The Interaction Effects of Board Independence and Digital Transformation on Environmental, Social, and Governance Performance: Complementary or Substitutive?

Abstract

Both board independence (BI) and digital transformation (DT) play important roles in promoting Environmental, Social, and Governance (ESG) performance. However, few studies have focused on their interaction effects on ESG performance (ESGP). The study selected Chinese A-share listed companies from 2013 to 2023 as the research sample and used a moderating effect model to test the complementary or substitutive relationship between the two. The empirical results show that there is a substitutive effect, rather than a complementary one. Further analysis of the individual ESG pillars revealed that the substitutive effect of BI and DT is primarily reflected in corporate governance. Moreover, this substitutive effect is more pronounced in state-owned enterprises and non-manufacturing enterprises, and digital supervision mechanisms may have a stronger substitutive role than traditional independent director oversight mechanisms. These findings uncover the complex relationship between the two governance mechanisms and corporate ESGP, offering important insights for managers; companies need to strike a balance between “human governance” and “digital governance” to maximize ESGP.

1. Introduction

The Environmental, Social, and Governance (ESG) concept represents a strategic approach designed to encourage companies to transition from a focus on short-term profits to a broader commitment to societal sustainable development [1]. ESG provides a holistic framework for evaluating a company’s non-financial performance, and, within the context of global sustainability, it has become a crucial metric for assessing a company’s sustainability efforts [2]. Before exploring ESG in depth, it is important to recognize the significant role that corporate social responsibility (CSR) has historically played in advancing sustainable development. CSR, rooted in social responsibility theory dating back to the 1930s, paved the way for ESG, which first emerged in December 2004 through the United Nations’ “Who Cares Wins” initiative. With the adoption of the Paris Agreement and the United Nations’ 2030 Agenda for Sustainable Development, which outlines 17 Sustainable Development Goals (SDGs), ESG has gained rapid prominence, gradually overtaking CSR as the primary focus in sustainability research. Some scholars argue that ESG and CSR share a similar foundational philosophy, as both frameworks emphasize the dual responsibility of companies to generate profits for shareholders while also addressing the interests of stakeholders such as employees, consumers, the environment, and the community. Despite both being rooted in stakeholder theory and focusing on environmental and social performance, key differences exist between the two. The primary distinction is that CSR is often viewed as voluntary actions taken by companies to give back to society—typically through monetary donations, resources, or other measurable contributions—after profits have been made. CSR emphasizes altruism and philanthropy. ESG, on the other hand, focuses on balancing profitability with social responsibility (“Doing Well and Doing Good”). ESG considers not only how companies create value for shareholders and stakeholders, but also the broader impacts of corporate actions on the environment and society, as well as the reciprocal influence of environmental and social factors on the company. This approach promotes a “win-win” scenario for all parties [3]. Although the concept of ESG is prevalent in developed countries, its development in China has been relatively slow, with a generally low rate of ESG participation [4]. According to the “Rising Tide: An Overview of China A-Share ESG Performance 2024” issued by SynTao Green Finance, (https://en.syntaogf.com/products/asesg2023, accessed on 3 June 2024), the ESG disclosure rate among Chinese A-share companies reached a historical high but still only stood at 38.8%. Various obstacles have hindered the improvement of corporate ESG performance (ESGP) [5]. Therefore, enhancing corporate ESGP has garnered increasing attention from the academic community and remains a significant challenge for policymakers and practitioners.

Corporate governance plays a crucial role when examining the factors driving ESG P in companies [6]. In recent years, a growing body of research has investigated how strong corporate governance can contribute to improved ESG outcomes. The core function of corporate governance is to monitor management and ensure that their decisions reflect the interests of all stakeholders [7]. As a result, board independence (BI) has gained significant attention in governance research [8]. This study focuses on exploring the relationship between BI and ESGP. The importance of this issue has already been well documented in the field of CSR, particularly within the framework of agency theory [9,10]. According to agency theory, the separation of ownership and control creates information asymmetry between external shareholders (principals) and management (agents), which can lead to self-interested actions and biased decision making [11]. Studies on CSR have highlighted that BI is vital for maintaining the effectiveness of corporate governance [12]. However, as the focus of research shifts from CSR to ESG, it remains to be seen whether these findings still apply. Nevertheless, some early studies have shown that BI continues to exert a positive impact on ESGP [13,14,15,16,17].

