, and decreasing to around 0.8 in 2021.Sample composition and the average firm-level political risk
Note(s):Figure 2 demonstrates the yearly distribution (frequency) of the firm-year observations and the distribution of average firm-level political risk over the sample period
Source: Authors’ own work
The chart plots years from 2002 to 2021 on the horizontal axis. The left vertical axis shows number of firms, and the right vertical axis shows average political risk. The number of firms increases overall from about 150 in 2002 to about 1180 in 2021, with a decline around 2013 to 2014 before rising sharply after 2015. Average political risk fluctuates across the period, starting around 0.8 in 2002, peaking near 1.0 in 2012, declining to about 0.8 in 2015, rising to about 1.25 in 2020, and decreasing to around 0.8 in 2021.Sample composition and the average firm-level political risk
Note(s):Figure 2 demonstrates the yearly distribution (frequency) of the firm-year observations and the distribution of average firm-level political risk over the sample period
Source: Authors’ own work
Table 1 depicts the country distribution for 12,517 observations and the percentage of firms under each country in the sample. Also, from Table 1, a US dominance in the sample is evident, with US firms accounting for 63.92% of the sample. France, Italy and Germany exhibit relatively high ESGC (19.45, 19.24 and 18.88, respectively) and high PRisk (0.93, 1.08 and 1.06, respectively), suggesting that an unstable political environment may trigger high controversies.
Sample distribution
| Country | Frequency | % | Average ESGC | Average political risk |
|---|---|---|---|---|
| Canada | 650 | 5.19 | 7.73 | 0.86 |
| Germany | 816 | 6.52 | 18.88 | 1.06 |
| France | 567 | 4.53 | 19.45 | 0.93 |
| UK | 1,054 | 8.42 | 17.16 | 1.07 |
| Italy | 139 | 1.11 | 19.24 | 1.08 |
| Japan | 1,092 | 8.72 | 11.84 | 0.75 |
| US | 8,199 | 65.5 | 11.68 | 0.87 |
| Full sample | 12,517 | 100 |
| Country | Frequency | % | Average | Average political risk |
|---|---|---|---|---|
| Canada | 650 | 5.19 | 7.73 | 0.86 |
| Germany | 816 | 6.52 | 18.88 | 1.06 |
| France | 567 | 4.53 | 19.45 | 0.93 |
| 1,054 | 8.42 | 17.16 | 1.07 | |
| Italy | 139 | 1.11 | 19.24 | 1.08 |
| Japan | 1,092 | 8.72 | 11.84 | 0.75 |
| 8,199 | 65.5 | 11.68 | 0.87 | |
| Full sample | 12,517 | 100 |
This table exhibits the sample, ESG controversy score and firm-level political risk distribution across the countries of the G7 group
3.2 Dependent variable
The dependent variable of this study is the ESG controversy score. Refinitiv ESG controversy score is computed based on 23 ESG controversy topics, with the default value of the measure being 0. The score of the firms with “no controversies” is 100. These controversy topics include controversies related to anti-competition, business ethics, intellectual property, critical and undemocratic countries, public health, tax fraud, child labour, human rights, management compensation, consumer complaints, customer health and safety, privacy, product access, responsible marketing, research and development, environmental impact, accounting issues, insider dealings, shareholder rights, workforce diversity, employee health and safety, working conditions and strikes. Refinitiv defines the ESG controversy score as a metric of a firm’s vulnerability to negative incidents and ESG controversies that are reported in global media. If an incident happens over the course of the year, the firm involved is subject to a penalty or punishment, and this will affect the ESG controversy score. The impact of the incident may persist into the following year if further legal actions or financial penalties develop in connection with the incident [11]. We rely on the Refinitiv ESGC score for several reasons − the ESG data provider’s reliability, the calculation’s user-friendliness and the data’s availability in quantitative terms. Furthermore, the Refinitiv ESG controversy score uses the concept of severity weighting to address the market cap bias from which large firms suffer due to the more media attention they would get compared to small firms. Appendix 3 shows the market cap grouping used for the score calculation [12]. We follow the reverse decimal basis method used by Carbó-Valverde and Cuadros-Solas (2023) to rearrange the score values as “100 – Refinitiv ESG controversy score value” and express the ESG controversy values for a more convenient interpretation. This transformation allows us to interpret the findings more conveniently and less confusingly to the readers. After this transformation, an increase in the controversy score implies an increase in the controversies of a firm.
3.3 Independent variable
We adopt the firm-level political risk measure of Hassan et al. (2019) for this study. Firm-level political risk is constructed using textual analysis of quarterly earnings conference-call transcripts. Hassan et al. (2019) count the occurrence of bigrams related to the discussion of a political topic within the set of 10 words associated with synonyms for the words “risk” or “uncertainty” on either side and divide by the number of bigrams in the transcript.
Here, [·] is the indicator function, while P\N is the set of bigrams contained in P but not N. The r stands for the position of the nearest synonym for uncertainty or risk. The first two terms in the numerator count the number of bigrams associated with discussing political but not nonpolitical topics that occur in proximity to a synonym for risk or uncertainty (within 10 words). They also weigh each bigram with a score that reflects how strongly the bigram is associated with discussing political topics (the third term in the numerator) for the standard specification. fb,P is the frequency of bigram b in the political training library, and BP is the total number of bigrams in the political training library. The weighted sum of bigrams associated with political (rather than nonpolitical) text used in conjunction with synonyms for risk or uncertainty is the overall measure of the share of the conversation devoted to risk associated with political topics (Hassan et al., 2019). This measure is based on the discussions at conference calls, which focus on political environment risks and uncertainties. We aggregate firm-year-level average political risk from the quarterly political risk scores. Following Hassan et al. (2019), we standardise political risk using its standard deviation. So, a higher PRisk score indicates a higher firm-level political risk.
3.4 Control variables
We include a set of control variables that may affect ESG controversies (Hasan and Jiang, 2023; Kuang et al., 2025; Schiemann and Tietmeyer, 2022). The firm-level control variables are market-to-book ratio, financial leverage, R&D expenses, ROA, capex, cash volatility, firm size, board size and stock volatility. The macro-level control variables are GDP, inflation, unemployment, control of corruption, government effectiveness, regulatory quality and EPU (economic policy uncertainty). All these variables are explained in Appendix 1. We control for country, industry and year-fixed effects to capture unobserved heterogeneity.
3.5 Econometric model
We estimate the following baseline regression specification to examine the impact of firm-level political risk on ESG controversies:
ESGC is the inverted Refinitiv controversy score, and PRisk is the firm-level political risk introduced in Sections 3.2.1 and 3.2.2, respectively. We control for firm-level and macro-level factors demonstrated in Section 3.2.3. The coefficient, β1, is the parameter of interest to test H1, expecting a positive and significant value.
Further, we estimate the regression specification in equation (3) to examine the moderating effect of ESG-linked CEO compensation on the impact of firm-level political risk on ESG controversies:
Here, the β2 is the parameter of interest that tests H2, expecting either an increase or a decrease in β1 after accumulating the moderating effect of ESG-linked CEO compensation.
4. Empirical results
4.1 Summary statistics and correlations
Table 2 exhibits the descriptive statistics of ESGC, PRisk and the control variables. The mean value for the ESGC is 12.85. As per the statistics, 75% of firm-year observations have an ESGC of 0, which means no controversies exist. The mean PRisk is 0.89, which implies a high level of political risk exposure, considering the median value is 0.63. Our sample’s mean logarithm of firm market capitalisation (SIZE) is 15.54. The average ROA of 2.3% indicates that a typical firm in our sample generates profits. The financial leverage of 24% implies that an average firm uses debt capital to finance about a quarter of its asset base. The average market-to-book ratio of 2.2% demonstrates that investors value a typical firm in our sample more than its book value. Furthermore, an average capital expenditure to total assets ratio of 3.8% means that firms in the sample generate the cash required for their capital investments. At the same time, an average cash flow volatility of 0.05 indicates that a typical firm has a low risk based on the operating cash flow of the particular firm. The average ratio of R&D expenditure to firms’ total assets is 5.6%. Regarding board governance, a typical board in our sample has ten members, translating into an average logarithm of 2. The average stock volatility of a firm in the sample was reported as 5.05. Turning into macro-economic, social and governance characteristics, the average yearly GDP per capita growth is 2.28% over the 20 years in G7 countries, while average unemployment is 3% of the total workforce. The average annual inflation rate over the sample period is around 17.5%. As far as the world governance indicators are concerned, the Regulatory Quality estimate (REGQ) has a higher average score of 5.85 compared to the Control of Corruption estimate (CoC) and Government Effectiveness estimate (GOVEF) of 0.99 and 1.82. The mean EPU score of the sample is 1.4, suggesting that firms are exposed to marginally high EPU, considering the median value is 1.34.
