On the Relationship Between Financial Distress and ESG Scores
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Lohmann, Christian; Möllenhoff, Steffen; Lehner, Sebastian Article — Published Version On the Relationship Between Financial Distress and ESG Scores Corporate Social Responsibility and Environmental Management Provided in Cooperation with: John Wiley & Sons Suggested Citation: Lohmann, Christian; Möllenhoff, Steffen; Lehner, Sebastian (2025) : On the Relationship Between Financial Distress and ESG Scores, Corporate Social Responsibility and Environmental Management, ISSN 1535-3966, John Wiley & Sons, Inc., Chichester, UK, Vol. 32, Iss. 5, pp. 6377-6401, https://doi.org/10.1002/csr.70033 This Version is available at: https://hdl.handle.net/10419/329808 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/
Corporate Social Responsibility and Environmental Management, 2025; 32:6377–6401 https://doi.org/10.1002/csr.70033 6377 Corporate Social Responsibility and Environmental Management REVIEW ARTICLE OPEN ACCESS On the Relationship Between Financial Distress and ESG Scores ChristianLohmann1 | SteffenMöllenhoff1,2 | SebastianLehner1,3 1Schumpeter School of Business and Economics, University of Wuppertal, Wuppertal, Germany | 2neXDos GmbH, Munich, Germany | 3Invesco Asset Management Deutschland GmbH, Frankfurt am Main,Germany Correspondence: Christian Lohmann ([email protected]) Received: 3 April 2025 | Revised: 4 June 2025 | Accepted: 11 June 2025 Keywords: ESG| ESG score| financial distress| nonparametric regression| shareholderstakeholder orientation ABSTRACT This empirical study analyzes the relationship between a company's financial distress obtained from a bankruptcy prediction model and ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Applying a nonparametric regression technique on panel data of listed US companies for 2003–2022 reveals a pronounced and statistically significant Ushaped relationship between financial distress and ESG scores. Financially distressed companies exhibit high ESG scores. Further empirical analysis shows that the most plausible interpretation is that companies anticipate their upcoming financial distress and intensify ESGsupporting disclosures to manage their ESG scores upward. The empirical results underline the importance of including the financial health of a company in ESG assessments. Only by taking into account both the ESG performance and the financial sustainability of a company is it possible to assess responsible corporate governance. JEL Classification: C33, G33, M41, Q56 1 | Introduction There is an ongoing debate on the informativeness of ESG scores that measure the ESG activities of a company. This debate also includes the effective relationship between a company's financial performance and ESG scores. Thereby, the latter constructed by various ESG rating agencies are used in business to inform operational corporate and longterm investment decisions (e.g., investment decisions by ESG funds; Raghunandan and Rajgopal2022) and in science for empirical research. The present study contributes to this research line by introducing the measure of bankruptcy risk as a new variable related to ESG scores. The measure of bankruptcy risk is the result of a bankruptcy prediction model and indicates the level of a company's financial distress. The present study analyzes the effective relationship between the measure of bankruptcy risk and ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG by applying a nonparametric regression technique on panel data of listed US companies from 2003 to 2022. The analysis reveals a pronounced and statistically significant Ushaped relationship between the level of financial distress and ESG scores at the company level. An increase in the measure of bankruptcy risk above a certain threshold is associated with increasing ESG scores. As a result, financially distressed companies exhibit ESG scores comparable to ESG scores from financially healthy companies. This empirical finding is robust as it is observable across the four different ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. However, one challenge of the analysis is to identify sufficiently strong evidence for the causality between financial distress and ESG scores. We address concerns related to reverse causality by applying a set of alternative variables. Based on our analysis, the most plausible explanation for the observed Ushaped relationship between financial distress and ESG scores is that companies anticipate financial distress and focus on costeffective ESG activities, such as ESGsupporting disclosures, to increase their ESG scores. We add to the existing literature by showing that managing ESG scores by the group of financially distressed companies reduces This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Author(s). Corporate Social Responsibility and Environmental Management published by ERP Environment and John Wiley & Sons Ltd.
6378 Corporate Social Responsibility and Environmental Management, 2025 the validity and credibility of ESG scores and makes them less reliable. Therefore, it is imperative to consider a company's financial situation when interpreting ESG scores. The empirical findings to date document a negative linear relationship between the level of financial distress and ESG scores (e.g., Zheng etal.2019; Boubaker etal.2020; Aslan etal.2021; Badayi etal. 2021; Truong etal. 2025). These empirical findings support two lines of argument that postulate a negative linear relationship between the level of financial distress and ESG scores. The first line of argument is that financially unconstrained companies have more financial resources than financially constrained companies to pursue ESG objectives and invest in projects demonstrating corporate goodness. The second line of argument relates to the predominantly positive relationship between financial performance and the ESG activities of a company. If increasing financial distress of a company is generally associated with lower financial performance, there should be a negative relationship between the level of financial distress and ESG scores. However, the previous results that show a consistently negative relationship between the level of financial distress and ESG scores are only reliable to a limited extent as these studies (e.g., Zheng etal.2019; Boubaker etal.2020; Aslan etal.2021; Badayi etal.2021; Truong etal.2025) exclusively use linear regression techniques and thus exclude a possible nonlinear relationship in the data from the outset although there are three good reasons for at least a partially positive relationship in the case of financially distressed companies. First, financially distressed companies have a solid interest in achieving high ESG scores to obtain equity and debt capital at relatively low cost as they rely on the low cost of capital to avoid or at least delay bankruptcy. Second, financially distressed companies may conceal information about the company's failures and opportunistically pursue ESG objectives to distract from their financial distress. Third, the intracompany incentive system may set incentives to achieve ESG objectives rather than financial ones that are particularly difficult to achieve in the case of financially distressed companies. To identify the effective and unrestricted relationship between the level of financial distress and ESG scores, a comprehensive empirical study has to apply a nonparametric regression technique to the available panel data. The interpretation of the revealed Ushaped relationship between the level of financial distress and ESG scores has to address the problem of reverse causality (Gow etal.2016). On the one hand, financial distress could be the cause, and a high ESG score could be the effect. In this case, one interpretation could be that financially distressed companies intensify ESGsupporting disclosures and manage their ESG scores upward, very likely to decrease the cost of capital, improve the financial conditions, and distract from their financial failure. Another reason for that observation could be the incentive system and the desire of management to increase personal benefits by achieving stakeholderoriented ESG objectives rather than shareholderoriented financial objectives. On the other hand, measures that lead to a high ESG score could be the cause, and financial distress could be the financial consequence of these measures. Such a relationship would be more likely to be observed with ESG investments, which have a greater impact on cash flows and corporate finances, than operational measures or ESGsupporting disclosures. We narrowed down the problem of reverse causality by conducting several extensions of empirical analysis. Based on the results of the empirical analysis, the most plausible interpretation of the revealed Ushaped relationship between the level of financial distress and ESG scores is that companies anticipate their upcoming financial distress and intensify costeffective ESG activities such as ESGsupporting disclosures to manage their ESG scores upward. This interpretation is also consistent with empirical findings that emphasize the importance of the quantity of ESG disclosures and consider the content of these ESG disclosures to be of secondary importance (Lyon and Maxwell2011; Marquis etal.2016). ESG scores are presumably influenced to a greater extent by the existence of ESG disclosures and less by their content (Drempetic etal.2020; LopezdeSilanes etal.2020). ESG scores are also enhanced by excessive and overexpectant disclosures on diversity, equity, and inclusion (Baker etal.2024), and ESG funds pay more attention to the existence and less to the content of ESG disclosures (Raghunandan and Rajgopal2022). The empirically observable, systematic management of ESG scores by the group of financially distressed companies makes it imperative to consider the degree of a company's financial distress when interpreting ESG scores. The paper is structured as follows: In the next section, we provide an overview of the literature and clarify the arguments in favor of a negative and positive relationship between financial distress and ESG scores. On the basis of these arguments, we formulate a hypothesis that is subsequently tested empirically. In Section3, we provide details on the applied data and describe the variables used for the main analysis. Section4 presents the empirical results on the estimated nonlinear relationship between the level of financial distress and ESG scores. This section also includes insights into the three ESG subfactors and applied control variables. In Section5, we extend the empirical analysis to address the problem of reverse causality and discuss the robustness of the results. Section6 concludes the paper with an overview of our findings and discusses the implications of the results. 2 | Literature Review The analysis of the relationship between financial distress and ESG scores refers to two strands of literature. The first strand of literature relates to the construction and significance of the measure of bankruptcy risk. Powerful, empirical bankruptcy prediction models (e.g., Beaver etal.2005; Balcaen and Ooghe2006; Bellovary etal.2007; Campbell etal.2008; Jones2017; Lohmann et al. 2023) can validly estimate the measure of bankruptcy risk. In contrast to periodic accounting indicators, such as return on equity and return on assets, and valuebased indicators, such as market value of equity, the measure of bankruptcy risk, which includes a large set of accountingbased, marketbased, companyspecific, and macroeconomic variables, enables a valid and robust estimation of a company's financial situation. The measure of bankruptcy risk is a forwardlooking indicator as it predicts an impending bankruptcy within the forecast horizon of the bankruptcy prediction model. Empirical findings show