At the same time, the rapid advancement of digital technologies, such as big data, artificial intelligence, cloud computing, and mobile connectivity, is accelerating the world’s entry into the digital era [18]. Within the thriving digital economy, corporate digital transformation (DT) is intensifying as new information technologies continue to merge with various industries, driving changes in both production processes and governance practices [19]. Digital technologies enhance corporate governance by upgrading monitoring systems and facilitating a fundamental shift in governance paradigms. This transformation enhances corporate governance by improving supervision, communication, and transparency. Through the integration of advanced digital technologies, companies can analyze vast amounts of data, leading to more effective management models and information architectures [20]. This fosters a transparent operational environment, with better data-sharing mechanisms that break down information silos and accelerate communication within the company [21]. Shareholders gain greater access to accurate information, reducing information asymmetry with management and enabling more effective oversight. As a result, corporate governance benefits from improved evaluation of management performance and decision-making capabilities. This improved oversight helps reduce internal agency conflicts [22]. As a result, DT provides a novel governance mechanism to address agency issues [23]. A significant body of empirical research from China highlights the positive impact of DT on ESGP [24,25,26,27] especially in specific sectors such as manufacturing, heavily polluting industries, high-carbon-emission sectors, and banking [28,29,30].

However, in academia, few studies have examined the combined effect of these two governance mechanisms on ESGP. A significant portion of the existing literature, grounded in resource dependence theory or upper echelons theory, explores how elements such as executive team and board diversity influence the relationship between DT and corporate sustainability or ESGP, either by mediating or moderating this impact. For instance, research by Luo et al. indicates that DT has a stronger influence on ESGP when executive teams possess an information technology background [18]. Zhu and Jin’s study suggests that an innovative mindset and technological expertise among executives serve as positive moderators for DT, thus enhancing ESG outcomes [30]. Similarly, Zhang et al. demonstrated that DT substantially improves corporate sustainability, with board diversity acting as a moderating factor [31]. Specifically, they found that while age and gender diversity diminish the positive effects of DT, experiential diversity enhances it. Additionally, Zhu et al. propose that CEOs with a background in digital technology can significantly drive improvements in corporate sustainability by facilitating DT [32]. However, research on independent directors is relatively lacking. This is surprising, as independent directors are a unique dual-role component of board structures, not only providing resources and advice but also bearing supervisory responsibilities. Most scholars seem to focus solely on one function of independent directors, neglecting the other. The dual nature of independent directors suggests that they may play a complex role in the process of DT promoting ESGP. This is an area worthy of in-depth exploration, yet current studies in the literature have scarcely addressed it. Although a small number of scholars have analyzed the moderating role of BI on the impact of DT from the perspective of environmental performance [33] and CSR performance [34], research expanding this scope to ESGP remains scarce. Some studies in the literature, however, suggest that the interaction between different governance mechanisms is complex [35]. From a complementary perspective, effective governance requires both incentive and monitoring mechanisms [36,37], while the substitution perspective argues that governance mechanisms can substitute for each other [38,39,40]. Therefore, based on these two viewpoints, two scenarios can be predicted; one possibility is that digital technology, through the integration and automation of business processes, makes information more timely, accurate, and transparent, providing strong decision-making support for independent directors in monitoring the company, thereby enhancing the quality of internal oversight [41] and promoting the positive fulfillment of social responsibility and improved corporate governance, ultimately contributing to better ESGP [42]. Another possibility is that, as DT deepens, monitoring mechanisms based on digital technology will further develop, while traditional governance mechanisms, such as BI, will become less significant. By improving transparency and enabling more effective monitoring of management activities, DT has the potential to reduce reliance on conventional governance tools, such as BI, and address agency conflicts more efficiently.