Summary statistics
| Variable | Mean | Median | SD | Min | Max |
|---|---|---|---|---|---|
| ESG controversy score (ESGC) | 12.8550 | 0.0000 | 26.4586 | 0.0000 | 99.4900 |
| Firm-level political risk (PRisk) | 0.8906 | 0.6334 | 0.8822 | 0.0000 | 7.2726 |
| Market-to-book ratio (MTB) | 2.2010 | 1.5374 | 2.0534 | 0.2546 | 13.9320 |
| Financial leverage (LEV) | 0.2402 | 0.2300 | 0.1793 | 0.0000 | 0.8519 |
| Research and development (RD) | 0.0567 | 0.0245 | 0.0858 | 0.0000 | 0.5388 |
| Return on assets (ROA) | 0.0228 | 0.0466 | 0.1426 | −0.7797 | 0.3227 |
| Capital expenditure (CAPX) | 3.8495 | 2.9852 | 3.2611 | 0.0257 | 33.7268 |
| Firm size (SIZE) | 15.5475 | 15.5148 | 1.6532 | 11.4140 | 19.4316 |
| Cashflow volatility (CashVOLT) | 0.0520 | 0.0295 | 0.0816 | 0.0022 | 0.6861 |
| Stock volatility (StockVOLT) | 5.0542 | 5.0226 | 0.4231 | 4.0967 | 6.2967 |
| Natural logarithm of board size (BSIZE) | 1.4853 | 1.5138 | 0.1750 | 0.1916 | 1.9007 |
| Gross domestic product (GDP) | 2.2825 | 2.3026 | 0.2965 | 1.0986 | 3.2958 |
| Inflation (INF) | 17.4908 | 16.6700 | 12.2121 | 0.0000 | 56.2500 |
| Unemployment (UNEMP) | 3.0891 | 0.9498 | 22.7907 | 0.1646 | 846.1912 |
| Control of corruption (CoC) | 0.9995 | 1.5404 | 2.7212 | −11.3557 | 7.6857 |
| Government effectiveness (GOVEF) | 1.8207 | 1.7939 | 1.2787 | −2.3140 | 8.1533 |
| Regulatory quality (REGQ) | 5.8514 | 5.3500 | 2.0137 | 2.3500 | 12.6800 |
| Economic policy uncertainty (EPU) | 1.3934 | 1.3404 | 0.2868 | 0.0060 | 2.0619 |
| Variable | Mean | Median | Min | Max | |
|---|---|---|---|---|---|
| 12.8550 | 0.0000 | 26.4586 | 0.0000 | 99.4900 | |
| Firm-level political risk (PRisk) | 0.8906 | 0.6334 | 0.8822 | 0.0000 | 7.2726 |
| Market-to-book ratio ( | 2.2010 | 1.5374 | 2.0534 | 0.2546 | 13.9320 |
| Financial leverage ( | 0.2402 | 0.2300 | 0.1793 | 0.0000 | 0.8519 |
| Research and development ( | 0.0567 | 0.0245 | 0.0858 | 0.0000 | 0.5388 |
| Return on assets ( | 0.0228 | 0.0466 | 0.1426 | −0.7797 | 0.3227 |
| Capital expenditure ( | 3.8495 | 2.9852 | 3.2611 | 0.0257 | 33.7268 |
| Firm size ( | 15.5475 | 15.5148 | 1.6532 | 11.4140 | 19.4316 |
| Cashflow volatility (CashVOLT) | 0.0520 | 0.0295 | 0.0816 | 0.0022 | 0.6861 |
| Stock volatility (StockVOLT) | 5.0542 | 5.0226 | 0.4231 | 4.0967 | 6.2967 |
| Natural logarithm of board size ( | 1.4853 | 1.5138 | 0.1750 | 0.1916 | 1.9007 |
| Gross domestic product ( | 2.2825 | 2.3026 | 0.2965 | 1.0986 | 3.2958 |
| Inflation ( | 17.4908 | 16.6700 | 12.2121 | 0.0000 | 56.2500 |
| Unemployment ( | 3.0891 | 0.9498 | 22.7907 | 0.1646 | 846.1912 |
| Control of corruption (CoC) | 0.9995 | 1.5404 | 2.7212 | −11.3557 | 7.6857 |
| Government effectiveness ( | 1.8207 | 1.7939 | 1.2787 | −2.3140 | 8.1533 |
| Regulatory quality ( | 5.8514 | 5.3500 | 2.0137 | 2.3500 | 12.6800 |
| Economic policy uncertainty ( | 1.3934 | 1.3404 | 0.2868 | 0.0060 | 2.0619 |
This table reports the summary statistics of the variables used in the baseline model over the period from 2002 to 2021. Specifically, the table summarises the mean, median, standard deviation, minimum and maximum values of the dependent, independent and control variables. The definitions of variables are shown in Appendix 1
The Pearson correlations are presented in Table 3. There is a significant positive correlation between PRisk and ESGC (ρ = 0.06, p < 0.01), providing preliminary evidence that the increase in firm-level political risk can contribute to the rise of ESG controversies. The highest correlation, 0.7267, is recorded between CoC and GOVEF. This supports the argument of Montes and Paschoal (2015) that control of corruption is correlated with government effectiveness. Since this highest pairwise correlation does not exceed 0.8, our model does not suffer from a multicollinearity problem (Gujarati and Porter, 2009).
Pairwise correlations
| Variable | ESGC | PRisk | MTB | LEV | RD | ROA | CAPX | SIZE | CashVOLT | StockVOLT | BSIZE | GDP | INF | UNEMP | CoC | GOVEF | REGQ | EPU |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ESGC | 1 | |||||||||||||||||
| PRisk | 0.0618*** | 1 | ||||||||||||||||
| MTB | −0.1155*** | −0.0024 | 1 | |||||||||||||||
| LEV | 0.0559*** | −0.0068 | −0.1367*** | 1 | ||||||||||||||
| RD | −0.1014*** | 0.0168 | 0.3934*** | −0.2012*** | 1 | |||||||||||||
| ROA | 0.0751*** | −0.0554*** | −0.0546*** | −0.0318*** | −0.5729*** | 1 | ||||||||||||
| CAPX | 0.0955*** | −0.0421*** | −0.0891*** | 0.0314*** | −0.1939*** | 0.1306*** | 1 | |||||||||||
| SIZE | 0.3963*** | 0.018 | 0.0583*** | 0.1130*** | −0.2622*** | 0.3977*** | 0.0996*** | 1 | ||||||||||
| CashVOLT | −0.0983*** | 0.0383*** | 0.2926*** | −0.1130*** | 0.5510*** | −0.4920*** | −0.1134*** | −0.3320*** | 1 | |||||||||
| StockVOLT | −0.0097 | −0.0112 | 0.0449*** | −0.0074 | −0.0142 | −0.018 | 0.0982*** | −0.0346*** | 0.0066 | 1 | ||||||||
| BSIZE | 0.2776*** | 0.0337*** | −0.2750*** | 0.1979*** | −0.2796*** | 0.1691*** | 0.1211*** | 0.5304*** | −0.2719*** | −0.0695*** | 1 | |||||||
| GDP | −0.0069 | −0.1177*** | 0.0372*** | −0.0133 | 0.0573*** | 0.0181 | −0.0055 | 0.0097 | 0.0361*** | 0.006 | −0.0233*** | 1 | ||||||
| INF | 0.0025 | −0.0089 | 0.1405*** | −0.0032 | 0.0657*** | −0.0337*** | −0.0356*** | −0.0865*** | 0.0925*** | 0.0236*** | −0.1373*** | 0.3112*** | 1 | |||||
| UNEMP | 0.1067*** | 0.0611*** | −0.0408*** | 0.0211 | −0.1293*** | 0.0659*** | 0.0328*** | 0.0704*** | −0.0540*** | 0.0106 | 0.1089*** | −0.2973*** | −0.1144*** | 1 | ||||
| CoC | 0.0474*** | −0.0091 | −0.2358*** | −0.0371*** | −0.1944*** | 0.1180*** | 0.2004*** | 0.0658*** | −0.1438*** | 0.0584*** | 0.1512*** | −0.0051 | −0.1095*** | −0.1709*** | 1 | |||
| GOVEF | 0.0262*** | −0.0599*** | −0.1433*** | −0.0677*** | −0.0931*** | 0.0872*** | 0.1670*** | 0.0304*** | −0.0688*** | 0.0321*** | 0.0450*** | 0.0737*** | −0.0406*** | −0.1858*** | 0.7267*** | 1 | ||
| REGQ | −0.0006 | −0.0073 | −0.0260*** | −0.0349*** | −0.0198 | 0.0268*** | 0.0773*** | −0.0998*** | 0.0176 | 0.0481*** | −0.0888*** | 0.1970*** | 0.3515*** | −0.2722*** | 0.5897*** | 0.5534*** | 1 | |
| EPU | −0.0537*** | 0.1355*** | 0.1205*** | −0.0006 | 0.0691*** | −0.1236*** | −0.1282*** | −0.1793*** | 0.0845*** | 0.0154 | −0.1367*** | −0.2852*** | 0.0363*** | 0.2064*** | −0.1189*** | −0.3186*** | −0.0169 | 1 |
| Variable | PRisk | CashVOLT | StockVOLT | CoC | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | ||||||||||||||||||
| PRisk | 0.0618*** | 1 | ||||||||||||||||
| −0.1155*** | −0.0024 | 1 | ||||||||||||||||
| 0.0559*** | −0.0068 | −0.1367*** | 1 | |||||||||||||||
| −0.1014*** | 0.0168 | 0.3934*** | −0.2012*** | 1 | ||||||||||||||
| 0.0751*** | −0.0554*** | −0.0546*** | −0.0318*** | −0.5729*** | 1 | |||||||||||||
| 0.0955*** | −0.0421*** | −0.0891*** | 0.0314*** | −0.1939*** | 0.1306*** | 1 | ||||||||||||
| 0.3963*** | 0.018 | 0.0583*** | 0.1130*** | −0.2622*** | 0.3977*** | 0.0996*** | 1 | |||||||||||
| CashVOLT | −0.0983*** | 0.0383*** | 0.2926*** | −0.1130*** | 0.5510*** | −0.4920*** | −0.1134*** | −0.3320*** | 1 | |||||||||
| StockVOLT | −0.0097 | −0.0112 | 0.0449*** | −0.0074 | −0.0142 | −0.018 | 0.0982*** | −0.0346*** | 0.0066 | 1 | ||||||||
| 0.2776*** | 0.0337*** | −0.2750*** | 0.1979*** | −0.2796*** | 0.1691*** | 0.1211*** | 0.5304*** | −0.2719*** | −0.0695*** | 1 | ||||||||
| −0.0069 | −0.1177*** | 0.0372*** | −0.0133 | 0.0573*** | 0.0181 | −0.0055 | 0.0097 | 0.0361*** | 0.006 | −0.0233*** | 1 | |||||||
| 0.0025 | −0.0089 | 0.1405*** | −0.0032 | 0.0657*** | −0.0337*** | −0.0356*** | −0.0865*** | 0.0925*** | 0.0236*** | −0.1373*** | 0.3112*** | 1 | ||||||
| 0.1067*** | 0.0611*** | −0.0408*** | 0.0211 | −0.1293*** | 0.0659*** | 0.0328*** | 0.0704*** | −0.0540*** | 0.0106 | 0.1089*** | −0.2973*** | −0.1144*** | 1 | |||||
| CoC | 0.0474*** | −0.0091 | −0.2358*** | −0.0371*** | −0.1944*** | 0.1180*** | 0.2004*** | 0.0658*** | −0.1438*** | 0.0584*** | 0.1512*** | −0.0051 | −0.1095*** | −0.1709*** | 1 | |||