6379 that professional investors likely apply bankruptcy prediction models to optimize their risk position as professional investors sell the shares of financially distressed companies that file for bankruptcy at an early stage and retain the shares of financially distressed peer companies that remain solvent (Lohmann and Möllenhoff2023a). As a result, the measure of bankruptcy risk is very well suited to measuring the sustainable financial situation of a company and, thus, making a valid statement about its continued existence. The second strand of literature relates to the relationship between financial distress and ESG scores and the arguments in favor of a negative or positive relationship. The empirical findings to date document a negative linear relationship between financial distress and ESG scores. Particularly, Zheng etal.(2019) show a negative relationship between the Z score and the ESG score from MSCI, Boubaker etal.(2020) show a negative relationship between the Z score and measure of corporate social responsibility (CSR) that is based on the qualitative dimensions of the MSCI ESG index, Aslan etal.(2021) show a negative relationship between the S&P Credit Rating and the ESG score from Refinitiv, Badayi et al. (2021) show a negative relationship between the Z score and the ESG score from Refinitiv, and Truong etal.(2025) show a negative relationship between the Z score and the ESG rating from MSCI. Lisin etal.(2022) and Cohen (2023) provide further empirical evidence on negative and mixed correlations between a company's financial distress and ESG scores. However, the informativeness of the cited empirical studies is limited, as they only use one ESG score and less developed bankruptcy risk measures such as the Z score and apply linear regression technique to analyze older datasets that are smaller in size and do not include firmyear observations from more recent years. Nevertheless, the empirical findings on a negative linear relationship between financial distress and ESG scores can be justified by two lines of argument. The first line of argument refers to the financial constraints of a company. A negative relationship between financial distress and ESG scores is expected for financially constrained companies. Hong etal.(2012) and Xu and Taehyun(2022) find that financially unconstrained companies have more resources than financially constrained ones to pursue social and environmental objectives and invest in projects showing corporate status. A company in financial distress should have severe financial constraints and will have less freely available financial resources to invest in ESGrelated projects. In addition, there is empirical evidence that financially distressed companies prefer operational and investment decisions that place less strain on the current cash flow and have a positive impact on shortterm financial performance indicators (e.g., Eisfeldt and Rampini 2007; Ma etal. 2022; Thomas etal. 2022). Such shortsighted decisions address the financial constraints of a financially distressed company. ESG investments are likely to be associated with uncertainties regarding their impact on future cash flows and should therefore not be suitable for easing financial constraints in the short term. As a result, shortsighted decisions in the context of financial distress are expected to be associated with an effective reduction in ESG performance. The second line of argument refers to the financial performance of a company. Lower financial performance is generally associated with the increasing financial distress of a company. The relationship between a company's financial performance and ESG scores has been extensively studied by applying predominantly linear regression techniques. A significant fourdigit number of individual studies and over a dozen metastudies (e.g., Friede etal.2015; Del Mar MirasRodríguez etal.2015; Hou etal.2016; Lu and Taylor2016; Wang etal.2016; Jeong and Harrison2017; Plewnia and Guenther 2017; Rost and Ehrmann 2017; Busch and Friede2018; Hoobler etal.2018; LópezArceiz etal.2018; GallardoVázquez etal.2019; Hang etal.2019; Vishwanathan etal.2020) found a predominantly positive relationship between the financial performance figures, mainly including variables such as return on equity, return on assets, and market value of equity, and the ESG activities of a company, which were very often considered in ESG scores' quantified form. The positive relationship between a company's financial performance and these scores suggests that there should also be a negative relationship between the level of financial distress and ESG scores. The empirical evidence on the relationship between a company's financial performance and ESG ratings is ambiguous and does not allow for a clear interpretation. Previous studies on this relationship apply linear regression models to a large extent; however, the estimated coefficients fluctuate considerably (e.g., the metastudy of Del Mar MirasRodríguez etal.2015) and sometimes even show a negative correlation (e.g., the metastudy of Rost and Ehrmann2017). In addition to using different samples that differ in country, time, and company characteristics, an explanation for these only partially consistent results could be an effective nonlinear relationship between a company's financial performance and ESG scores. However, the latter are also inconclusive as there is empirical evidence for a Ushaped relationship (e.g., Nollet etal.2016; Nuber etal.2020; Naimy etal.2021; Agarwala et al. 2024) and an inverted Ushaped relationship (e.g., Buallay etal.2022; Teng etal.2022; El Khoury etal.2023; Pu2023). As a result, an effective nonpositive or nonlinear relationship between a company's financial performance and ESG scores may indicate an effective nonnegative or nonlinear relationship between financial distress and ESG scores. Besides the empirical arguments for a negative relationship between the level of financial distress and ESG scores, there are also three lines of argument that can justify a positive relationship. The first line of argument is based on the empirical observation that higher ESG scores are associated with a lower cost of capital. Companies that implement the ESG principles and, in particular, are considered to be highly environmentally friendly can reduce their cost of capital relative to environmentally harmful companies that do not comply with ESG principles (e.g., Chava2014; van der Beck2023; Pástor etal.2022; Kacperczyk and Peydro2024; AronDine etal.2024; Green and Vallee2024). The difference in the cost of capital between compliant and noncompliant ESG companies is difficult to estimate; however, an analysis of quarterly earnings conference calls of US companies for the period 2016–2021 revealed that the difference in the cost of capital could be in the range of 2%–3% (Gormsen etal.2023). As financially distressed companies aim to keep the cost of capital as low as possible to ease financial constraints or to avoid or delay bankruptcy, they have a solid interest in achieving high ESG scores to obtain equity and debt capital at relatively low costs. The negative relationship between ESG scores and the
6380 Corporate Social Responsibility and Environmental Management, 2025 cost of capital incentivizes financially distressed companies to strive for high ESG scores. As a result, there could be a positive relationship between the level of financial distress and ESG scores, at least for financially distressed companies that have to lower their cost of capital to increase the probability of survival. In a second line of argument, the “management obfuscation hypothesis” supports a positive relationship between the level of financial distress and ESG scores. Bloomfield(2002) introduced the management obfuscation hypothesis and argued that managers have incentives to provide clear information about the company's successes and to obfuscate information about the company's failures. Several studies provide empirical evidence that the management obfuscation hypothesis is valid concerning information on a company's performance and earnings (e.g., Lang and Lundholm1993; DeFond and Jiambalvo1994; Schrand and Walther2000; Jaggi and Lee2002; Li2010). The empirical evidence on the management obfuscation hypothesis is less conclusive about bankruptcy risk information in annual reports. HolderWebb and Cohen(2007) doubt that the annual reports of financially distressed companies are informative, as selfinterested managers are incentivized to delay the disclosure of severe risks to the continued existence of their company as a going concern. Mayew etal.(2015) analyzed a large sample of listed US companies and found that the management obfuscation hypothesis is largely invalid. On the contrary, Lohmann and Ohliger(2020) provided empirical evidence that supports it as the annual reports of companies that eventually went bankrupt contain, on average, longer and relatively more complex risk reports and generally exhibit a less negative linguistic tone than the reports of financially distressed companies that remained solvent. As imminent bankruptcy risks are complex and challenging to communicate (Bloomfield2008), the high linguistic complexity may result from both obfuscation and complex information (Bushee etal.2018). Another option to obfuscate a company's financial distress is to focus on disclosed information about nonfinancial ESG objectives. If the company management pursues ESG objectives, financial objectives become less critical. Corporate governance and monitoring target achievement are complex, as ESG objectives are often qualitative and can only be measured subjectively. The increased complexity of such a multidimensional objective system could be useful to distract from financial objectives or conceal poor financial performance (Bebchuk and Tallarita2020; Karpoff2021). As financial distress mounts, the need to explain complex financial information increases, unless the focus can be placed on ESG objectives and their achievement. The required explanation of financial information can be interpreted as transaction costs that the company management aims to minimize. This argument is reinforced by empirical evidence on the positive relationship between CSR and ESG activities and earnings management (e.g., Patten and Trompeter2003; Gargouri et al. 2010; Yip et al. 2011; Muttakin et al. 2015; MartínezFerrero etal.2016; Jordaan etal.2018; Buertey etal.2020; Yu etal.2020; Pasko etal.2021; Zhang etal.2021) that is in line with the stakeholder agency theory (Jensen and Meckling1976; Hill and Jones1992). Thereby, the positive relationship between ESG activities and earnings management is enhanced by a company's financial distress (Almubarak etal.2023). Another empirical finding that supports the validity of the management obfuscation hypothesis is the fact that the existence and scope of ESG disclosures are particularly relevant rather than the real circumstances reflected in the ESG disclosures (Lyon and Maxwell2011; Marquis etal.2016; Drempetic etal.2020; LopezdeSilanes etal.2020; Raghunandan and Rajgopal2022). Recently, Flugum and Souther (2025) analyzed the company communication after quarterly earnings reports and provided empirical evidence that the management distracts from missing earning expectations by highlighting nonfinancial ESG objectives. One conclusion might be that companies that report, particularly, extensively on their ESG activities and prioritize ESG objectives in a publicityeffective manner are financially underperforming (Bhagat 2022). It is conceivable that financially distressed companies will also show an affinity with ESG objectives to distract from their financial distress. As the financial resources of a financially distressed company are likely to be very scarce regularly (Myers1977), an improvement in ESG scores is more likely to result from costeffective extensions of ESGsupporting disclosure rather than ESG investments, which require upfront capital expenditures and will only lead to positive cash flows in the long term. Thereby, extensions of ESGsupporting disclosure should be effective as ESG scores are presumably influenced to a greater extent by the existence of ESG disclosures and less by their content (Drempetic etal.2020; LopezdeSilanes etal.2020) and investors such as ESG funds focus on the existence and less to the content of ESG disclosures (Raghunandan and Rajgopal2022). Suppose the management obfuscation hypothesis about stakeholderoriented ESG objectives and associated ESG disclosures is valid. In that case, there should be a positive relationship between the level of financial distress and ESG scores, at least for financially distressed companies incentivized to obfuscate their financial failure. The third line of argument relates to the intracompany incentive system. In addition to financial objectives, ESG objectives are also pursued and implemented within the intracompany incentive system, whereby the design options are still very heterogeneous (Reda2022). Recent analyses show a growing share of listed US companies that incorporate ESG measures in annual incentive plans (e.g., Salzbank etal.2023; Kuk etal.2023). A company's management likely allocates the available resources to maximize the remuneration based on financial and ESG objectives. Cohen etal.(2023) provide empirical evidence that ESGbased incentives increase ESG performance and the associated ESG scores, while ESGbased incentives tend to lead to poorer financial performance. Chang etal.(2015) analyzed the relationship between financial distress risk and the incentivebased compensation of new CEOs. They found new CEOs receive more equitybased incentives to ensure that the company's financial position is improved. As a result, the choice and design of incentivebased remuneration significantly influence management decisions and the achievement of both financial and ESG objectives. Furthermore, the difficulty of achieving the target must be taken into account. It may be possible that ESG objectives can be achieved more easily than financial objectives; this is the case for companies that do not have enough resources to recover from their financial distress and regularly fall short of financial objectives. If resources are scarce, it would be rational