To contribute to this debate and reveal how the two governance mechanisms—BI and DT—jointly influence ESGP, we selected an annual observation sample of 27,222 A-share listed companies in China from 2013 to 2023 to study this issue. Our research findings suggest that these two governance mechanisms are substitutes for each other regarding their effect on ESGP. The results are further validated through a variety of endogeneity and robustness checks. Further analysis shows that the substitution effect mainly occurs in the governance (G) pillar of ESG, and the substitution effect of DT is greater than that of BI. In addition, heterogeneity analysis reveals that the substitution effect is more significant in state-owned enterprises (SOEs) compared to non-state-owned enterprises (non-SOEs) and in non-manufacturing enterprises (non-MEs) compared to manufacturing enterprises (MEs). Overall, our study indicates that BI and DT do indeed have a substitution effect on promoting ESGP.

These findings provide valuable contributions to the current body of literature on DT, board structure, and ESGP. To begin with, this study, to the best of our knowledge, is the first to empirically confirm the substitution effect between BI and DT in enhancing ESGP, contrary to the complementary relationship suggested by prior studies. This finding challenges earlier assumptions and suggests that in companies with a high degree of DT, increasing the proportion of independent directors may not be necessary and vice versa. This insight adds nuance to the ongoing debate on the interplay between traditional governance structures and digital advancements. Second, this study refines the analysis by deconstructing ESGP into its individual components, finding that the substitution effect of DT on BI is primarily reflected in the governance pillar. This nuanced observation aligns with the core theory underlying our hypothesis, proving that the substitution relationship between DT and BI is mainly rooted in their overlapping roles within corporate governance mechanisms. This contributes to a deeper understanding of how digital tools can serve as alternative governance mechanisms, particularly in the governance pillar. Third, this study reveals important variations in the substitution effect across different types of companies, a point that has been underexplored in the literature. It shows that the effect is more pronounced in SOEs and non-MEs. This differential effect highlights the importance of context in governance decisions, providing more detailed guidance for companies when selecting supervisory mechanisms. It emphasizes the need for differentiated governance strategies based on the characteristics of both the company and its industry, offering a more tailored approach than previous one-size-fits-all recommendations. Moreover, this study finds that the digital supervision mechanism may exhibit a stronger substitution effect than traditional independent director oversight. This insight adds a new dimension to corporate governance discussions, suggesting that DT-driven governance can be more efficient in some cases. From a corporate governance perspective, this finding contributes to the evolving literature on “human governance” versus “digital governance,” uncovering the complex relationship between these two governance mechanisms and corporate ESGP. It offers valuable insights for decision makers, emphasizing the importance of balancing these mechanisms to maximize ESG outcomes.

The rest of this paper is structured as follows: Section 2 discusses previous research and develops the theoretical foundation. Section 3 explains the research methods in detail. Section 4 focuses on the empirical analysis, including regression outcomes, examining both direct and moderating effects, and performing robustness and heterogeneity tests. Section 5 provides an interpretation and discussion of the major findings. Section 6 addresses the theoretical and practical implications, recognizes limitations, and proposes directions for future research.

2. Literature Review and Hypotheses Development

2.1. Theoretical Background

Both agency theory and stakeholder theory suggest that independent directors hold a crucial role in corporate governance. Independent directors, due to their position independent of management, can evaluate management performance more objectively [43], effectively safeguarding the interests of shareholders [13,44], avoiding the influence of corporate executives and enhancing the supervisory function of the company [45]. This enhances the efficiency of corporate governance and mitigates conflicts of interest between stakeholders [46,47,48]. Their independence allows them to promote ESG disclosure and ESGP [49,50], as they focus more on long-term interests and reputation preservation, which helps to reduce agency conflicts and information asymmetry [51,52].