| 0.0262*** | −0.0599*** | −0.1433*** | −0.0677*** | −0.0931*** | 0.0872*** | 0.1670*** | 0.0304*** | −0.0688*** | 0.0321*** | 0.0450*** | 0.0737*** | −0.0406*** | −0.1858*** | 0.7267*** | 1 | |||
| −0.0006 | −0.0073 | −0.0260*** | −0.0349*** | −0.0198 | 0.0268*** | 0.0773*** | −0.0998*** | 0.0176 | 0.0481*** | −0.0888*** | 0.1970*** | 0.3515*** | −0.2722*** | 0.5897*** | 0.5534*** | 1 | ||
| −0.0537*** | 0.1355*** | 0.1205*** | −0.0006 | 0.0691*** | −0.1236*** | −0.1282*** | −0.1793*** | 0.0845*** | 0.0154 | −0.1367*** | −0.2852*** | 0.0363*** | 0.2064*** | −0.1189*** | −0.3186*** | −0.0169 | 1 |
This table demonstrates the pairwise correlations for the variables used in the baseline regression. Coefficients with ***are significant at the 0.01 level. All variables are defined in the Appendix 1
4.2 Regression results: political risk and ESG controversies
Table 4 illustrates the results by estimating equation (2), which tests our H1 on the impact of PRisk on ESG controversies. Specifically, we present the regression results for the relationship between firm-level political risk and ESG controversies after controlling for firm-level determinants of financial and governance in Column (1), country-level variables in Column (2) and firm-fixed effects in Columns (3) and (4). In Columns (1) and (2), the coefficients of PRisk are positive and statistically significant at the 1% level (1.0029, 0.9809, respectively), implying that those firms exposed to political risks tend to result in more ESG controversies. The results remain consistent in Columns (3) and (4), with the adjusted R2 values ranging from 23% to 40.4% across the four models, indicating a moderate explanatory level. Our result is also economically significant. For example, in Column 4, a 1-standard-deviation increase in PRisk causes an increase of 1.98 (antilog of [0.3364 * 0.882]) value of ESGC, with the standard deviation of ESGC being 26.45.
Baseline regression: firm-level political risk and ESG controversies
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Variable | ESGC | ESGC | ESGC | ESGC |
| PRisk | 1.0029*** | 0.9809*** | 0.3494** | 0.3364** |
| (3.1806) | (3.1289) | (2.0688) | (2.1535) | |
| MTB | −2.0341*** | −2.0186*** | −0.6843*** | −0.6728*** |
| (−11.5790) | (−11.4953) | (−2.7674) | (−2.6712) | |
| LEV | −3.6893* | −3.4949 | −2.2150 | −1.7633 |
| (−1.6997) | (−1.6031) | (−1.1862) | (−0.9016) | |
| RD | 6.8204 | 6.7251 | −5.2939 | −5.7323 |
| (1.4656) | (1.4456) | (−0.6008) | (−0.6217) | |
| ROA | −16.8370*** | −16.9759*** | −9.3566*** | −9.7342*** |
| (−6.8481) | (−6.9188) | (−3.4283) | (−3.5342) | |
| CAPX | 0.2794** | 0.2810** | −0.0036 | −0.0046 |
| (2.1821) | (2.1938) | (−0.0275) | (−0.0362) | |
| SIZE | 6.7959*** | 6.7822*** | −0.3073 | −0.1456 |
| (13.5241) | (13.4710) | (−0.5404) | (−0.2361) | |
| CashVOLT | 15.9788*** | 15.8884*** | −4.4325 | −4.5860 |
| (4.8694) | (4.8430) | (−0.7630) | (−0.7885) | |
| StockVOLT | 0.0119 | 0.0113 | 0.0348 | 0.0336 |
| (1.1943) | (1.1129) | (1.2602) | (1.2372) | |
| BSIZE | 3.0631* | 2.8974* | 3.0152 | 2.9999* |
| (1.8299) | (1.7222) | (1.6232) | (1.7288) | |
| GDP | −0.1053 | 0.0642 | ||
| (−0.4271) | (0.2416) | |||
| INF | 0.5312* | 0.6841** | ||
| (1.8415) | (2.2034) | |||
| UNEMP | 0.5432 | 0.6326* | ||
| (1.4874) | (1.7290) | |||
| CoC | −1.8840 | 3.7402 | ||
| (−0.5126) | (1.0943) | |||
| GOVEF | 8.6883* | 7.5977 | ||
| (1.8193) | (1.1836) | |||
| REGQ | 1.1594 | 2.2330 | ||
| (0.3661) | (0.7149) | |||
| EPU | −2.7641* | −3.3738** | ||
| (−1.7049) | (−2.0539) | |||
| Constant | −98.4451*** | −100.0873*** | 11.6182 | 1.1135 |
| (−15.4100) | (−7.5863) | (1.0741) | (0.0570) | |
| Year effects | Yes | Yes | Yes | Yes |
| Firm effects | No | No | Yes | Yes |
| Country effects | Yes | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 12,517 | 12,517 | 12,517 | 12,517 |
| Adj R2 | 0.235 | 0.236 | 0.403 | 0.404 |
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Variable | ||||
| PRisk | 1.0029 | 0.9809 | 0.3494 | 0.3364 |
| (3.1806) | (3.1289) | (2.0688) | (2.1535) | |
| −2.0341 | −2.0186 | −0.6843 | −0.6728 | |
| (−11.5790) | (−11.4953) | (−2.7674) | (−2.6712) | |
| −3.6893 | −3.4949 | −2.2150 | −1.7633 | |
| (−1.6997) | (−1.6031) | (−1.1862) | (−0.9016) | |
| 6.8204 | 6.7251 | −5.2939 | −5.7323 | |
| (1.4656) | (1.4456) | (−0.6008) | (−0.6217) | |
| −16.8370 | −16.9759 | −9.3566 | −9.7342 | |
| (−6.8481) | (−6.9188) | (−3.4283) | (−3.5342) | |
| 0.2794 | 0.2810 | −0.0036 | −0.0046 | |
| (2.1821) | (2.1938) | (−0.0275) | (−0.0362) | |
| 6.7959 | 6.7822 | −0.3073 | −0.1456 | |
| (13.5241) | (13.4710) | (−0.5404) | (−0.2361) | |
| CashVOLT | 15.9788 | 15.8884 | −4.4325 | −4.5860 |
| (4.8694) | (4.8430) | (−0.7630) | (−0.7885) | |
| StockVOLT | 0.0119 | 0.0113 | 0.0348 | 0.0336 |
| (1.1943) | (1.1129) | (1.2602) | (1.2372) | |
| 3.0631 | 2.8974 | 3.0152 | 2.9999 | |
| (1.8299) | (1.7222) | (1.6232) | (1.7288) | |
| −0.1053 | 0.0642 | |||
| (−0.4271) | (0.2416) | |||
| 0.5312 | 0.6841 | |||
| (1.8415) | (2.2034) | |||
| 0.5432 | 0.6326 | |||
| (1.4874) | (1.7290) | |||
| CoC | −1.8840 | 3.7402 | ||
| (−0.5126) | (1.0943) | |||
| 8.6883 | 7.5977 | |||
| (1.8193) | (1.1836) | |||
| 1.1594 | 2.2330 | |||
| (0.3661) | (0.7149) | |||
| −2.7641 | −3.3738 | |||
| (−1.7049) | (−2.0539) | |||
| Constant | −98.4451 | −100.0873 | 11.6182 | 1.1135 |
| (−15.4100) | (−7.5863) | (1.0741) | (0.0570) | |
| Year effects | Yes | Yes | Yes | Yes |
| Firm effects | No | No | Yes | Yes |
| Country effects | Yes | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 12,517 | 12,517 | 12,517 | 12,517 |
| Adj R2 | 0.235 | 0.236 | 0.403 | 0.404 |
This table reports the regression results for the impact of firm-level political risk on ESG controversies. Column (1) shows the regression results for the impact of firm-level political risk on ESG controversies after controlling for both firm-level financial and governance variables. Column (2) shows the regression results of the main model specified in equation (2). The regressions shown in Columns (1) and (2) are controlled for year, country and industry fixed effects. The regressions shown in Columns (3) and (4) are tested under firm fixed effects along with year, country and industry fixed effects, respectively. Under these fixed effect conditions, Column (3) shows the regression results for the impact of firm-level political risk on ESG controversies after controlling for both firm-level financial and governance variables. Column (4) shows the regression results of the main model specified in equation (2). The independent and control variables are winsorised at the 1 and 99% levels to mitigate the effect of outliers. Coefficients with ***, ** and *are significant at the 0.01, 0.05 and 0.10 levels, respectively. Robust t-statistics in parentheses. All variables are defined in the Appendix 1
The strong and positive coefficients of PRisk on ESGC across all models support the hypothesis that firm-level political risk increases ESG controversies. Firstly, our main finding is consistent with the theoretical predictions of moral disengagement theory (Bandura et al., 1996), which states that negative perceptions resulting from increased firm-level political risk can give rise to unethical behaviour or ESG controversies in a firm. Our findings suggest that perceptions arising from political risk could impact the motive for unethical firm behaviour. Hence, the increased political risk would result in reduced capital investment and financial performance shortfall, which could eventually lead to unethical motives (Choi et al., 2022; Banerjee and Dutta, 2022; Bernanke, 1983; Santoso, 2019; DasGupta, 2022; Chari et al., 2019; Kahneman and Tversky, 1979).