6381 to pursue the more easily achievable ESG objectives first and maximize the associated remuneration components. In addition to increasing variable remuneration, management could follow alternative career paths (Song and Thakor2006; Zhang2021) or reputational objectives (Jiang etal.2016) that are promoted by achieving the highest possible ESG scores. The priority allocation of scarce resources to achieve ESG objectives is a further argument for a positive relationship between the level of financial distress and ESG scores that could be observed for financially distressed companies. Both the negative and positive relationship between financial distress and ESG scores can be valid simultaneously but for different companies. A negative relationship might be proper for financially healthy companies, and a positive one might be valid for financially distressed companies. The lines of argument from the existing literature can be summarized in the following hypothesis: The relationship between financial distress and ESG scores should have a Ushaped progression, where not only financially healthy companies, but also financially distressed companies should have a high ESG score. While a high ESG score for financially healthy companies underpins the positive relationship between a company's ESG activities and financial performance, a high ESG score for financially distressed companies would call into question the meaningfulness of ESG scores for this group of companies. Therefore, nonparametric regression models must be used to reveal the actual relationship between the level of financial distress and ESG scores (Hastie and Tibshirani1990, 1995; Wood2017) and interpreted accordingly. However, a Ushaped relationship between financial distress and ESG scores has not yet been empirically documented, as the findings to date exclusively use linear regression techniques and thus exclude a possible nonlinear relationship in the data from the outset. 3 | Data, Variables, and Descriptive Statistics 3.1 | Raw Data, Final Dataset, and Final Samples The starting point of our empirical analysis is a merged database of listed US companies. This database includes accounting and market data from the CRSP/Compustat database, ownership data derived from Form 13F and Form 13D(/A) filings, and data on financial distress from BRAINKRUPTCY. Overall, the database of listed US companies includes 71,794 firmyear observations for the period 2003–2022. The bankruptcy prediction model of BRAINKRUPTCY was recently described in detail by Lohmann and Möllenhoff(2023b), and the bankruptcy predictions were applied by Lohmann and Möllenhoff(2023c). The bankruptcy prediction model uses numerous variables proven informative for financial distress and prospective corporate bankruptcy (e.g., the explanatory variables introduced by Campbell etal.2008) and is available for all listed US companies. BRAINKRUPTCY estimates the measure of bankruptcy risk by using a logistic bankruptcy prediction model and a gradientboosting model. The outoftime validations of the annually updated bankruptcy prediction models provide empirical evidence that the classifications based on the measure of bankruptcy risk are accurate. The AUC values for the annual outoftime validations fluctuate within a small range of around 0.9. In comparison, the outoftime validity of a reestimated bankruptcy prediction model that includes the variables of the Z score (Altman1968, 2013) achieves only an AUC value of 0.716 (Lohmann etal.2023). Our main analysis applies the measure of bankruptcy risk based on a logistic bankruptcy prediction model; the measure of bankruptcy risk based on a gradientboosting model is used for a robustness check. In the next step, databases on ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG were matched to the merged database of listed US companies. An empirical analysis of ESG has to consider that the content of ESG and the associated ESG scores are heterogeneous (e.g., Meuer etal.2019). Although ESG rating agencies claim their scores are a reliable indicator of a company's ESG activities, research on ESG rating disagreement shows that the ESG scores of different ESG rating agencies can differ considerably (Chatterji et al. 2016; Dimson et al. 2020; Billio etal.2021; Berg etal.2022). In addition, there are indications of a deliberate distortion of ESG scores by ESG rating agencies if the ESG rating agencies are subject to a conflict of interest and generate significant revenue from ESG scorebased indices (Agrawal etal.2023). Nevertheless, the ESG scores are positively correlated with each other, and therefore, corporate activities such as new ESG investments, additional ESGrelated operations, and the provision of ESG information will likely increase all ESG scores. As a result, valid empirical findings must be reproducible for all material ESG scores, or the empirical results must not contradict each other, at least for different ESG scores. Overall, after matching the databases on ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG to our merged database of listed US companies, we ended up with 34,723 firmyear observations with at least one ESG score. One hundred fortysix firmyear observations (or 0.42% of the 34,723 firmyear observations, respectively) were eliminated due to missing control variables; 142 firmyear observations did not include the dividend yield, and the stock return volatility was missing in four firmyear observations. As a result, the final dataset includes 34,577 firmyear observations with at least one ESG score and all other applied control variables. We further processed the final dataset by winsorizing variables that we consider in the empirical analysis at the 1st and 99th percentiles if a variable has no natural limit and exhibits recognizable outliers. Four samples of four different ESG scores applied as dependent variables were extracted from the final dataset. Table1 provides information on the sample sizes, the number of firmyear observations for which a pair of ESG scores is available, and the correlations between the ESG scores. The ESG scores from Refinitiv, MSCI, and ESG Book form the basis for the larger samples, whereas the ESG scores from Moody's ESG form the basis for the smaller sample. The larger samples include a large share of firmyear observations from the smaller sample. The pairwise correlations of the ESG scores range between 0.254 and 0.673. Very similar correlations can be determined if the
6382 Corporate Social Responsibility and Environmental Management, 2025 quantile ranks of the various ESG scores are taken into account. That empirical finding is consistent with Chatterji etal.(2016), who showed that the pairwise correlations between CSR ratings of different rating agencies are generally low and mostly below 0.5, and Berg etal.(2022), who showed that the average pairwise correlation between prominent ESG scores ranges between 0.38 and 0.71. Figure 1 shows the number of companies for which the ESG scores were available over time. The ESG databases are distinguished from each other in terms of the history of data availability and the number of companies covered. The ESG scores from Refinitiv, MSCI, and ESG Book cover a more significant number of companies. Hence, the ESG scores from Refinitiv, MSCI, and ESG Book also provide data on smaller companies that receive less attention from investors and stakeholders. Nevertheless, Refinitiv, MSCI, ESG Book, and Moody's ESG provide large ESG databases that include longterm ESG information. 3.2 | Applied Variables and Descriptive Statistics The relationship between financial distress and ESG scores is analyzed using linear and nonparametric regression techniques. Four ESG rating agencies provide the dependent variables for the regressions, while a comprehensive set of companyrelated independent variables is used. The ESG data were obtained from Refinitiv, MSCI, ESG Book, and Moody's ESG, and each includes a company's overall ESG score and the respective ESG subfactors. As a result, we apply four overall ESG scores as dependent variables in the principal analysis and 12 ESG subfactors as dependent variables in additional regressions. The independent variable of interest is the measure of bankruptcy risk that BRAINKRUPTCY provides. A company's level of financial distress is indicated by the measure of bankruptcy risk. The measure of bankruptcy risk is the outcome of a logistic bankruptcy prediction model calibrated for listed US companies and can be interpreted as the probability of bankruptcy. The measure of bankruptcy risk ranges in the interval [0, 1]. Table 2 provides the definitions of the applied dependent and independent variables. Besides the dependent variables on ESG scores and the measure of bankruptcy risk, the regression analysis considers 14 control variables on company fundamentals and ownership. Additionally, the regression analysis controls for yearfixed, industryfixed, and companyfixed effects in the panel data. Table3 reports the descriptive statistics on the dependent and independent variables of the four final samples. The mean and the standard deviation of the ESG scores from Moody's ESG are lower as the value range is [0, 10] instead of [0, 100] for all other applied ESG scores. The larger samples based on ESG scores from Refinitiv, MSCI, and ESG Book include a more significant number of firmyear observations from smaller companies than the smaller sample based on ESG scores from Moody's ESG. This difference between the four final samples becomes apparent in the mean of the independent variables. The larger samples are associated with lower mean values of total assets, market value of equity, return on equity, and return on assets. Regarding the other independent variables, the final samples have comparable characteristics and do not differ significantly. Table4 shows the correlations between the independent variables when considering the final dataset of 34,723 firmyear observations. The measure of bankruptcy risk shows no meaningful correlation with any other independent variable. In unreported results, we find similar correlations in the four final samples and for the case where the quantile ranks of the independent variables are considered in the correlation analysis. There are meaningful correlations between two independent variables if the variables are comparable in content. This applies to the pairing total assets and market value of equity, Tobin's Q and market value to book value, and return on equity and return on assets. We address these correlations by examining the variance inflation factors of the independent variables used in the TABLE 1 | Summary statistics on the four samples of four different ESG scores. Sample Firmyear observations Refinitiv MSCI ESG Book Moody's ESG Refinitiv 23,980 0.383 0.489 0.673 MSCI 26,657 17,634 0.254 0.417 ESG Book 25,217 20,170 19,722 0.432 Moody's ESG 8157 7208 7639 7925 Note: The table reports the final sample sizes (Column 2), overlapping firmyear observations with at least two specific ESG scores (lower section in Columns 3–5), and the respective pairwise correlations between the ESG scores (upper section in Columns 4–6). FIGURE 1 | Panel data structure of the final samples of four different ESG scores. This figure shows the number of companies for which ESG scores are available over time after crosschecking against our merged database of listed US companies. Our empirical analysis considers ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. While Refinitiv and ESG Book cover companies for the entire history up to 2003, MSCI and Moody's ESG data are available since 2007.