Corporate DT can also improve the level of supervision and governance by reducing information asymmetry and lowering agency costs, thereby promoting corporate ESGP. First, the application of digital technologies makes organizational structures more networked and flattened, enhancing communication efficiency between management and employees, thus alleviating agency conflicts [53]. Digital technologies also optimize internal control systems, such as smart accounting systems and automated business processes, enhancing interdepartmental collaboration and supervision, improving the efficiency of information acquisition and processing, reducing information asymmetry, mitigating opportunistic behavior by managers, and improving corporate transparency, supervisory quality, and management norms [41,54,55]. Furthermore, the decentralized nature of DT enables employees to become transaction supervisors, reducing managerial discretion and further mitigating agency issues between shareholders and management [53,56], while also promoting stakeholder oversight of the company [57,58]. In conclusion, DT motivates managers to prioritize the company’s long-term objectives, leading to improved ESGP [59] while simultaneously strengthening corporate environmental governance and boosting market competitiveness [33].

Enhancing corporate governance serves as an effective strategy to tackle agency problems [23]. Measures like raising BI and advancing DT can help alleviate these issues [51,52,53,56]. Therefore, increasing BI and DT can be seen as a substitute for strengthening corporate governance [60]. Agency theory suggests that principals can select between two governance mechanisms: incentives or supervision [11]. From a corporate governance standpoint, the key way in which both BI and DT enhance ESGP is through supervision. As digital technologies are increasingly integrated, process-based monitoring systems are strengthened, while reward-based governance mechanisms are diminished. These two mechanisms—supervision and incentives—tend to act as substitutes. Within the agency theory framework, digital technology has led to a transformation in corporate governance models, improving governance through process-focused, constraint-based supervision and reducing dependence on incentive mechanisms tied to outcomes [23]. As for independent directors, they are generally considered to play dual roles of supervision and resource provision [61]. Most scholars emphasize their supervisory role in mitigating agency problems and promoting ESGP [62], which leads to positive governance effects [45].

Thus, from the corporate governance perspective, governance by BI and governance through DT are essentially both supervisory mechanisms, though differing in approach. Governance by BI relies on “human governance,” primarily using human supervision, whereas governance through DT depends on “digital governance,” utilizing digital technologies for supervision. However, whether the combined influence of these two governance mechanisms on ESGP is complementary or substitutive remains underexplored, with a lack of theoretical and empirical research in this area. So far, only two closely related studies, by Lu et al. [41] and Meng et al. [34], have been identified. Both studies indicate that BI strengthens the positive association between DT and ESG or CSR performance. However, Lu et al.’s study [41] oversimplified the complex reality by using the median method. Additionally, Lu et al.’s interpretation of the results lacks theoretical grounding, offering only a surface-level understanding. They argue that DT can improve a company’s information transparency and governance, and that independent directors aim to play a role in corporate governance and oversight of management. Therefore, independent directors are likely to adopt digital technologies to enhance supervision and governance. According to their explanation, the more independent directors there are, the more attention is paid to digital transformation. However, this explanation is overly simplistic and unreasonable. In fact, when there are fewer independent directors, they may be more inclined to leverage the convenience brought by DT to reduce the pressure of overseeing the company. Meanwhile, although Meng et al. [34] demonstrated the positive impact of DT on corporate social performance (CSP) and that BI significantly strengthens this positive impact, their study is based on the perspective of CSR, which is not comprehensive regarding ESGP. Furthermore, the explanation of why a higher number of independent directors tend to adopt digital means to respond to stakeholder demands, while fewer independent directors do not, is unclear. Overall, these two studies lack detailed theoretical discussion and further empirical support regarding how BI and DT work together to enhance ESGP.