Secondly, our main finding aligns with the theoretical framework of the Cressey fraud triangle (Cressey, 1953), which states that when firms are under pressure, a possibility for fraudulent behaviour is created by converting the pressure into a reason to commit fraud, finding a chance to commit fraud and finding a justification for the adverse behaviour. Our finding suggests that pressure created through Prisk could be converted into a reason to commit unethical behaviour. Hence, increased political exposure would create additional pressure for managers, likely motivating them to manipulate and make the earnings announcements and other reporting intentionally non-transparent, resulting in ethical dilemmas (Hoang et al., 2023).
Turning to control variables, we observe that MTB, LEV and ROA variables negatively impact ESG controversies. This explains that the firms that are profitable and financially strong experience fewer controversies. In addition, firms in countries with higher EPU experience low ESG controversies. In times of high economic policy uncertainties, firms increase their ESG performance (Zhao, 2023). In contrast, larger firms (SIZE), firms with high capital expenditure (CAPX) and higher board governance (BSIZE) tend to experience more ESG controversies. Further, we identify that firms in countries with increased inflation (INF) experience more ESG controversies.
4.3 Channel analysis: role of ESG-linked CEO compensation
To test the moderating effect of ESG-linked CEO compensation as a channel on the impact of the firm-level political risk on ESG controversies, we use ESG-linked CEO compensation, obtained from Refinitiv DataStream, which takes the value of one if CEO compensation is linked to ESG and takes the value of zero otherwise. Then, we create an interaction term between PRisk and ESG-linked CEO compensation and rerun the baseline model to test the impact of ESG-linked CEO compensation on the PRisk−ESGC relationship. Corresponding results in Column (1) of Table 5 show that while the coefficient of the primary variable (PRisk) remains positive and significant, the coefficient of the interaction term (CEO*compensation*PRisk) is negative and significant (−0.9022), indicating that ESG-linked CEO compensation reduces the effect of PRisk on ESGC. Hence, the results support the hypothesis that ESG-linked CEO compensation is a moderating channel that explains the impact of firm-level political risk on ESG controversies. Our results further suggest that integrating ESG incentives into CEO compensation is an effective strategy to mitigate ESG controversies through increasing ESG performance during high political uncertainty (Peng and Smith, 2023; Derchi et al., 2020).
Channel analysis and use of alternative dependent variables
| (1) | (2) | (3) | |
|---|---|---|---|
| Variable | ESGC | ESGCn+1 | Centred ESGC |
| PRisk | 0.9585** | 0.6590** | 0.9809*** |
| (2.3417) | (1.9866) | (3.1289) | |
| Prisk*CEO compensation | −0.9022* | ||
| (−1.7112) | |||
| CEO_compensation | 1.2016 | ||
| (1.5862) | |||
| Constant | −31.0389*** | −134.8858*** | −112.6903*** |
| (−2.6691) | (−8.7829) | (−8.5415) | |
| Controls | Yes | Yes | Yes |
| Year effects | No | Yes | Yes |
| Country effects | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes |
| Observations | 12,361 | 10,960 | 12,517 |
| Adj R2 | 0.4 | 0.247 | 0.236 |
| (1) | (2) | (3) | |
|---|---|---|---|
| Variable | Centred | ||
| PRisk | 0.9585 | 0.6590 | 0.9809 |
| (2.3417) | (1.9866) | (3.1289) | |
| Prisk | −0.9022 | ||
| (−1.7112) | |||
| CEO_compensation | 1.2016 | ||
| (1.5862) | |||
| Constant | −31.0389 | −134.8858 | −112.6903 |
| (−2.6691) | (−8.7829) | (−8.5415) | |
| Controls | Yes | Yes | Yes |
| Year effects | No | Yes | Yes |
| Country effects | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes |
| Observations | 12,361 | 10,960 | 12,517 |
| Adj R2 | 0.4 | 0.247 | 0.236 |
This table reports the channel analysis and additional analyses of the main regression results. Column (1) describes the role of ESG-linked CEO compensation as a channel to the relationship in the baseline model. Columns (2) and (3) demonstrate the use of alternative measures to the dependent variable: use of mean adjusted ESG controversy score [column (1)] and use of (n + 1) of the ESG controversy score [column (2)]. Coefficients with ***, ** and *are significant at the 0.01, 0.05 and 0.10 levels, respectively. Robust t-statistics in parentheses
4.4 Endogeneity analysis
4.4.1 Addressing reverse causality.
Reverse causation, one of the sources of endogeneity (Antonakis et al., 2010b), is a phenomenon that explains the association of two variables in a direction opposite to what a researcher expects. In our setting, there may be a concern that due to the public pressure, firms may intentionally reduce disclosing ESG controversies to the media in the current period. Using lagged independent variables is popular in addressing reverse causality (Bellemare et al., 2017). Hence, to check any effects of reverse causality, we regress the one-year-ahead dependent variable with independent and control variables in the baseline model. As per Column (2) of Table 5, the coefficient of PRisk (0.6590) remains positive and significant on ESGC, supporting our baseline results.
4.4.2 Addressing model misspecification.
One might contend that firms demonstrating high firm-level political risk fundamentally differ from the firms that do not. This may raise the concern that our regression analysis has not adequately controlled for these differences. Hence, we use the entropy-balanced sample to reduce selection bias and model misspecifications (Hasan and Jiang, 2023). Entropy balancing helps to retain sample size, unlike propensity score matching (PSM), which reduces the size of one group to match with the other. Hence, entropy balancing enhances the covariate balance compared to PSM (McMullin and Schonberger, 2022). According to Hainmueller (2012), entropy balancing checks for the systematic and random differences between treatment and control groups, hence mitigating the concerns about results being affected by the choices of design.
We re-estimate equation (2) using the entropy balancing approach to estimate the influence of firm-level political risk on ESG controversies. We use the median of the PRisk to identify our treatment (high-PRisk firms such that PRisk above the median) and control (low-PRisk firms such that PRisk below the median) groups. The results of the entropy balancing approach are in Table 6. Panel A exhibits the pre- and post-balancing mean, variance and skewness of the covariates of the treatment and control groups. We witness that the covariates of the treatment and control groups depict balanced mean, variance and skewness after the entropy balancing. As in Panel B, the coefficient of PRisk is still positive and significant on ESGC (p < 0.01), which confirms our baseline results.