6383 TABLE 2 | Definitions of the applied dependent and independent variables. Variable Definition ESG variables ESG score Refinitiv(j,a)Company j's ESG score from Refinitiv at the end of the fiscal year a. E subfactor Refinitiv(j,a)Company j's E subfactor from Refinitiv at the end of the fiscal year a. S subfactor Refinitiv(j,a)Company j's S subfactor from Refinitiv at the end of the fiscal year a. G subfactor Refinitiv(j,a)Company j's G subfactor from Refinitiv at the end of the fiscal year a. ESG score MSCI(j,a)Company j's ESG score from MSCI at the end of the fiscal year a. E subfactor MSCI(j,a)Company j's E subfactor from MSCI at the end of the fiscal year a. S subfactor MSCI(j,a)Company j's S subfactor from MSCI at the end of the fiscal year a. G subfactor MSCI(j,a)Company j's G subfactor from MSCI at the end of the fiscal year a. ESG score ESG Book(j,a)Company j's ESG score from ESG Book at the end of the fiscal year a. E subfactor ESG Book(j,a)Company j's E subfactor from ESG Book at the end of the fiscal year a. S subfactor ESG Book(j,a)Company j's S subfactor from ESG Book at the end of the fiscal year a. G subfactor ESG Book(j,a)Company j's G subfactor from ESG Book at the end of the fiscal year a. ESG score Moody's ESG(j,a)Company j's ESG score from Moody's ESG at the end of the fiscal year a. E subfactor Moody's ESG(j,a)Company j's E subfactor from Moody's ESG at the end of the fiscal year a. S subfactor Moody's ESG(j,a)Company j's S subfactor from Moody's ESG at the end of the fiscal year a. G subfactor Moody's ESG(j,a)Company j's G subfactor from Moody's ESG at the end of the fiscal year a. Company variables Measure of bankruptcy risk(j,a)Company j's measure of bankruptcy risk from BRAINKRUPTCY based on a logistic bankruptcy prediction model at the end of the fiscal year a. Total assets(j,a)Company j's total assets in fiscal year a, winsorized at the 99th percentile. B&h stock return(j,a)Company j's buyandhold stock return in fiscal year a, winsorized at the 99th percentile. Stock return volatility(j,a)Company j's annualized stock return volatility in fiscal year a, winsorized at the 99th percentile. Tobin's Q(j,a)Company j's market value of equity plus its book value of total assets, minus its book value of equity, divided by its book value of total assets in fiscal year a, winsorized at the 99th percentile. Market value of equity(j,a)Company j's market value of equity in fiscal year a, winsorized at the 99th percentile. Market value to book value(j,a)Company j's market value of equity divided by its book value of equity in fiscal year a, winsorized at the 99th percentile. Leverage(j,a)Company j's total debt divided by its total assets in fiscal year a, winsorized at the 99th percentile. Return on equity(j,a)Company j's earnings before tax divided by its book value of equity in fiscal year a, winsorized at the 1st and 99th percentiles. Return on assets(j,a)Company j's earnings before interest, tax, depreciation and amortization (EBITDA) divided by its total assets in fiscal year a, winsorized at the 1st and 99th percentiles. CapEx(j,a)Company j's capital expenditures (CapEx) divided by its total assets in fiscal year a, winsorized at the 99th percentile. R&D(j,a)Company j's research and development expenditures (R&D) divided by its total assets in fiscal year a, winsorized at the 99th percentile. (Continues)
6384 Corporate Social Responsibility and Environmental Management, 2025 linear regression models and by checking the robustness of the results by varying the independent variables. However, we apply all control variables in the regressions presented in the results section. 4 | Empirical Results 4.1 | Linear Regression Results The linear regression models estimate a linear relationship between the independent and dependent variables. Table 5 shows the results of the linear regression models LRM1– LRM4 in terms of the estimated regression coefficients of the metric independent variables. The linear regression models apply different ESG scores as dependent variables. In the ESG scores applied, there are differences in the samples used with regard to the available firmyear observations. Based on the applied ESG scores, there are also differences in the applied samples with regard to the available firmyear observations. However, all four linear regression models use the same independent variables, and we estimated all regressions with standard errors clustered at the company level. The linear regression models estimate a negative and statistically significant relationship between the measure of bankruptcy risk and the ESG scores; there is always a negative coefficient. The coefficient of the linear regression model LRM2 is much smaller as the ESG score from MSCI is in the value range [0, 10] instead of [0, 100] for all other applied ESG scores. Based on the linear regression models, we can conclude that, ceteris paribus, the ESG score decreases if a company's financial distress increases. This result is fully consistent with comparable studies using a linear regression technique (e.g., Zheng etal.2019; Boubaker etal. 2020; Aslan et al. 2021; Badayi etal.2021; Truong etal.2025). Overall, the linear regression models show mixed results concerning the metric control variables. The sign of the estimated coefficient or its statistical significance often changes when the results of a specific control variable are compared for the linear regression models LRM1–LRM4. The heterogeneity of the results is to be expected and can be explained by the low correlation between the various ESG scores. As expected, the estimated coefficients differ significantly in magnitude, as the control variables have a different range of values (see the descriptive statistics in Table3). Numerically large control variables have very small coefficients, and numerically small control variables have very large coefficients. We calculated and reviewed the variance inflation factors of the independent variables that are applied in the linear regression models LRM1–LRM4 to check for multicollinearity caused by the correlations between the independent variables. The independent variable Stock return volatility has the highest variance inflation factor in each regression; this lies between 2.00 (LRM2) and 2.41 (LRM4). All other variance inflation factors are always below 2 and mostly close to 1. Multicollinearity is therefore only present to a negligible extent and there is no indication that the linear regression results and the following additive regression analysis are distorted by multicollinearity. 4.2 | Additive Regression Results The basic assumption of a linear regression model is that there are linear relationships whose slopes (i.e., coefficients) are estimated by the model. As we need to know whether there are linear relationships, a linear regression model will produce misleading results if nonlinear relationships exist between the independent and dependent variables. To overcome this shortcoming of a linear regression model, we have to apply additive regression models that estimate the effective relationship in terms of an unspecified function, which could be linear or nonlinear (Stone1985; Hastie and Tibshirani1990). We apply penalized splines (see for an application in business research, e.g., Lohmann and Ohliger2017; Lohmann etal.2023) to model the nonlinear relationships between independent variables, including the measure of bankruptcy risk and a set of control variables, and dependent variables consisting of several ESG scores and their ESG subfactors. We put each spline function in concrete terms by using seven equidistant intervals and polynomials of rank g = 3 for each penalized spline. The minimum of the generalized crossvalidation criterion determines the smoothing parameter (Eilers and Marx1996; Green and Silverman1994). Table 6 shows the results of the additive regression models ARM1–ARM4. The output value for the metric independent variables is the equivalent degree of freedom. The equivalent degree of freedom indicates the nonlinearity in the estimated spline function. An equivalent degree of freedom dff = 1.000 Variable Definition Dividend yield(j,a)Company j's total annual dividend divided by its market value of equity in fiscal year a, winsorized at the 1st and 99th percentiles. Active ownership(j,a)The total active ownership of company j, based on Form 13D(/A) filings at the end of the fiscal year a. Institutional ownership(j,a)The share of institutional ownership of company j, based on Form 13F filings at the end of the fiscal year a. IndustryjCompany j's Standard Industrial Classification (SIC) code: SIC1–SIC9 Note: This table presents and defines the variables we used. We include ESG scores and ESG subfactors from Refinitiv, MSCI, ESG Book, and Moody's ESG as dependent variables. Furthermore, we derived the measure of bankruptcy risk from BRAINKRUPTCY, further company variables from the CRSP and Compustat databases, active ownership from 13D(/A) filings, and institutional ownership from 13F filings. TABLE 2 | (Continued)
6391 there should be a Ushaped relationship between the measure of bankruptcy risk and the variable Shareholderstakeholder orientation that is comparable to the Ushaped relationship between the measure of bankruptcy risk and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Second, there should be no correlation between the variable Shareholderstakeholder orientation and the residuals of the additive regression models ARM1–ARM4 presented in Table 6. Third, the variable Shareholderstakeholder orientation should be able to explain the variance of the ESG scores to a similar extent as the variable Measure of bankruptcy risk and should show a positive, statistically significant, and almost linear relationship with the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Table7 shows the results of the linear regression model LRMSSO and the additive regression model ARMSSO when the variable Shareholderstakeholder orientation is applied as the dependent variable. The linear regression model indicates a negative and statistically significant linear relationship between the measure of bankruptcy risk and a company's stakeholder orientation. On the contrary, the additive regression model reveals a statistically significant nonlinear relationship between the measure of bankruptcy risk and a company's stakeholder orientation, depicted in Figure 5. The nonlinear relationship between the measure of bankruptcy risk and the Shareholderstakeholder orientation can be described by a Ushaped relationship comparable to the relationships between the measure of bankruptcy risk and ESG scores in Figure 2. Financially distressed companies generally exhibit a high stakeholder orientation, although all management efforts should be directed toward preserving the company and protecting the shareholders' equity. The comprehensive stakeholder orientation is consistent with the financially distressed companies' high ESG scores. Table 8 shows the correlations between the variable Shareholderstakeholder orientation and the residuals of the additive regression models ARM1–ARM4 presented in Table6. No anomalies were found in the review of the correlations. The residuals of the additive regression models ARM1–ARM4 are not correlated with the variable Shareholderstakeholder orientation. The variable Shareholderstakeholder orientation does not explain any variance in the ESG scores that is not explained by the additive regression models ARM1–ARM4. When the variable Shareholderstakeholder orientation replaces the variable Measure of bankruptcy risk, and there is a FIGURE 3 | The nonlinear relationship between the market value (MV) of equity and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. This figure shows the statistically significant spline patterns for the relationship between the market value of equity and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. The market value of equity is plotted on the x axis, and the ESG score is plotted on the y axis. Due to the within transformation, the values on the axes represent deviations from the company mean value. The bold black line represents the estimated spline function and the dashed line represents the estimated linear function. The 95% confidence band is shaded gray.
6392 Corporate Social Responsibility and Environmental Management, 2025 statistically significant positive relationship between the variable Shareholderstakeholder orientation and the ESG scores, we provide empirical evidence that financially distressed companies mimic the stakeholderoriented behavior of financially healthy companies by emphasizing stakeholderoriented ESG objectives and intensifying ESGsupporting disclosures. FIGURE 4 | Relationships between the measure of bankruptcy risk (MBR) and the ESG subfactors from Refinitiv, MSCI, ESG Book, and Moody's ESG. This figure shows the spline patterns for the relationship between the measure of bankruptcy risk and the ESG subfactors from Refinitiv, MSCI, ESG Book, and Moody's ESG. The measure of bankruptcy risk is plotted on the x axis, and the ESG subfactors are plotted on the y axis. Due to the within transformation, the values on the axes represent deviations from the company mean value. The bold black line represents the estimated spline function, and the dashed line represents the estimated linear function. The 95% confidence band is shaded gray.
6393 Table 9 shows the results of the additive regression models ARM1SSO–ARM4SSO, where the variable Shareholderstakeholder orientation replaces the variable Measure of bankruptcy risk. There is a statistically significant relationship between the variable Shareholderstakeholder orientation and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Furthermore, the adjusted R2 of the additive regression models ARM1SSO–ARM4SSO is comparable to the adjusted R2 of the additive regression models ARM1–ARM4 presented in Table6. The independent variable Shareholderstakeholder orientation explains the variance of the ESG scores to a similar extent as the independent variable Measure of bankruptcy risk. Figure6 depicts the spline patterns of the independent variable Shareholderstakeholder orientation for the additive regression models ARM1SSO–ARM4SSO that consider four different ESG scores as dependent variables. Figure6 reveals a positive and almost linear relationship between the independent variable Shareholderstakeholder orientation and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. The 95% confidence bands are very narrow and do not allow any other conclusion. As a result, the most plausible interpretation of the TABLE 7 | Results of the linear regression model LRMSSO and the additive regression model ARMSSO with the dependent variable Shareholderstakeholder orientation (SSO). Regression model LRMSSO ARMSSO Dependent variable Shareholderstakeholder orientation Shareholderstakeholder orientation Measure of bankruptcy risk −0.085** 2.892*** Total assets 0.000 2.983*** B&h stock return −0.016*** 1.000*** Stock return volatility 0.064*** 2.773*** Tobin's Q 0.005 1.000** Market value of equity 0.000*** 1.000*** Market value to book value 0.001 1.813** Leverage 0.095*** 1.000*** Return on equity 0.013*1.000** Return on assets −0.022 1.000 CapEx −0.173 1.000** R&D −0.153*1.366** Dividend yield −0.257 1.000* Active ownership 0.064*1.000*** Institutional ownership −0.016 2.864* Yearfixed effects Yes Yes Industryfixed effects Yes Yes Firmfixed effects Yes Yes N30,329 30,329 Adjusted R20.09 0.09 Note: This table shows the results of the linear regression model LRMSSO (i.e., estimated coefficients) and the additive regression model ARMSSO (i.e., equivalent degrees of freedom) where Shareholderstakeholder orientation is applied as the dependent variable. The independent variables include the company variables that are defined in Table2. Both regression models take into account yearfixed, industryfixed, and firmfixed effects. All standard errors of the linear regression model LRMSSO are clustered at the company level. *, **, and *** represent significance levels of 0.05 [or 5%], 0.01 [or 1%], and 0.001 [or 0.1%], respectively. *pvalue < 0.05. **pvalue < 0.01. ***pvalue < 0.001. FIGURE 5 | The Ushaped relationship between the measure of bankruptcy risk (MBR) and the Shareholderstakeholder orientation. This figure shows the statistically significant spline patterns for the relationship between the measure of bankruptcy risk and the Shareholderstakeholder orientation. The measure of bankruptcy risk is plotted on the xaxis, and the Shareholderstakeholder orientation is plotted on the yaxis. Due to the within transformation, the values on the axes represent deviations from the company mean value. The bold black line represents the estimated spline function of the additive regression model ARMSSO, and the dashed line represents the estimated linear function of the linear regression model LRMSSO. The 95% confidence band is shaded gray. TABLE 8 | Correlations between the variable Shareholderstakeholder orientation and the residuals from the additive regression models ARM1–ARM4. Shareholderstakeholderorientation Number of observations Residuals from ARM1 0.017 21,088 Residuals from ARM2 0.012 23,208 Residuals from ARM3 0.016 21,936 Residuals from ARM4 0.029 7406 Note: This table shows the correlations between the variable Shareholderstakeholder orientation and the residuals from the additive regression models ARM1–ARM4.