In reviewing the existing literature, it is evident that significant research gaps remain in the study of the combined influence of BI and DT on ESGP. Therefore, this paper intends to explore the comprehensive impact of BI and DT on corporate ESGP through empirical research from a Chinese perspective. Through this study, we aim to provide theoretical foundations and practical guidance for companies enhancing their ESGP, promoting sustainable corporate development. At the same time, this research hopes to enrich the theoretical research on governance by BI and DT, offering new perspectives and empirical evidence for academia.

2.2. Hypotheses Development

The research framework was established as shown in Figure 1.

Numerous studies in the past have demonstrated that effective governance, such as oversight [63,64], along with the alignment of incentive structures [65,66], has a positive impact on CSR. This is because, without proper monitoring or incentives, corporate managers may prioritize short-term personal gains and avoid long-term investments such as CSR, which take time to produce results [67]. Therefore, many studies, such as Torres [62] and Nguyen and Nguyen [68], argue that a strong governance structure tends to result in higher ESG disclosure or performance. However, the mechanisms through which governance practices affect ESGP remain a subject of ongoing debate. For instance, Rediker and Seth [39] and Zajac and Westphal [40] suggest that governance mechanisms can work together to improve organizational outcomes, as their combined benefits outweigh individual drawbacks. This implies that one governance practice can enhance the impact of another, leading to synergistic effects on ESGP [69]. The notion of complementarity implies that the combination of governance mechanisms enhances their effectiveness through “mutual reinforcement” [36]. Therefore, to maximize a company’s ESGP, it is necessary to implement specific combinations of governance practices simultaneously. An effective governance structure, through cumulative mechanisms of increased supervision and aligned incentives, induces synergies between governance practices, thereby promoting ESGP. For instance, combining increased BI with DT can create positive synergy in supervising company information, enhancing transparency and responding to stakeholder needs, ultimately helping improve corporate ESGP. On the one hand, firms with a greater percentage of independent directors are more likely to utilize digital technologies to gain a deeper understanding of stakeholder expectations and address their needs throughout the process of DT. This is because companies in the midst of DT are more likely to use digital tools (such as social media, forums, or microsites) to engage with stakeholders on matters related to CSR. These companies can also respond more effectively to social responsibility demands, and, with an increasing number of independent directors, they are more likely to show a greater commitment to social concerns. Furthermore, as DT enhances oversight by various stakeholders, independent directors will be more motivated to encourage companies undergoing digital changes to invest in ESG initiatives that align with societal values [34]. In addition, independent directors are likely to push companies to embrace digital technologies in order to meet ESG objectives, improving the company’s reputation and, by extension, their own [70]. As a result, boards with a greater presence of independent directors are more responsive to stakeholder pressures driven by DT and more proactive in using digital tools to address social responsibility issues, leading to a notable improvement in social responsibility performance. This is especially true when independent directors have a background in information technology (either professionally or academically) or participate in the board’s IT (information technology) governance [71], which can further drive DT [32,72,73] and better supervise managerial behavior [74], ultimately leading to a greater impact of DT on ESGP [18]. Empirical research has further demonstrated that the effect of DT on ESG or CSR is more significant in organizations where the proportion of independent directors is high [34,41]. On the other hand, the digital technologies used in the company’s internal management systems provide independent directors with more accurate and scientific information, significantly improving the level of independent director oversight, thus enhancing the quality of internal supervision [41]. The improvement in internal supervision quality positively impacts the fulfillment of social responsibility and the enhancement of corporate governance, which helps to improve ESGP [42]. Thus, in organizations with an advanced level of DT, independent directors have a greater influence on ESGP.