Entropy balancing and robust regression results
| Panel A: Before and after matching | Panel B: Baseline regression after entropy balancing | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Treatment group | Control group | ||||||||||
| Before/after matching | Before matching | After matching | |||||||||
| Variable | Mean | Variance | Skewness | Mean | Variance | Skewness | Mean | Variance | Skewness | Dep variable: ESGC | Coefficient |
| MTB | 1.740 | 2.461 | 3.460 | 2.357 | 4.714 | 2.659 | 1.740 | 2.461 | 3.460 | PRisk | 1.4761** |
| LEV | 0.260 | 0.026 | 0.626 | 0.234 | 0.034 | 0.671 | 0.26 | 0.026 | 0.626 | (2.5781) | |
| RD | 0.039 | 0.003 | 3.614 | 0.063 | 0.009 | 2.746 | 0.039 | 0.003 | 3.614 | Constant | 24.7453 |
| ROA | 0.049 | 0.009 | −2.845 | 0.014 | 0.024 | −2.547 | 0.049 | 0.009 | −2.845 | (0.8842) | |
| CAPX | 4.325 | 10.06 | 1.405 | 3.689 | 10.73 | 2.399 | 4.325 | 10.06 | 1.405 | Controls | Yes |
| SIZE | 16.83 | 2.260 | −0.436 | 15.11 | 2.146 | −0.132 | 16.83 | 2.26 | −0.436 | Year effects | Yes |
| CashVOLT | 0.035 | 0.002 | 7.488 | 0.058 | 0.008 | 4.538 | 0.035 | 0.002 | 7.488 | Country effects | Yes |
| StockVOLT | 2.435 | 399.1 | 32.360 | 3.31 | 559.9 | 27.010 | 2.436 | 399.2 | 32.36 | Industry effects | Yes |
| BSIZE | 2.449 | 0.076 | 0.035 | 2.226 | 0.079 | 0.156 | 2.449 | 0.076 | 0.035 | Observations | 12,517 |
| GDP | 0.960 | 6.779 | −1.111 | 1.013 | 7.617 | −0.992 | 0.960 | 6.779 | −1.111 | Adj R2 | 0.0708 |
| INF | 1.790 | 1.460 | 0.468 | 1.831 | 1.694 | 0.584 | 1.790 | 1.460 | 0.468 | ||
| UNEMP | 6.271 | 4.468 | 0.284 | 5.710 | 3.836 | 0.638 | 6.271 | 4.468 | 0.284 | ||
| CoC | 1.423 | 0.086 | −0.680 | 1.384 | 0.080 | −0.013 | 1.423 | 0.086 | −0.68 | ||
| GOVEF | 1.496 | 0.036 | −2.961 | 1.482 | 0.029 | −2.014 | 1.496 | 0.036 | −2.961 | ||
| REGQ | 1.442 | 0.043 | −0.581 | 1.448 | 0.037 | −0.408 | 1.442 | 0.043 | −0.581 | ||
| EPU | 5.005 | 0.176 | 0.319 | 5.071 | 0.179 | 0.303 | 5.005 | 0.176 | 0.319 | ||
| Panel A: Before and after matching | Panel B: Baseline regression after entropy balancing | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Treatment group | Control group | ||||||||||
| Before/after matching | Before matching | After matching | |||||||||
| Variable | Mean | Variance | Skewness | Mean | Variance | Skewness | Mean | Variance | Skewness | Dep variable: | Coefficient |
| 1.740 | 2.461 | 3.460 | 2.357 | 4.714 | 2.659 | 1.740 | 2.461 | 3.460 | PRisk | 1.4761 | |
| 0.260 | 0.026 | 0.626 | 0.234 | 0.034 | 0.671 | 0.26 | 0.026 | 0.626 | (2.5781) | ||
| 0.039 | 0.003 | 3.614 | 0.063 | 0.009 | 2.746 | 0.039 | 0.003 | 3.614 | Constant | 24.7453 | |
| 0.049 | 0.009 | −2.845 | 0.014 | 0.024 | −2.547 | 0.049 | 0.009 | −2.845 | (0.8842) | ||
| 4.325 | 10.06 | 1.405 | 3.689 | 10.73 | 2.399 | 4.325 | 10.06 | 1.405 | Controls | Yes | |
| 16.83 | 2.260 | −0.436 | 15.11 | 2.146 | −0.132 | 16.83 | 2.26 | −0.436 | Year effects | Yes | |
| CashVOLT | 0.035 | 0.002 | 7.488 | 0.058 | 0.008 | 4.538 | 0.035 | 0.002 | 7.488 | Country effects | Yes |
| StockVOLT | 2.435 | 399.1 | 32.360 | 3.31 | 559.9 | 27.010 | 2.436 | 399.2 | 32.36 | Industry effects | Yes |
| 2.449 | 0.076 | 0.035 | 2.226 | 0.079 | 0.156 | 2.449 | 0.076 | 0.035 | Observations | 12,517 | |
| 0.960 | 6.779 | −1.111 | 1.013 | 7.617 | −0.992 | 0.960 | 6.779 | −1.111 | Adj R2 | 0.0708 | |
| 1.790 | 1.460 | 0.468 | 1.831 | 1.694 | 0.584 | 1.790 | 1.460 | 0.468 | |||
| 6.271 | 4.468 | 0.284 | 5.710 | 3.836 | 0.638 | 6.271 | 4.468 | 0.284 | |||
| CoC | 1.423 | 0.086 | −0.680 | 1.384 | 0.080 | −0.013 | 1.423 | 0.086 | −0.68 | ||
| 1.496 | 0.036 | −2.961 | 1.482 | 0.029 | −2.014 | 1.496 | 0.036 | −2.961 | |||
| 1.442 | 0.043 | −0.581 | 1.448 | 0.037 | −0.408 | 1.442 | 0.043 | −0.581 | |||
| 5.005 | 0.176 | 0.319 | 5.071 | 0.179 | 0.303 | 5.005 | 0.176 | 0.319 | |||
This table reports the covariate balances of entropy balancing. Panel A exhibits the pre- and post-matching mean, variance and skewness of all the covariates in the treatment and control groups. The treatment group is high PRisk firms. Panel B exhibits the results of the re-estimation of the baseline regression after the entropy balancing. Robust t-statistics in parentheses. Coefficients with ***, ** and *are significant at the 0.01, 0.05 and 0.10 levels, respectively. Appendix 1 shows the variable definitions
4.4.3 Exogenous shocks.
We have established that PRisk significantly increases ESG controversies. We now investigate if this relationship is held during unprecedented circumstances. Firstly, the world community had high hopes for the G7, that as the group of the largest economies, they would coordinate and take actions to mitigate the impact of the global financial crisis (GFC) during 2007–2009 (Chris Isidore, 2008). Prior literature establishes that economic crises can lead to unethical behaviours in firms (Lupuleac et al., 2012). At the same time, firms increase their CSR performance following the financial crisis to regain lost trust and reputation in their businesses (Giannarakis and Theotokas, 2011). Following this strand of literature, we test whether GFC acts as an exogenous shock on the association between firm-level political risk and ESG controversies. We create a GFC dummy variable, which takes the value of one for years after 2009 and takes the value of zero otherwise. Then, we test the direct impact of the GFC on ESGC and its interaction effect on the PRisk−ESGC relationship. In Column (1) of Table 7, the coefficient of the Post GFC dummy (5.4950) is positive and significant on ESGC. This suggests that ESG controversies increased overall in the post-GFC period, independent of political risk, consistent with the prior literature (Lupuleac et al., 2012). In contrast, the significant negative coefficient (−2.3301) of the interaction term (PRisk*Post GFC) on ESGC implies that the marginal effect of political risk on ESG controversies is weakened after the GFC. As the overall net coefficient of PRisk becomes positive (3.1688–2.3301 = 0.8387), this suggests that while political risk remained a driver of ESG controversies, its marginal effect diminished after the GFC. This could be possibly due to increased scrutiny, reputational regain efforts or regulatory reforms such as introducing a code of conduct by the International Organisation of Securities Commission and a framework for audit quality by the International Auditing and Assurance Standards Board (Borio et al., 2020). The findings are consistent with the prior literature (Giannarakis and Theotokas, 2011).
Exogenous shock effect of the global financial crisis and Paris Agreement
| (1) | (2) | |
|---|---|---|
| Variable | ESGC | ESGC |
| PRisk | 3.1688** (2.4084) | 1.3606*** (2.9992) |
| Post GFC | 5.4950*** (3.9326) | |
| PRisk*Post GFC | −2.3301* (−1.7413) | |
| Post Paris | −2.7753** (−2.4039) | |
| PRisk*Post Paris | −0.6154 (−1.1336) | |
| Constant | −113.7765*** (−11.6140) | −109.2095*** (−11.6876) |
| Controls | Yes | Yes |
| Industry effects | Yes | Yes |
| Country effects | Yes | Yes |
| Observations | 11,035 | 11,887 |
| Adj R2 | 0.231 | 0.233 |
| (1) | (2) | |
|---|---|---|
| Variable | ||
| PRisk | 3.1688** (2.4084) | 1.3606*** (2.9992) |
| Post | 5.4950*** (3.9326) | |
| PRisk*Post | −2.3301* (−1.7413) | |
| Post Paris | −2.7753** (−2.4039) | |
| PRisk*Post Paris | −0.6154 (−1.1336) | |
| Constant | −113.7765*** (−11.6140) | −109.2095*** (−11.6876) |
| Controls | Yes | Yes |
| Industry effects | Yes | Yes |
| Country effects | Yes | Yes |
| Observations | 11,035 | 11,887 |
| Adj R2 | 0.231 | 0.233 |
This table reports the exogenous shock effect of the global financial crisis (GFC) and the Paris Agreement on the relationship between political risk and ESG controversies. This demonstrates the interaction effect of the global crisis with political risk on ESG controversies and the interaction effect of the Paris Agreement with political risk on ESG controversies. Column (1) depicts the results of the GFC interaction effect with PRisk on ESGC for the full sample. Column (2) reports the interaction effect of Paris with PRisk on ESGC for the full sample. Coefficients with ***, ** and * are significant at the 0.01, 0.05 and 0.10 levels, respectively. Robust t-statistics in parentheses
Secondly, committing to the Paris Agreement and reducing the firms’ exposure to carbon risk bears both advantages and disadvantages to firms’ environmental and ethical practices (Agliardi et al., 2023; Perera et al., 2023). Hence, we explore whether the PA acts as an exogenous shock on the relationship between political risk and ESG controversies. We create a Paris Agreement dummy variable, which takes the value of one after 2015 and zero otherwise. Then, we test the direct impact of the PA on ESGC and its interaction effect with the PRisk−ESGC relationship. The results presented in Column (2) of Table 7 report that the coefficient of the Paris Agreement dummy (2.7753) is negative and significant on ESGC. This result appropriately indicates that ESG controversies have declined on average due to the Paris Agreement. This can be attributable to the positive spillovers in highly environmentally committed firms following the Paris Agreement (Agliardi et al., 2023).