6394 Corporate Social Responsibility and Environmental Management, 2025 revealed Ushaped relationship between the measure of bankruptcy risk and ESG scores is that companies anticipate their upcoming financial distress and mimic the stakeholderoriented behavior of financially healthy companies by emphasizing ESG objectives and intensifying ESGsupporting disclosures. The empirical finding suggests that financially distressed companies manage their ESG scores upward by intensifying their stakeholderoriented reporting, expressed by the groups addressed in the 10K filings. 5.2 | Further Extensions We carried out three further extensions of the regression analysis, the results of which are worth mentioning. The first extension is a regression analysis that takes into account multiyear capital expenditures and R&D expenditures. The idea behind this extension is that companyrelated measures that lead to a high ESG score could induce financial consequences that increase the measure of bankruptcy risk. Such a relationship should be particularly evident in ESG investments, as these impact a company's financial position more than operational measures or ESGsupporting disclosures. It is a plausible argument that past cash floweffective ESG investments can increase a company's current ESG score while worsening its current financial position. A previous increase in capital and R&D expenditures with a subsequent increase in the ESG score would indicate this causeeffect relationship. We tested this hypothesis by taking into account the capital expenditures of the previous 2 and 3 years when we calculated the independent variable CapEx(j,a) and the R&D expenditures of the previous 2 and 3 years when we calculated the independent variable R&D(j,a). The results of the extended regression analyses are comparable to Table6 and remain almost unchanged, demonstrating the robustness of our research. As a second extension, we applied the CDP (formerly the Carbon Disclosure Project) database to analyze investments to reduce greenhouse gas emissions in more detail. Based on a subsample TABLE 9 | Results of the additive regression models ARM1SSO–ARM4SSO. Regression model ARM1SSO ARM2SSO ARM3SSO ARM4SSO Dependent variable ESG score Refinitiv ESG score MSCI ESG score ESG book ESG score Moody's ESG Shareholderstakeholder orientation 1.374*** 2.920*** 2.535*** 2.797*** Total assets 2.981*** 2.884** 1.921*1.000*** B&h stock return 2.943*** 1.000*** 1.000*** 1.000 Stock return volatility 2.856*1.000** 1.916*** 2.362*** Tobin's Q 1.000 1.000 2.970*** 2.866*** Market value of equity 2.990*** 2.990*** 2.923*** 2.988*** Market value to book value 1.294 1.910*1.000** 1.307 Leverage 1.000*** 1.000 1.830*** 1.760* Return on equity 1.000 1.000*2.519 1.000 Return on assets 1.000 1.000*2.984*** 2.839* CapEx 1.000*** 2.715 2.765*** 1.273*** R&D 2.047*** 1.000 2.931*** 2.258 Dividend yield 2.957*** 2.827** 1.550*** 2.575*** Active ownership 1.000*** 1.000*2.797*** 1.000 Institutional ownership 1.000*** 1.000 1.470*1.000 Yearfixed effects Yes Yes Yes Yes Industryfixed effects Yes Yes Yes Yes Firmfixed effects Yes Yes Yes Yes N21,318 23,450 22,356 7437 Adjusted R20.379 0.101 0.095 0.309 Note: This table shows the results (i.e., equivalent degrees of freedom) of the additive regression models ARM1SSO–ARM4SSO, where different ESG scores are applied as the dependent variable. The independent variables include the company variables that are defined in Table2; the variable Shareholderstakeholder orientation replaces the variable Measure of bankruptcy risk. All additive regression models take into account yearfixed, industryfixed, and firmfixed effects. *, **, and *** represent significance levels of 0.05 [or 5%], 0.01 [or 1%], and 0.001 [or 0.1%], respectively. *pvalue < 0.05. **pvalue < 0.01. ***pvalue < 0.001.
6395 of 2877 analyzable firmyear observations, all conducted regression analyses show that there is no statistically significant relationship between the investments to reduce greenhouse gas emissions and the measure of bankruptcy risk. Consequently, the hypothesis that past ESG investments worsen a company's current financial situation and, at the same time, increase the current ESG scores must be rejected with a probability bordering on certainty. The third extension is a regression analysis that takes into account a company's energy intensity. Companyrelated measures that lead to high ESG scores and an increase in financial distress do not necessarily require ESG investments. However, these measures should mainly affect the environmental subfactor, as the revealed Ushaped relationship is primarily based on this subfactor. An important indicator of a company's environmental footprint is the energy intensity of its valueadded process. If companyrelated measures lead to a high ESG score and an increase in the measure of bankruptcy risk, we would expect a reversed Ushaped or at least negative relationship between a company's energy intensity and the measure of bankruptcy risk. Financially distressed companies should have a smaller environmental footprint regarding their energy intensity. We sourced from the Refinitiv database a score that is based on the total direct and indirect energy consumption in gigajoules divided by net sales or revenue in US dollars. Based on a subsample of 6068 analyzable firmyear observations all conducted regression analyses show that neither the estimated linear nor the nonlinear relationship between a company's energy intensity and the measure of bankruptcy risk is statistically significant. As a result, it can be ruled out with sufficient certainty that financially distressed companies reduce their environmental footprint through real effective measures to an extent that plausibly explains the observed increase in the environmental subfactor. 5.3 | Robustness The regression analysis revealed comparable Ushaped relationships between the measure of bankruptcy risk and four different FIGURE 6 | The linear relationship between the Shareholderstakeholder orientation (SSO) and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. This figure shows the statistically significant spline patterns for the relationship between the Shareholderstakeholder orientation and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Shareholderstakeholder orientation is plotted on the x axis, and the ESG score is plotted on the y axis. The values on the axes are deviations from the company mean value due to the within transformation. For better comparability, the value range of the x axis was limited to [−1, 1]. The bold black line represents the estimated spline function, and the dashed line represents the estimated linear function from unreported linear regression model. The 95% confidence band is shaded gray.
6396 Corporate Social Responsibility and Environmental Management, 2025 ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. The empirical analysis consistently documents that financially distressed companies are associated with higher ESG scores and that this finding is mainly based on the environmental subfactor. We substantiated the robustness of the empirical results in great detail by varying the applied data and the number of intervals for which each spline function was estimated. A critical design element of additive regression models is the number of intervals for which each spline function was estimated. First, we checked the appropriateness of the applied number of intervals by performing the test recommended by Wood(2017, section5.9). The test is based on randomly sampled data and produces a test statistic that may widely vary if the test is replicated. Therefore, we repeated the test 100 times for the additive regression models ARM1–ARM4 to make a valid statement about the robustness of the additive regression models. The evaluation of the test statistics shows that the applied number of intervals is appropriate to model the relationship between the measure of bankruptcy risk and the ESG scores. Nevertheless, we estimated the additive regression models using a larger number of intervals for all metricindependent variables. The Ushaped relationship between the measure of bankruptcy risk and the ESG scores is always recognizable and robust against any reasonable change in the number of intervals. The applied final dataset and the derived final samples include firmyear observations from 2003 to 2022. We repeated the complete analysis for the periods 2003–2019 and 2011–2022. The analysis for the period 2003–2019 excluded from the dataset the last 3 years in which the COVID19 pandemic occurred. The COVID19 pandemic and the associated governmental measures to get the pandemic under control possibly led to distortions in the applied data. An additional advantage of the shorter observation period 2003–2019 is that we do not consider ESG scores from Refinitiv, which are still changing. If there is new ESGrelevant information from an earlier year, the ESG scores from Refinitiv are adjusted. This applies in particular to ESG scores of the more recent observation years. In the analysis for the period 2011–2022, the first 8 years in which the ESG ratings began were excluded from the dataset. Initial valuation difficulties and valuation inconsistencies possibly led to distortions in the early ESG scores. All linear and additive regressions with the ESG scores and the ESG subfactors as independent variables show similar relationships and thus confirm the robustness of the empirical results. Another variation in the samples used was that we created a new sample in which the ESG scores from Refinitiv, MSCI, and ESG Book are available in all firmyear observations. This combined sample comprises a total of 16,184 firmyear observations. All linear and additive regressions with the ESG scores and the ESG subfactors as independent variables show similar relationships. The empirical results provide additional evidence for the robustness of the Ushaped relationship between the measure of bankruptcy risk and the ESG scores. We examined the robustness of the regression models by varying the independent variable Measure of bankruptcy risk, and highly correlated independent variables. Thereby, the independent variable Measure of bankruptcy risk is varied by applying an alternative measure of bankruptcy risk that results from a gradientboosting model rather than a logistic bankruptcy prediction model. The results of the regression models are robust against this variation of the measure of bankruptcy risk. Furthermore, we observed that the Ushaped relationships between the measure of bankruptcy risk and the ESG scores and their subfactors (mainly, the environmental subfactor) are more pronounced if the number of control variables is reduced and/ or if yearfixed and industryfixed effects are not taken into account. However, as we do not want to exaggerate the empirical results, we have only presented the regression analyses with 14 independent (control) variables as well as yearfixed, industryfixed, and firmfixed effects. We varied the procedure several times to refine the final dataset and analyze whether the applied procedure influenced the empirical results. We replaced missing values using the knearest neighbors algorithm, winsorized outliers at the 2nd and 98th percentiles, and eliminated outliers. The review showed that the empirical results are robust and not affected by the applied procedure to refine the final dataset. 6 | Conclusion 6.1 | Insights The analysis provides empirical evidence of a Ushaped relationship between the level of financial distress measured by the measure of bankruptcy risk from BRAINKRUPTCY and the ESG scores from Refinitiv, MSCI, ESG Book, and Moody's ESG. Above a certain threshold, financially distressed companies are associated with high ESG scores. The study not only provides a concrete answer to the empirical relationship between financial distress and ESG scores, but also works out a concrete cause– effect relationship between financial distress and ESG scores. Severe financial distress is the cause, and a high ESG score is the effect. Financially distressed companies intensify their costeffective ESG activities, such as ESGsupporting disclosures and manage their ESG scores upward. The causeeffect relationship between financial distress and ESG scores has been elaborated by first invalidating the reverse causeeffect relationship, stating that measures leading to a high ESG score are the cause and financial distress is the financial consequence of these measures based on empirical evidence. The interpretation of the narrow confidence bands of the additive regression models ARM1–ARM4, the unchanged regression analyses with multiyear capital expenditures and R&D expenditures, and the empirically unverifiable relationship between the measure of bankruptcy risk and a company's energy intensity are significant arguments against a reverse causeeffect relationship. Subsequently, empirical evidence was presented to support the actual causeeffect relationship between financial distress and ESG scores. This empirical evidence consists of the interpretation of the narrow confidence bands of the additive regression models ARM1–ARM4 and the regression analyses with the variable Shareholderstakeholder orientation. Overall, there is empirical evidence of a pronounced stakeholder orientation for the group of financially distressed companies.
6397 In addition, further insights were gained, which are fundamentally neutral in terms of a causeeffect relationship. Particularly, the Ushaped relationship can be observed regarding the environmental subfactor of the analyzed ESG scores. Furthermore, there is a consistent finding that a negative relationship exists between the market value of equity and the ESG score when the market value of equity is below a company's mean value, indicating that companies with a shrunken market value of equity are associated with higher ESG scores. An overall picture can be put together from the individual empirical results. Financially distressed companies systematically manage their ESG scores upward by reinforcing the perception of their ESG activities and intensifying ESGsupporting disclosures. We can rule out with a probability bordering on certainty that ESG investments or other ESGrelated operational measures increase ESG scores and cause financial distress as a side effect. The intensification of ESGsupporting disclosures relates primarily to the environmental subfactor, meaning that the group of financially distressed companies can speak of systematic greenwashing. The motivation for such behavior could be based on the need for a lower cost of capital and improved financing conditions that can be achieved through higher ESG scores, the management's desire to divert attention from financial failure, or the managerial incentive system which rewards ESG objectives that may be easier to achieve than financial objectives. In a followup study, the motivations behind the ESG disclosure policies of financially distressed companies need to be analyzed in more detail. Presumably, however, a mixture of arguments will be responsible for a Ushaped relationship between the level of financial distress and ESG scores. One limitation of the analysis is that only listed US companies are taken into account. Although by far the largest stock market in the world is considered, this does not guarantee that the empirical results are transferable to other countries. However, it can be assumed with a high degree of probability that a similar Ushaped relationship between financial distress and ESG scores should also be observed in other western industrialized countries. In this context, the question arises as to whether an internationally composed sample could not also be empirically investigated. In our view, there are two main reasons against such an approach: First, the measure of bankruptcy risk based on a bankruptcy prediction model can only be accurately estimated for a specific country, as the legal definition of bankruptcy differs across countries. Second, ESG regulations, ESG practices and corporate culture may differ in different countries, which also has an impact on ESG ratings. For these two reasons, there is a tradeoff between analyzing a consistent and unbiased sample of a single country and analyzing an inconsistent and potentially biased sample of multiple countries. We have chosen to analyze a consistent and unbiased sample of listed US companies. 6.2 | Implications The management of ESG scores by the group of financially distressed companies reduces the validity and credibility of ESG scores and makes them less reliable. Due to the observed management of ESG scores by the group of financially distressed companies, it is imperative to consider a company's financial situation when interpreting ESG scores. If ESG scores are viewed in isolation, they may lead to an incorrect evaluation of financially distressed companies. As a result, the methodology for determining ESG scores in particular and ESG performance in general must be scrutinized. A methodology must be designed so that extensive distortions, such as those practiced by the group of financially distressed companies, are not possible or are at least shown transparently. On the other hand, the incentives that lead to such opportunistic behavior by the group of financially distressed companies must be examined more closely. The Ushaped relationship between the measure of bankruptcy risk and four different ESG scores shows some tension between ESG and the sustainable financial situation of a company when the company is financially distressed. Companies have to be sustainable in two respects: First, a company's activities have to be sustainable in terms of environmental, social, and governance aspects, measured by ESG scores. That is an overall objective of the company's stakeholders. Second, every company strives to ensure its continued existence and not become bankrupt. The viability of a company can be described by the measure of bankruptcy risk, which can be interpreted as the probability of bankruptcy if a logistic bankruptcy prediction model is applied. The company's shareholders have the objective that the company is financially healthy and that the continued existence of the company is not jeopardized. As a result, the company has to take into account ESG and its financial sustainability as two objectives at the same time. A company's ESG activities and sustainable financial situation can be measured by two separate performance measures, such as an ESG score and the measure of bankruptcy risk. However, this separation, particularly, leads to two issues. First, the ranking of companies in relation to one of the two performance measures is inconsistent, as we observed increasing ESG scores when a company becomes more financially distressed. A consistent evaluation of a company's sustainability concerning both ESG and its financial situation requires a single measure of its overall sustainability, including the financial situation. Second, two company objectives and the associated performance measures, such as an ESG score and the measure of bankruptcy risk, may lead to severe incentive problems if variable remuneration or longterm career development depends on these performance measures. If both performance measures are additively linked, measures are first taken to increase the performance measure that leads to the most significant benefit with the least use of resources. It is plausible that a high ESG score can be achieved by corporate activities such as extensive ESG reporting rather than actual ESG investments. In that case, increasing an ESG score can be achieved easier than improving a company's financial situation, and a rational decisionmaker is likely to focus first on a company's ESG score. A company's ESG activities and sustainable financial situation can be measured by one overall sustainability score if the ESG score and the measure of bankruptcy risk are multiplicatively linked. If the measure of bankruptcy risk is interpreted as the probability of bankruptcy, the product of the ESG score and the
6398 Corporate Social Responsibility and Environmental Management, 2025 complementary probability that the company remains solvent can be interpreted as the expected ESG score. The expected ESG score assumes that the current ESG score represents the future period and additionally weights that ESG score with the complementary probability that the company remains solvent. This overall sustainability score provides a sustainability evaluation of a company that will likely become bankrupt in the foreseeable future. If a company is financially distressed and exhibits an increased measure of bankruptcy risk, the overall sustainability score is much smaller than the associated ESG score. The overall sustainability score of a financially distressed company will be decreased as a company on the verge of bankruptcy is not sustainable in any criteria. The empirical analysis demonstrates a Ushaped relationship between the measure of bankruptcy risk and ESG scores. This finding leads to the implication that the environmental, social, and governance impacts of a company that are condensed in the ESG score and the financial sustainability that the measure of bankruptcy risk must be considered together, which applies, in particular, to financially distressed companies. If ESG scores are viewed in isolation, they may lead to an incorrect evaluation of financially distressed companies. In order to ensure the validity and informative value of ESG scores, including ESG scores for financially distressed companies, the ESG scores should be adjusted by incorporating the probability of bankruptcy or the complementary probability that the company remains solvent. This task is urgent, as an increasing proportion of listed US companies are experiencing financial difficulties (Lohmann and Möllenhoff2023b). Notwithstanding this, the present study shows that stakeholders and ESG rating agencies should not be blinded by extensive ESG disclosure and that shareholders of financially distressed companies must insist that boards address financial matters first. Acknowledgements The authors have nothing to report. Open Access funding enabled and organized by Projekt DEAL. References Agarwala, N., S. Jana, and T. N. Sahu. 2024. “ESG Disclosures and Corporate Performance: A NonLinear and Disaggregated Approach.” Journal of Cleaner Production 437: 140517. https:// doi. org/ 10. 1016/j. jclep ro. 2023. 140517. Agrawal, S., L. Y. Liu, S. Rajgopal, S. A. Sridharan, Y. Yan, and T. L. Yohn. 2023. “ESG Ratings of ESG Index Providers.” Columbia Business School Research Paper No. 4468531, Accessed October 30, 2024. https:// doi. org/ 10. 2139/ ssrn. 4468531. Almubarak, W. I., K. Chebbi, and M. A. Ammer. 2023. “Unveiling the Connection Among ESG, Earnings Management, and Financial Distress: Insights From an Emerging Market.” Sustainability 15, no. 16: 12348. https:// doi. org/ 10. 3390/ su151 612348. Altman, E. I. 1968. “Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy.” Journal of Finance 23, no. 4: 589– 609. https:// doi. org/ 10. 2307/ 2978933. Altman, E. I. 2013. “Predicting Financial Distress of Companies: Revisiting the ZScore and ZETA Models.” In Handbook of Research Methods and Applications in Empirical Finance, edited by A. R. Bell, C. Brooks, and M. Prokopczuk, 428–456. Edward Elgar Publishing. https:// doi. org/ 10. 4337/ 97808 57936 097. 00027 . AronDine, S., J. Beutel, M. Piazzesi, and M. Schneider. 2024. “Household Climate Finance: Theory and Survey Data on Safe and Risky Green Assets.” NBERWorking Paper 32615. National Bureau of Economic Research, Accessed June 2024. http:// www. nber. org/ papers/ w32615. Aslan, A., L. Poppe, and P. Posch. 2021. “Are Sustainable Companies More Likely to Default? Evidence From the Dynamics Between Credit and ESG Ratings.” Sustainability 13, no. 15: 8568. https:// doi. org/ 10. 