However, the coefficient of the interaction term (PRisk* Post Paris) on ESGC is statistically insignificant. This suggests that the effect of PRisk on ESGC did not change meaningfully after the Paris Agreement. It further implies that while ESG controversies declined overall, the relative impact of political risk on ESG controversies remained stable. This implies that the Paris Agreement may have resulted in general improvements in ESG performance, as the prior literature suggests that the agreement has been able to establish politically successful climate negotiations in countries through a flexible approach, which has allowed countries to set unique targets to mitigate climate change (Dimitrov, 2016; Falkner, 2016). However, the results suggest that these general improvements from the Paris Agreement did not substantially alter how firm-level political risk translates into ESG controversies.
4.5 Robustness and additional analysis
4.5.1 Use of alternative measures.
We use a mean-adjusted ESGC score as an alternative to ESGC, as it removes outliers and provides a more concise picture of the effect of PRisk on ESGC. Following Hasan and Jiang (2023), we derive the mean adjusted ESGC to verify the impact of political risk on ESG controversies. Column (3) of Table 5 exhibits a significantly positive coefficient of PRisk (p < 0.01) for mean-adjusted ESGC. Thus, the result indicates that our documented finding is robust for different measures of ESGC.
4.5.2 US vs non-US context.
Given that US dominance exists in our sample, we validate whether the results prevail in G7 countries other than the US. For this purpose, we estimate equation (2) for non-US and US sub-samples and the corresponding results are illustrated in Table 8, Columns (1) and (2). The results reveal the positive and significant coefficients of PRisk on the ESGC in both sub-samples, resulting in the coefficient of PRisk as 1.0422 (p < 0.05) for non-US firms, while 0.8145 (p < 0.05) for US firms. Thus, the results confirm that the US dominance does not violate our baseline findings.
Sub-sample analysis
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| US | Non-US | Without environmental initiatives | With environmental initiatives | |
| Variable | ESGC | ESGC | ESGC | ESGC |
| PRisk | 1.0422** | 0.8145** | 1.1478*** | 0.0603 |
| (2.1082) | (2.1325) | (3.2219) | (0.0845) | |
| Constant | −125.2485*** | −123.8454*** | −71.0393*** | −209.5724*** |
| (−5.7799) | (−9.5969) | (−4.9644) | (−6.5477) | |
| Controls | Yes | Yes | Yes | Yes |
| Year effects | Yes | No | Yes | Yes |
| Country effects | Yes | No | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 4,318 | 8,199 | 10,397 | 2,115 |
| Adj R2 | 0.253 | 0.247 | 0.197 | 0.323 |
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Non-US | Without environmental initiatives | With environmental initiatives | ||
| Variable | ||||
| PRisk | 1.0422 | 0.8145 | 1.1478 | 0.0603 |
| (2.1082) | (2.1325) | (3.2219) | (0.0845) | |
| Constant | −125.2485 | −123.8454 | −71.0393 | −209.5724 |
| (−5.7799) | (−9.5969) | (−4.9644) | (−6.5477) | |
| Controls | Yes | Yes | Yes | Yes |
| Year effects | Yes | No | Yes | Yes |
| Country effects | Yes | No | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 4,318 | 8,199 | 10,397 | 2,115 |
| Adj R2 | 0.253 | 0.247 | 0.197 | 0.323 |
This table reports the results of the impact of firm-level political risk on ESG controversies tested on sub-samples. Columns (1) and (2) demonstrate US and non-US sub-samples and columns (3) and (4) demonstrate firms without and with environmental initiatives sub-samples. Coefficients with ***, ** and *are significant at the 0.01, 0.05 and 0.10 levels, respectively. Robust t-statistics in parentheses
4.5.3 Role of environmental initiatives.
Moral conscience and ethical value orientation influence employees’ involvement in environmental engagements (Wahab, 2021). Furthermore, environmental initiatives and engagements are vital to improving a firm’s reputation through being an ethical and sustainable corporate citizen. On the contrary, environmentally irresponsible firms will experience adverse effects on perceived corporate ethics and corporate reputation (Kao et al., 2016). Consequently, firms adopting environmental initiatives could reduce the harmful consequences of firms’ irresponsible environmental behaviours (Kao et al., 2016). Hence, when the firms exhibit higher levels of environmental engagement, ESG controversies or any related unethical behaviours could be reduced.
Thus, we anticipate that the presence of environmental initiatives of a firm is an important factor that could influence the impact of firm-level political risk on ESG controversies due to the moral disengagement resulting from political uncertainties. To test this proposition, we conduct a sub-sample analysis for the firms that have and do not have environmental initiatives. Our results in columns (3) and (4) in Table 8 show that the impact of political risk on ESG controversies exists predominantly in firms that lack environmental initiatives. This finding suggests that environmental initiatives can significantly reduce the occurrence of ESG controversies in times of high political risk (Kao et al., 2016).
4.5.4 Managerial power.
Thus far, we have reported that firm-level political risk increases ESG controversies. One could argue that good corporate governance encourages firms to counter corporate fraud (Salleh and Othman, 2016). Hence, the impact of PRisk on ESGC could vary in the presence of good governance.
Using principal component analysis, we developed a corporate governance index and obtained principal components representing managerial power and board composition. Appendix 4 provides the detailed description of the index. Our first principal component strongly correlates with two input variables, as the first component increases with the instances where the same person holds both the chairman and CEO positions and when the chairman has held a CEO position previously. Hence, the first component can be viewed as a measure of managerial power. Based on the sample median of the first component, we divide the sample into two subgroups of firms with high managerial power and low managerial power. This sample partition captures the difference in the effects of firm-level political risk on ESG controversies between firms with high managerial power and those with low managerial power. We re-estimate equation (2) for each sub-sample and present results in Columns (1) and (2) of Table 9. High managerial power raises concerns about effective corporate governance, as CEOs may misuse their power to control business operations and decisions (Salleh and Othman, 2016; Agrawal and Chadha, 2005). For instance, Agrawal and Chadha (2005) document that CEO duality increases the probability of earnings manipulations in firms.
The role of managerial power and board monitoring
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Managerial power | Board monitoring | |||
| High | Low | Low | High | |
| Variable | ESGC | ESGC | ESGC | ESGC |
| PRisk | 1.2417*** | 0.2065 | 1.3000*** | 0.6860 |
| (3.2537) | (0.7475) | (3.2580) | (1.3943) | |
| Constant | −24.5653 | 28.8364 | −81.7168*** | −98.5971*** |
| (−0.6595) | (1.0474) | (−5.3637) | (−3.1342) | |
| Controls | Yes | Yes | Yes | Yes |
| Year effects | Yes | Yes | Yes | Yes |
| Country effects | Yes | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 5,526 | 6,240 | 7,188 | 4,770 |
| Adj R2 | 0.443 | 0.370 | 0.199 | 0.299 |
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Managerial power | Board monitoring | |||
| High | Low | Low | High | |
| Variable | ||||
| PRisk | 1.2417 | 0.2065 | 1.3000 | 0.6860 |
| (3.2537) | (0.7475) | (3.2580) | (1.3943) | |
| Constant | −24.5653 | 28.8364 | −81.7168 | −98.5971 |
| (−0.6595) | (1.0474) | (−5.3637) | (−3.1342) | |
| Controls | Yes | Yes | Yes | Yes |
| Year effects | Yes | Yes | Yes | Yes |
| Country effects | Yes | Yes | Yes | Yes |
| Industry effects | Yes | Yes | Yes | Yes |
| Observations | 5,526 | 6,240 | 7,188 | 4,770 |
| Adj R2 | 0.443 | 0.370 | 0.199 | 0.299 |
This table reports the results of the baseline regression run on sub-samples based on the extent of managerial power and governance score. Column (1) depicts the results of the sub-sample with high managerial power, while Column (2) depicts the results of the sub-sample with low managerial power. Column (3) presents the results of the sub-sample with a low governance score, while column (4) presents the results of the sub-sample with a high governance score. Coefficients with ***, ** and * are significant at the 0.01, 0.05 and 0.10 levels, respectively. Robust t-statistics in parentheses
Our results in Columns (1) and (2) in Table 9 show that the impact of political risk on ESG controversies exists predominantly in firms with high managerial power. This finding explains that firms with high managerial power, such as CEO duality, lack effective corporate governance and have a tendency to engage in more fraudulent activities compared to the firms that have effective corporate governance mechanisms. Further, the finding supports the prior literature (Salleh and Othman, 2016; Agrawal and Chadha, 2005) that documents the concerns about the impacts of high managerial power on corporate governance.
4.5.5 Board monitoring.
We next consider the influence of board monitoring and examine whether this corporate governance mechanism impacts the primary relationship uncovered in our study. Prior literature (Mallidis et al., 2024) documents that more female directors on a board would reduce the occurrences of ESG controversies. Concurrently, Issa and Hanaysha (2023) further discuss this argument through the critical mass concept [13], that three or more females on the board will likely decrease ESG controversies in firms. At the same time, independent directors would increase a board’s transparency and monitoring (Terjesen et al., 2015), which is a phenomenon supported by several global corporate governance codes, such as the Sarbanes–Oxley Act, mandating the inclusion of a significant share of independent directors on boards. Moreover, firms concerned about board independence are more likely to have gender-diverse boards (Terjesen et al., 2015).
From the index we created on corporate governance in the previous section, we now consider the second principal component. Our second principal component strongly correlates with two input variables, as the second component increases with board independence and the percentage of female directors. Hence, the second component can be viewed as a measure of board monitoring. Based on the sample median of the second component, we then split the sample into two subgroups of firms with high board monitoring and low board monitoring. This sample partition captures the difference in the effects of firm-level political risk on ESG controversies between firms with low board monitoring and those with high board monitoring. Results illustrated in Columns (3) and (4) of Table 9 show that the political risk’s impact on ESG controversies exists predominantly in the sub-sample of low board monitoring. This result explains that the impact of firm-level political risk on ESG controversies is more evident in firms with low board independence and fewer female directors. The finding in this context extends the prior studies (Mallidis et al., 2024; Terjesen et al., 2015), which suggest the influence of strong board independence and gender diversity on ESG controversies.
5. Conclusion
The increasing number of ESG controversies has put firms under media and public scrutiny. Thus, corporate reputation and investor trust in firms are at risk. Furthermore, controversies cost firms a lot in terms of fines and litigation. Concurrently, political uncertainty influences firms’ business operations in several ways. This paper examines the impact of firm-level political risk on ESG controversies. We investigate the underlying impact of firm-level political risk on ESG controversies for a sample of G7 countries from 2002 to 2021.
Our study reveals a significant positive impact of firm-level political risk on ESG controversies. Thus, aligned with prior literature on moral disengagement theory (Bandura et al., 1996) and the fraud triangle hypothesis (Jacobs and Cressey, 1954), our findings support the notion that negative perceptions arising from an adverse political environment can lead to unethical firm behaviour.
Our study further identifies that the ESG-linked CEO compensation acts as a channel and moderates the impact of firm-level political risk on ESG controversies. Our sub-sample analysis documents that the positive impact of firm-level political risk on ESG controversies prevails in both US and non-US firms despite the US dominance in the sample. Furthermore, the impact of political risk on ESG controversies is more pronounced in firms that lack environmental initiatives. Thus, it highlights the importance of environmental initiatives for firms. Our findings further have empirical implications for climate-related policies from the evidence of the Paris Agreement’s effect on the underlying relationship. We find that the Paris Agreement influences ESG controversies through general improvements in ESG performance. Nevertheless, the results suggest that these general improvements from the Paris Agreement did not substantially alter how firm-level political risk translates into ESG controversies. Thus, in line with the prior literature on the Paris Agreement, our findings support the notion of literature that the Paris Agreement has mixed interventions with political uncertainties and ESG practices. Our findings further reveal that the financial crisis moderates the impact of political risk on ESG Controversies. Thus, the findings have economic implications.
Our findings have important implications for various stakeholders, including investors, managers, regulators and policymakers. On practical grounds, our study suggests that both shareholders and managers need to evaluate the mechanisms and strategies to manage the political risks of firms to minimise the chances of controversies. This study has significant policy implications, leading to increased accountability of firms in their ESG performance and strengthened corporate governance. Importantly, the findings inform the policies towards increased regulatory scrutiny to enhance enforcement and ensure environmental regulation compliance. Our study indicates to policymakers the potential role of ESG assurance as a governance function and a risk mitigation strategy. Further, our study draws attention to policies on ESG assurance. Our study assists in improving the policies aimed at broader assurance requirements implied by frameworks such as the European Union’s Corporate Sustainability Reporting Directive. Our findings highlight that political risk leads to ESG controversies. Hence, without proper mechanisms to manage or mitigate political risk, firms would be more prone to ESG controversies. Therefore, our study provides policy insights on how to prevent the occurrence of ESG controversies when firms are exposed to high political risk and provides implications for policymakers to devise frameworks for managing political risk. Thus, the findings will help firms adopt frameworks focused on risk management while maintaining good governance. Our study will further assist firms in gaining investor confidence in the firms’ reactions to manage political risk and reduce controversies. Regulators may find our research an important and supportive reference point to draw up necessary rules to maintain and enhance the governance and credibility of firms in times of high political uncertainty.
However, our study is not without limitations. The findings of our study are limited to the G7 countries. Considering the influential economic and political landscape of the G7 group, we assumed it is an ideal setting to test the underlying association. Nevertheless, the extent to which the findings of this study would hold in the rest of the world’s economies can be questionable. For instance, the implications of political exposure on firms in G7 countries are much more advanced than those of developing economies. Moreover, given that the sample is limited to G7 countries, our study findings do not take into account other influential economies in relation to the ESG context, such as China. This could stand as a constraint on the generalisability of the findings.
Future research may extend our study in several ways. Although this study found empirical evidence that validates our hypothesis, case studies in different economic landscapes would be beneficial in further investigations on ESG controversies linked with political uncertainties. Moreover, given the significance of the Chinese economy in relation to the ESG context, future studies could test the underlying proposition of our study on a sample of Chinese firms. Finally, future studies may extend our findings over time and by different sample selections to get a deeper understanding of the rationale and to examine the effectiveness of strategies that firms have implemented to reduce the adverse effects of political risk on firms.
Notes
See the following URL for the media report: www.nytimes.com/2018/04/04/us/politics/cambridge-analytica-scandal-fallout.html
See the following URL for the media report: Link to the cited article.
See the following URL for the media report: www.reuters.com/business/autos-transportation/german-labour-court-finds-tesla-broke-union-election-rules-2024-02-13/
See the following URL for the source of reference: www.ey.com/en_gl/insights/forensic-integrity-services/global-integrity-report
See the following URL for additional information on the Cressey hypothesis: Link to the cited article.
Refer following URL for more information on macro level political risk: www.ey.com/en_nz/insights/strategy/how-political-risk-affects-five-areas-at-the-top-of-the-c-suite-agenda
https://www.firmlevelrisk.com/available for the 2002-2021 period
World development report 2021: DATA FOR BETTER LIVES. (n.d.). Link to the cited article.
World Bank Group. (2024). Home | Worldwide Governance Indicators. In World Bank. www.govindicators.org/
First, Refinitiv extracts the values pertaining to controversies for all firms of a particular year and gets the count of controversies per firm. Then multiply the count of controversies by the severity weight. Next, they sort the firms from lowest to highest (lowest being better) considering the values after applying the severity weights. Finally, they apply the percentile rank formula to derive the ESG controversies scores.