3390/ su131 58568 . Badayi, S. A., B. T. Matemilola, B. A. An, and L. W. Theng. 2021. “Does Corporate Social Responsibility Influence Firm Probability of Default?” International Journal of Finance and Economics 26, no. 3: 3377–3395. https:// doi. org/ 10. 1002/ ijfe. 1966. Baker, A. C., D. F. Larcker, C. G. McClure, D. Saraph, and E. M. Watts. 2024. “Diversity Washing.” Journal of Accounting Research 62, no. 5: 1661–1709. https:// doi. org/ 10. 1111/ 1475679X. 12542 . Balcaen, S., and H. Ooghe. 2006. “35 Years of Studies on Business Failure: An Overview of the Classic Statistical Methodologies and Their Related Problems.” British Accounting Review 38, no. 1: 63–93. https:// doi. org/ 10. 1016/j. bar. 2005. 09. 001. Beaver, W. H., M. F. McNichols, and J. W. Rhie. 2005. “Have Financial Statements Become Less Informative? Evidence From the Ability of Financial Ratios to Predict Bankruptcy.” Review of Accounting Studies 10, no. 1: 93–122. https:// doi. org/ 10. 1007/ s1114 200463419. Bebchuk, L. A., and R. Tallarita. 2020. “The Illusory Promise of Stakeholder Governance.” Cornell Law Review 106, no. 1: 91–177. ht tps: // li vec o rne ll - la w - r e vi ew. pant h eonsi t e. i o/ wpc o nte nt/ uplo a d s/ 2021/ 02/ TheIllus oryPromi seofStake holde rGover nance. pdf. Bellovary, J. L., D. E. Giacomino, and M. D. Akers. 2007. “A Review of Bankruptcy Prediction Studies: 1930 to Present.” Journal of Financial Education 33, no. 4: 1–43. https:// www. jstor. org/ stable/ 41948574. Berg, F., J. F. Kölbel, and R. Rigobon. 2022. “Aggregate Confusion: The Divergence of ESG Ratings.” Review of Finance 26, no. 6: 1315–1344. https:// doi. org/ 10. 1093/ rof/ rfac033. Bhagat, S. 2022. “An Inconvenient Truth About ESG Investing.” Harvard Business Review, Accessed March 31, 2022. https:// hbr. org/ 2022/ 03/ anincon venie nttruth - about - esginves ting. Billio, M., M. Costola, I. Hristova, C. Latino, and L. Pelizzon. 2021. “Inside the ESG Ratings: (Dis)agreement and Performance.” Corporate Social Responsibility and Environmental Management 28, no. 5: 1426– 1445. https:// doi. org/ 10. 1002/ csr. 2177. Billio, M., A. Fitzpatrick, C. Latino, and L. Pelizzon. 2024. “Unpacking the ESG Ratings: Does One Size Fit All?” SAFE Working Paper 415. Leibniz Institute for Financial Research SAFE: Sustainable Architecture for Finance in Europe, Accessed March 1, 2024. https:// doi. org/ 10. 2139/ ssrn. 4742445. Bloomfield, R. J. 2002. “The “Incomplete Revelation Hypothesis” and Financial Reporting.” Accounting Horizons 16, no. 3: 233–243. https:// doi. org/ 10. 2308/ acch. 2002. 16.3. 233. Bloomfield, R. J. 2008. “Discussion of “Annual Report Readability, Current Earnings, and Earnings Persistence”.” Journal of Accounting and Economics 45, no. 2–3: 248–252. https:// doi. org/ 10. 1016/j. jacce co. 2008. 04. 002. Boubaker, S., A. Cellier, R. Manita, and A. Saeed. 2020. “Does Corporate Social Responsibility Reduce Financial Distress Risk?” Economic Modelling 91: 835–851. https:// doi. org/ 10. 1016/j. econm od. 2020. 05. 012. Buallay, A., J. AlAjmi, and E. Barone. 2022. “Sustainability Engagement's Impact on Tourism Sector Performance: Linear and Nonlinear Models.” Journal of Organizational Change Management 35, no. 2: 361–384. https:// doi. org/ 10. 1108/ JOCM1020200308.
6399 Buertey, S., E. J. Sun, J. S. Lee, and J. Hwang. 2020. “Corporate Social Responsibility and Earnings Management: The Moderating Effect of Corporate Governance Mechanisms.” Corporate Social Responsibility and Environmental Management 27, no. 1: 256–271. https:// doi. org/ 10. 1002/ csr. 1803. Busch, T., and G. Friede. 2018. “The Robustness of the Corporate Social and Financial Performance Relation: A SecondOrder MetaAnalysis.” Corporate Social Responsibility and Environmental Management 25, no. 4: 583–608. https:// doi. org/ 10. 1002/ csr. 1480. Bushee, B. J., I. D. Gow, and D. J. Taylor. 2018. “Linguistic Complexity in Firm Disclosures. Obfuscation or Information?” Journal of Accounting Research 56, no. 1: 85–121. https:// doi. org/ 10. 1111/ 1475679X. 12179 . Campbell, J. Y., J. Hilscher, and J. Szilagyi. 2008. “In Search of Distress Risk.” Journal of Finance 63, no. 6: 2899–2939. https:// doi. org/ 10. 1111/j. 15406261. 2008. 01416. x. Chang, W.- J., R. M. Hayes, and S. A. Hillegeist. 2015. “Financial Distress Risk and New CEO Compensation.” Management Science 62, no. 2: 479–501. https:// doi. org/ 10. 1287/ mnsc. 2014. 2146. Chatterji, A., R. Durand, D. Levine, and S. Touboul. 2016. “Do Ratings of Firms Converge? Implications for Managers, Investors and Strategy Researchers.” Strategic Management Journal 37, no. 8: 1597–1614. https:// doi. org/ 10. 1002/ smj. 2407. Chava, S. 2014. “Environmental Externalities and Cost of Capital.” Management Science 60, no. 9: 2223–2247. https:// doi. org/ 10. 1287/ mnsc. 2013. 1863. Cohen, G. 2023. “ESG Risks and Corporate Survival.” Environment Systems and Decisions 43, no. 1: 16–21. https:// doi. org/ 10. 1007/ s1066 902209886 - 8. Cohen, S., I. Kadach, G. Ormazabal, and S. Reichelstein. 2023. “Executive Compensation Tied to ESG Performance: International Evidence.” Journal of Accounting Research 61, no. 3: 805–853. https:// doi. org/ 10. 1111/ 1475679X. 12481 . DeFond, M. L., and J. Jiambalvo. 1994. “Debt Covenant Violation and Manipulation of Accruals.” Journal of Accounting and Economics 17, no. 1–2: 145–176. https:// doi. org/ 10. 1016/ 01654101(94) 90008 - 6. Del Mar MirasRodríguez, M., A. CarrascoGallego, and B. EscobarPérez. 2015. “Are Socially Responsible Behaviors Paid Off Equally? A CrossCultural Analysis.” Corporate Social Responsibility and Environmental Management 22, no. 4: 237–256. https:// doi. org/ 10. 1002/ csr. 1344. Dimson, E., P. Marsh, and M. Staunton. 2020. “Divergent ESG Ratings.” Journal of Portfolio Management 47, no. 1: 75–87. https:// doi. org/ 10. 3905/ jpm. 2020.1. 175. Drempetic, S., C. Klein, and B. Zwergel. 2020. “The Influence of Firm Size on the ESG Score: Corporate Sustainability Ratings Under Review.” Journal of Business Ethics 167, no. 2: 333–360. https:// doi. org/ 10. 1007/ s1055 101904164 - 1. Eilers, P. H. C., and B. D. Marx. 1996. “Flexible Smoothing With BSplines and Penalties.” Statistical Science 11, no. 2: 89–121. https:// doi. org/ 10. 1214/ ss/ 10384 25655 . Eisfeldt, A. L., and A. A. Rampini. 2007. “New or Used? Investment With Credit Constraints.” Journal of Monetary Economics 54, no. 8: 2656–2681. https:// doi. org/ 10. 1016/j. jmone co. 2007. 06. 030. El Khoury, R., N. Nasrallah, and B. Alareeni. 2023. “ESG and Financial Performance of Banks in the MENAT Region: ConcavityConvexity Patterns.” Journal of Sustainable Finance & Investment 13, no. 1: 406– 430. https:// doi. org/ 10. 1080/ 20430 795. 2021. 1929807. Flugum, R., and M. E. Souther. 2025. “Stakeholder Value: A Convenient Excuse for Underperforming Managers?” Journal of Financial and Quantitative Analysis 60, no. 1: 135–168. https:// doi. org/ 10. 1017/ S0022 10902 3001308. Friede, G., T. Buch, and A. Bassen. 2015. “ESG and Financial Performance: Aggregated Evidence From More Than 2000 Empirical Studies.” Journal of Sustainable Finance & Investment 5, no. 4: 210–233. https:// doi. org/ 10. 1080/ 20430 795. 2015. 1118917. GallardoVázquez, D., M. J. BarrosoMéndez, M. L. PajueloMoreno, and J. SánchezMeca. 2019. “Corporate Social Responsibility Disclosure and Performance: A MetaAnalytic Approach.” Sustainability 11, no. 4: 1115–1148. https:// doi. org/ 10. 3390/ su110 41115 . Gargouri, M., R. Shabou, and C. Francoeur. 2010. “The Relationship Between Corporate Social Performance and Earnings Management.” Canadian Journal of Administrative Sciences 27, no. 4: 320–334. https:// doi. org/ 10. 1002/ cjas. 178. Gormsen, N. J., K. Huber, and S. Oh. 2023. “Climate Capitalists.” SSRN Working Paper, Accessed February 22, 2023. https:// doi. org/ 10. 2139/ ssrn. 4366445. Gow, I. D., D. F. Larcker, and P. C. Reiss. 2016. “Causal Inference in Accounting Research.” Journal of Accounting Research 54, no. 2: 477–523. https:// doi. org/ 10. 1111/ 1475679X. 12116 . Green, D., and B. Vallee. 2024. “Can Finance Save the World? Measurement and Effects of Coal Divestment Policies by Banks.” Working Paper. Proceedings of the EUROFIDAI-ESSEC Paris December Finance Meeting 2024, Accessed November 11, 2024. https:// doi. org/ 10. 2139/ ssrn. 4090974. Green, P. J., and B. W. Silverman. 1994. Nonparametric Regression and Generalized Linear Models: A Roughness Penalty Approach. Chapman and Hall/CRC. Hang, M., J. GeyerKlingeberg, and A. W. Rathgeber. 2019. “It Is Merely a Matter of Time: A MetaAnalysis of the Causality Between Environmental Performance and Financial Performance.” Business Strategy and the Environment 28, no. 2: 257–273. https:// doi. org/ 10. 1002/ bse. 2215. Hastie, T. J., and R. J. Tibshirani. 1990. Generalized Additive Models. Chapman and Hall/CRC. https:// doi. org/ 10. 1002/ sim. 47801 10717 . Hastie, T. J., and R. J. Tibshirani. 1995. “Generalized Additive Models for Medical Research.” Statistical Methods in Medical Research 4, no. 3: 187–196. https:// doi. org/ 10. 1177/ 09622 80295 00400302. Hill, C. W. L., and T. M. Jones. 1992. “StakeholderAgency Theory.” Journal of Management Studies 29, no. 2: 131–154. https:// doi. org/ 10. 1111/j. 14676486. 1992. tb006 57. x. HolderWebb, L., and J. R. Cohen. 2007. “The Association Between Disclosure, Distress and Failure.” Journal of Business Ethics 75, no. 3: 301–314. https:// doi. org/ 10. 1007/ s1055 100692547. Hong, H., J. Kubik, and J. A. Scheinkman. 2012. “Financial Constraints on Corporate Goodness.” SSRN Working Paper, Accessed April 12, 2012. https:// doi. org/ 10. 2139/ ssrn. 1734164. Hoobler, J. M., C. R. Masterson, S. M. Nkomo, and E. J. Michel. 2018. “The Business Case for Women Leaders: MetaAnalysis, Research Critique, and Path Forward.” Journal of Management 44, no. 6: 2473– 2499. https:// doi. org/ 10. 1177/ 01492 06316 628643. Hou, M., H. Liu, P. Fan, and Z. Wei. 2016. “Does CSR Practice Pay Off in East Asian Firms? A MetaAnalytic Investigation.” Asia Pacific Journal of Management 33, no. 1: 195–228. https:// doi. org/ 10. 1007/ s1049 001594312. Jaggi, B., and P. Lee. 2002. “Earnings Management Response to Debt Covenant Violations and Debt Restructuring.” Journal of Accounting, Auditing and Finance 17, no. 4: 295–324. https:// doi. org/ 10. 1177/ 01485 58X02 01700402. Jensen, M. C., and W. H. Meckling. 1976. “Theory of the Firm. Managerial Behaviour, Agency Costs and Ownership Structure.” Journal of Financial Economics 3, no. 4: 305–360. https:// doi. org/ 10. 1016/ 0304405X(76) 90026 - X.