See the following empirical study for further understanding of the critical mass concept: https://doi.org/10.1057/palgrave.ejis.3000358
References
Further reading
Appendix 1
Variable definitions
| Variable | Variable description and measurement | Source |
|---|---|---|
| Dependent variable | ||
| ESGC | Rearranged ESG controversy score: The original values of the Refinitiv ESG controversy score are rearranged as 100- the original Refinitiv ESG controversy score, for the convenience of interpretation of the results. In this modified score, the value of 0 implies no controversies and the value of 100 demonstrates the highest level of controversies level | DataStream |
| Independent variable | ||
| PRisk | The average firm-level political risk over the four quarters developed by Hassan et al. (2019). The PRisk variable is standardized by the standard deviations | www.firmlevelrisk.com/ |
| Control variables | ||
| Firm-level variables | ||
| MTB | The market value of assets is scaled by the book value of assets | Compustat |
| LEV | Financial leverage: calculated as the total debt scaled by total assets | Compustat |
| RD | Research and development expenses: scaled by the total assets | Compustat |
| ROA | A firm’s profitability: calculated as its income before extraordinary items scaled by the lagged total assets | Compustat |
| CAPX | The ratio of capital expenditure to total assets*100 | Compustat |
| CashVOLT | Cashflow volatility: the standard deviation of the operating cash flow scaled by the total assets for the previous five years | Compustat |
| SIZE | Natural logarithm of market capitalization | Compustat |
| BSIZE | Natural logarithm of board size | DataStream |
| StockVOLT | Stock volatility: the standard deviation of daily stock returns | CRSP |
| Macro level variables | ||
| GDP | GDP per capita growth (annual %) | https://databank.worldbank.org/ |
| INF | Inflation, GDP deflator (annual %) | https://databank.worldbank.org/ |
| UNEMP | Unemployment, total (% of total labour force) (modelled ILO estimate). | https://databank.worldbank.org/ |
| CoC | Control of corruption estimate from WGI | www.worldbank.org/en/publication/worldwide-governance-indicators |
| GOVEF | Government effectiveness estimate from WGI | www.worldbank.org/en/publication/worldwide-governance-indicators |
| REGQ | Regulatory quality estimate from WGI | www.worldbank.org/en/publication/worldwide-governance-indicators |
| EPU | EPU index by Baker et al. (2016) | www.policyuncertainty.com/ |
| Channel analysis | ||
| CEO_compensation | A dummy variable that takes the value of one if a firm links ESG targets with CEO compensation and takes the value of zero otherwise | DataStream |
| Additional analysis | ||
| Post Paris | A dummy variable that takes the value of one for years after 2015 and takes the value of zero otherwise | Perera et al. (2023) |
| Post GFC | A dummy variable that takes the value of one for post-global financial crisis (years after 2007) and takes the value of zero otherwise | Andrieș and Ursu (2016) |
| Environmental initiatives | A dummy variable that takes the value of one if a firm reports environmental investment initiatives in a given year and takes the value of zero otherwise | DataStream |
| Variable | Variable description and measurement | Source |
|---|---|---|
| Dependent variable | ||
| Rearranged | DataStream | |
| Independent variable | ||
| PRisk | The average firm-level political risk over the four quarters developed by | |
| Control variables | ||
| Firm-level variables | ||
| The market value of assets is scaled by the book value of assets | Compustat | |
| Financial leverage: calculated as the total debt scaled by total assets | Compustat | |
| Research and development expenses: scaled by the total assets | Compustat | |
| A firm’s profitability: calculated as its income before extraordinary items scaled by the lagged total assets | Compustat | |
| The ratio of capital expenditure to total assets*100 | Compustat | |
| CashVOLT | Cashflow volatility: the standard deviation of the operating cash flow scaled by the total assets for the previous five years | Compustat |
| Natural logarithm of market capitalization | Compustat | |
| Natural logarithm of board size | DataStream | |
| StockVOLT | Stock volatility: the standard deviation of daily stock returns | |
| Macro level variables | ||
| Inflation, | ||
| Unemployment, total (% of total labour force) (modelled | ||
| CoC | Control of corruption estimate from | |
| Government effectiveness estimate from | ||
| Regulatory quality estimate from | ||
| Channel analysis | ||
| CEO_compensation | A dummy variable that takes the value of one if a firm links | DataStream |
| Additional analysis | ||
| Post Paris | A dummy variable that takes the value of one for years after 2015 and takes the value of zero otherwise | |
| Post | A dummy variable that takes the value of one for post-global financial crisis (years after 2007) and takes the value of zero otherwise | |
| Environmental initiatives | A dummy variable that takes the value of one if a firm reports environmental investment initiatives in a given year and takes the value of zero otherwise | DataStream |
Appendix 2
National Business Systems
| National Business System | Examples of countries | Nature of the corporate governance mechanism |
|---|---|---|
| Liberal market economies (LMEs) | United States of America, United Kingdom, Australia, Canada, Ireland and New Zealand | Corporate governance norms are guided by agency theory and shareholder value maximisation |
| Coordinated market economies (CMEs) | Austria, Belgium, Denmark, Finland, the Netherlands, Norway, Sweden and Switzerland | Greater focus on value maximisation for multiple stakeholders, influencing how firms perceive corporate governance |
| Highly coordinatedeconomies | Japan | General prevalence of insider-dominated governance structures |
| European peripheral economies | France, Greece, Italy, Portugal, Spain, Czech Republic, Hungary, Poland, Romania and Slovakia | Moderately strong corporate governance norms |
| Advanced emerging economies | Chile, Turkey, Israel, South Africa, Korea and Taiwan | Well-defined corporate governance norms |
| Advanced city economies | Hong Kong and Singapore | Superior corporate governance norms |
| Arab oil-based economies | Kuwait, Qatar, Saudi Arabia and the United Arab Emirates | Poor to average corporate governance norms |
| Emerging economies | Algeria, Argentina, Bangladesh, Brazil, China, Colombia, Egypt, India, Indonesia, Kazakhstan, Malaysia, Mexico, Morocco, Nigeria, Pakistan, Peru, Philippines, Russia, Thailand, Ukraine and Vietnam | Poor corporate governance norms |
| Socialist economies | Cuba and Venezuela | Very weak corporate governance norms |
| National Business System | Examples of countries | Nature of the corporate governance mechanism |
|---|---|---|
| Liberal market economies (LMEs) | United States of America, United Kingdom, Australia, Canada, Ireland and New Zealand | Corporate governance norms are guided by agency theory and shareholder value maximisation |
| Coordinated market economies (CMEs) | Austria, Belgium, Denmark, Finland, the Netherlands, Norway, Sweden and Switzerland | Greater focus on value maximisation for multiple stakeholders, influencing how firms perceive corporate governance |
| Highly coordinatedeconomies | Japan | General prevalence of insider-dominated governance structures |
| European peripheral economies | France, Greece, Italy, Portugal, Spain, Czech Republic, Hungary, Poland, Romania and Slovakia | Moderately strong corporate governance norms |
| Advanced emerging economies | Chile, Turkey, Israel, South Africa, Korea and Taiwan | Well-defined corporate governance norms |
| Advanced city economies | Hong Kong and Singapore | Superior corporate governance norms |
| Arab oil-based economies | Kuwait, Qatar, Saudi Arabia and the United Arab Emirates | Poor to average corporate governance norms |
| Emerging economies | Algeria, Argentina, Bangladesh, Brazil, China, Colombia, Egypt, India, Indonesia, Kazakhstan, Malaysia, Mexico, Morocco, Nigeria, Pakistan, Peru, Philippines, Russia, Thailand, Ukraine and Vietnam | Poor corporate governance norms |
| Socialist economies | Cuba and Venezuela | Very weak corporate governance norms |
Appendix 3
Market cap grouping for firm size in determining the ESG controversy score by Refinitiv
| Global benchmark | Cap class | Severity rate |
|---|---|---|
| ≥ 10 billion | Large | 0.33 |
| ≥ 2 billion | Mid | 0.67 |
| < 2 billion | Small | 1 |
| Global benchmark | Cap class | Severity rate |
|---|---|---|
| ≥ 10 billion | Large | 0.33 |
| ≥ 2 billion | Mid | 0.67 |
| < 2 billion | Small | 1 |
Appendix 4
Author constructed index for managerial power and board monitoring
| Variable | Variable name | Variable description |
|---|---|---|
| Panel A: Variables included in the author-constructed index | ||
| CEO | CEO duality | A dummy variable that takes the value of one if the chairman is also the CEO and takes the value of zero otherwise |
| Ex_CEO | Chairman is an ex-CEO | A dummy variable that takes the value of one if the chairman has held a CEO position and takes the value of zero otherwise |
| BOARDIND | Availability of a policy for board independence | A dummy variable that takes the value of one if a firm has a policy for board independence and takes the value of zero otherwise |
| FEMALEDIR | Female directors | A dummy variable that takes the value of one if the female participation in the board is higher than the median value and takes the value of zero otherwise |
| Variable | Variable name | Variable description |
|---|---|---|
| Panel A: Variables included in the author-constructed index | ||
| A dummy variable that takes the value of one if the chairman is also the | ||
| Ex_CEO | Chairman is an ex-CEO | A dummy variable that takes the value of one if the chairman has held a |
| BOARDIND | Availability of a policy for board independence | A dummy variable that takes the value of one if a firm has a policy for board independence and takes the value of zero otherwise |
| FEMALEDIR | Female directors | A dummy variable that takes the value of one if the female participation in the board is higher than the median value and takes the value of zero otherwise |
| Component | Eigenvalue | Difference | Proportion | Cumulative |
|---|---|---|---|---|
| Panel B: Principal components/correlation | ||||
| Comp1 | 1.8604 | 0.742617 | 0.4651 | 0.4651 |
| Comp2 | 1.11778 | 0.267347 | 0.2794 | 0.7445 |
| Comp3 | 0.850433 | 0.679043 | 0.2126 | 0.9572 |
| Comp4 | 0.17139 | . | 0.0428 | 1 |
| Component | Eigenvalue | Difference | Proportion | Cumulative |
|---|---|---|---|---|
| Panel B: Principal components/correlation | ||||
| Comp1 | 1.8604 | 0.742617 | 0.4651 | 0.4651 |
| Comp2 | 1.11778 | 0.267347 | 0.2794 | 0.7445 |
| Comp3 | 0.850433 | 0.679043 | 0.2126 | 0.9572 |
| Comp4 | 0.17139 | . | 0.0428 | 1 |
| Variable | Comp1 | Comp2 | Comp3 | Comp4 |
|---|---|---|---|---|
| Panel C: Scoring coefficients | ||||
| CEO | 0.6884 | −0.1703 | 0.0358 | 0.7042 |
| Ex_CEO | 0.6944 | −0.1052 | 0.0746 | −0.708 |
| BOARDIND | 0.1885 | 0.6523 | −0.7341 | 0.0107 |
| FEMALEDIR | 0.092 | 0.7311 | 0.674 | 0.0526 |
| Variable | Comp1 | Comp2 | Comp3 | Comp4 |
|---|---|---|---|---|
| Panel C: Scoring coefficients | ||||
| 0.6884 | −0.1703 | 0.0358 | 0.7042 | |
| Ex_CEO | 0.6944 | −0.1052 | 0.0746 | −0.708 |
| BOARDIND | 0.1885 | 0.6523 | −0.7341 | 0.0107 |
| FEMALEDIR | 0.092 | 0.7311 | 0.674 | 0.0526 |