6400 Corporate Social Responsibility and Environmental Management, 2025 Jeong, S.- H., and D. A. Harrison. 2017. “Glass Breaking, Strategy Making, and Value Creating: MetaAnalytic Outcomes of Women as CEOs and TMT Members.” Academy of Management Journal 60, no. 4: 1219–1252. https:// doi. org/ 10. 5465/ amj. 2014. 0716. Jiang, W., H. Wan, and S. Zhao. 2016. “Reputation Concerns of Independent Directors: Evidence From Individual Director Voting.” Review of Financial Studies 29, no. 3: 655–696. https:// doi. org/ 10. 1093/ rfs/ hhv125. Jones, S. 2017. “Corporate Bankruptcy Prediction. A High Dimensional Analysis.” Review of Accounting Studies 22, no. 4: 1366–1422. https:// doi. org/ 10. 1007/ s1114 201794071. Jordaan, L. A., M. De Klerk, and C. J. Villiers. 2018. “Corporate Social Responsibility and Earnings Management of South African Companies.” South African Journal of Economic and Management Sciences 21, no. 1: 99–112. https:// doi. org/ 10. 4102/ sajems. v21i1. 1849. Kacperczyk, M. T., and J.- L. Peydro. 2024. “Carbon Emissions and the BankLending Channel.” Finance Working Paper (991/2024). European Corporate Governance Institute, Accessed July 2024. https:// www. ecgi. g lobal / s y st em / fi le s/ 20 2409/2_ ka c p e r c zy k - _ c a r b o nemi s s ion s - andthebanklendi ngchann el. pdf. Karpoff, J. M. 2021. “On a Stakeholder Model of Corporate Governance.” Financial Management 50, no. 2: 321–343. https:// doi. org/ 10. 1111/ fima. 12344 . Kirk, M., and J. Vincent. 2014. “Professional Investor Relations Within the Firm.” Accounting Review 89, no. 4: 1421–1452. https:// doi. org/ 10. 2308/ accr50724 . Kuk, K., R. Newbury, M. Montecelos, and X. Zhu. 2023. “Global Study on ESG Incentives in Executive Compensation.” WTW. https:// www. wtwc o. c om / enmy/ insi g ht s/ 20 2 3/ 01/ g loba lst udy - ones gi ncen t iv e s - inexecu tivecompe nsation. Lang, M., and R. Lundholm. 1993. “CrossSectional Determinants of Analyst Ratings of Corporate Disclosures.” Journal of Accounting Research 31, no. 2: 246–271. https:// doi. org/ 10. 2307/ 2491273. Li, F. 2010. “The Information Content of ForwardLooking Statements in Corporate Filings: A Naїve Bayesian Machine Learning Approach.” Journal of Accounting Research 48, no. 5: 1049–1102. https:// doi. org/ 10. 1111/j. 1475679X. 2010. 00382. x. Lisin, A., A. Kushnir, A. G. Koryakov, N. Fomenko, and T. Shchukina. 2022. “Financial Stability in Companies With High ESG Scores: Evidence From North America Using the Ohlson OScore.” Sustainability 14, no. 1: 479. https:// doi. org/ 10. 3390/ su140 10479 . Lohmann, C., and S. Möllenhoff. 2023a. “Dark Premonitions: PreBankruptcy Investor Attention and Behavior.” Journal of Banking & Finance 151: 106853. https:// doi. org/ 10. 1016/j. jbank fin. 2023. 106853. Lohmann, C., and S. Möllenhoff. 2023b. “How Do Bankruptcy Risk Estimations Change in Time? Empirical Evidence From Listed US Companies.” Finance Research Letters 58: 104389. https:// doi. org/ 10. 1016/j. frl. 2023. 104389. Lohmann, C., and S. Möllenhoff. 2023c. “The Bankruptcy Risk Matrix as a Tool for Interpreting the Outcome of Bankruptcy Prediction Models.” Finance Research Letters 50: 103851. https:// doi. org/ 10. 1016/j. frl. 2023. 103851. Lohmann, C., S. Möllenhoff, and T. Ohliger. 2023. “Nonlinear Relationships in Bankruptcy Prediction and Their Effect on the Profitability of Bankruptcy Prediction Models.” Journal of Business Economics 93, no. 9: 1661–1690. https:// doi. org/ 10. 1007/ s1157 302201130 - 8. Lohmann, C., and T. Ohliger. 2017. “Nonlinear Relationships and Their Effect on Bankruptcy Prediction.” Schmalenbach Business Review 18, no. 3: 261–287. https:// doi. org/ 10. 1007/ s4146 40170034y. Lohmann, C., and T. Ohliger. 2020. “Bankruptcy Prediction and the Discriminatory Power of Annual Reports: Empirical Evidence From Financially Distressed German Companies.” Journal of Business Economics 90, no. 1: 137–172. https:// doi. org/ 10. 1007/ s1157 301900938 - 1. LópezArceiz, F. J., A. J. Bellostas, and P. Rivera. 2018. “Twenty Years of Research on the Relationship Between Economic and Social Performance: A MetaAnalysis Approach.” Social Indicators Research 140, no. 2: 453–484. https:// doi. org/ 10. 1007/ s1120 501717911. LopezdeSilanes, F., J. McCahery, and P. Pudschedl. 2020. “ESG Performance and Disclosure: A CrossCountry Analysis.” Singapore Journal of Legal Studies 2020, no. 1: 217–241. https:// www. jstor. org/ stable/ 27032607. Lu, W., and M. E. Taylor. 2016. “Which Factors Moderate the Relationship Between Sustainability Performance and Financial Performance? A MetaAnalysis Study.” Journal of International Accounting Research 15, no. 1: 1–15. https:// doi. org/ 10. 2308/ jiar51103 . Lyon, T. P., and J. W. Maxwell. 2011. “Greenwash: Corporate Environmental Disclosure Under Threat of Audit.” Journal of Economics and Management Strategy 20, no. 1: 3–41. https:// doi. org/ 10. 1111/j. 15309134. 2010. 00282. x. Ma, S., J. Murfin, and R. Pratt. 2022. “Young Firms, Old Capital.” Journal of Financial Economics 146, no. 1: 331–356. https:// doi. org/ 10. 1016/j. jfine co. 2021. 09. 017. Marquis, C., M. W. Toffel, and Y. Bird. 2016. “Scrutiny, Norms and Selective Disclosure: A Global Study of Greenwashing.” Organization Science 27, no. 2: 483–504. https:// doi. org/ 10. 1287/ orsc. 2015. 1039. MartínezFerrero, J., S. Banerjee, and I. M. GarcíaSánchez. 2016. “Corporate Social Responsibility as a Strategic Shield Against Costs of Earnings Management Practices.” Journal of Business Ethics 133, no. 2: 305–324. https:// doi. org/ 10. 1007/ s1055 10142399x. Mayew, W. J., M. Sethuraman, and M. Venkatachalam. 2015. “MD&A Disclosure and the Firm's Ability to Continue as a Going Concern.” Accounting Review 90, no. 4: 1621–1651. https:// doi. org/ 10. 2308/ accr50983 . Meuer, J., J. Koelbel, and V. H. Hoffmann. 2019. “On the Nature of Corporate Sustainability.” Organization & Environment 33, no. 3: 319– 341. https:// doi. org/ 10. 1177/ 10860 26619 850180. Muttakin, M. B., A. Khan, and M. Azim. 2015. “Corporate Social Responsibility Disclosures and Earnings Quality: Are They a Reflection of Managers' Opportunistic Behavior?” Managerial Auditing Journal 30, no. 3: 277–298. https:// doi. org/ 10. 1108/ MAJ0220140997. Myers, S. C. 1977. “Determinants of Corporate Borrowing.” Journal of Financial Economics 5, no. 2: 147–175. https:// doi. org/ 10. 1016/ 0304405X(77) 90015 - 0. Naimy, V., R. El Khoury, and S. Iskandar. 2021. “ESG Versus Corporate Financial Performance: Evidence From East Asian Firms in the Industrials Sector.” Sustainable Economics 39, no. 3: 1–24. https:// doi. org/ 10. 25115/ eea. v39i3. 4457. Nollet, J., G. Filis, and E. Mitrokostas. 2016. “Corporate Social Responsibility and Financial Performance: A NonLinear and Disaggregated Approach.” Economic Modelling 52, no. B: 400–407. https:// doi. org/ 10. 1016/j. econm od. 2015. 09. 019. Nuber, C., P. Velte, and J. Hörisch. 2020. “The Curvilinear and TimeLagging Impact of Sustainability Performance on Financial Performance: Evidence From Germany.” Corporate Social Responsibility and Environmental Management 27, no. 1: 232–243. https:// doi. org/ 10. 1002/ csr. 1795. Pasko, O., F. Chen, N. Proskurina, R. Mao, V. Gryn, and I. Pushkar. 2021. “Are Corporate Social Responsibility Active Firms Less Involved in Earnings Management? Empirical Evidence From China.” Business: Theory and Practice 22, no. 2: 504–516. https:// doi. org/ 10. 3846/ btp. 2021. 14940 .