However, although BI and DT may generate complementary effects in improving ESGP, some researchers argue that they may also have substitutive effects. The core logic of this perspective lies in the fact that governance decisions involve the allocation of resources, and using governance tools for supervision and management often entails significant costs [75]. In this case, when a company uses supervisory mechanisms to control management behavior, it may face a cost–benefit trade-off [76]. Overall, while the combined use of multiple governance mechanisms may offer certain benefits, the associated costs may outweigh the potential gains, making such an approach not always ideal [77]. In fact, Hoskisson et al. [78] proposed that a “systemic balance between governance devices’ may be more beneficial to a company’s long-term interests, meaning that when selecting and implementing governance mechanisms, the interaction and balance between different mechanisms should be considered rather than simply adding more governance tools. Particularly in the context of CSR, employing multiple governance mechanisms may lead to ‘diminishing returns on behavior” [40], where, after a certain point, additional governance measures actually reduce marginal benefits. All of the above arguments suggest that governance mechanisms can serve as substitutes for organizational outcomes rather than complements. This perspective is especially important in the context of ESG. For example, if a company enhances its governance capabilities through digital technologies, using advanced information systems for supervision and management to comprehensively monitor corporate activities [23], then additional governance mechanisms, like increasing the proportion of independent directors, may become redundant. Although, according to agency and signaling theory, diversity among independent directors can bring more varied perspectives and resources to the board, helping the company better respond to the needs of various stakeholders [14,79,80], the board may be more inclined to pressure executives to focus more on financial results than on social outcomes [81], especially when the board holds the view that strong ESGP does not necessarily result in improved financial outcomes. Consequently, further oversight from independent directors aimed at achieving profitability may have little impact on the company’s decisions regarding ESG investments, as management has already been advised against excessive investment in ESG initiatives [75]. Therefore, the substitution perspective suggests that adding any extra governance mechanism (such as supervision) to an existing governance practice may result in costs that outweigh the benefits. In this scenario, the simultaneous use of multiple governance mechanisms may not necessarily lead to more proactive ESG practices. In fact, the marginal effects of each governance mechanism in such cases may not increase and may even turn negative. This perspective is supported by some empirical studies. For example, the findings of Randoy and Goel [82] and Rediker and Seth [39] suggest that the presence of one governance mechanism may diminish the need for other governance mechanisms. Specifically, Rediker and Seth [39] found a substitutive relationship between the board’s monitoring capacity and other alternative governance mechanisms. These studies further support the view that when a company has already adopted an effective governance mechanism, other governance mechanisms may become redundant or unnecessary. In summary, in the pursuit of ESG goals, BI and other governance mechanisms, such as DT, may have a substitutive relationship. Excessive governance measures may not only bring additional costs but could also reduce the effectiveness of governance, thereby weakening the company’s investment and performance in ESG.

In conclusion, regardless of whether one considers complementary or substitutive effects, it is generally agreed that a notable interaction exists between BI and DT. This interplay will, together, shape the ESGP of a company. Hence, we propose the following hypothesis:

Hypothesis 1. 

Board independence and digital transformation will have either complementary effects or substitutive effects on ESG performance.

3. Research Design

3.1. Selection of Samples

This study focuses on listed companies on the main board of the Shanghai and Shenzhen A-shares, covering the period from 2013 to 2023. This decade was chosen due to China’s significant economic reforms and efforts toward sustainable development and DT, providing a critical period for analyzing their impact on ESG and DT across enterprises. The selection of 2013 as the starting point for this study is based on two key reasons. First, in 2013, China experienced a severe smog crisis, which drew significant public attention and concern over environmental issues. In response, the Chinese government introduced the Air Pollution Prevention and Control Action Plan that same year, setting clear emission reduction targets and strengthening environmental regulations for companies, which had a lasting impact on businesses. Second, in 2012, China released the 2012 Project Guide of Special Funds for Deep Integration of Informatization and Industrialization, offering special funds to promote the integration of information technology and industrialization. Since then, an increasing number of companies have begun to pursue DT [83]. Based on these two factors, 2013 was chosen as the starting point of this study. The main reason for selecting listed companies as research subjects is that key research topics such as “DT” and “ESGP” are more comprehensively disclosed in listed companies, allowing for reliable data from authoritative databases. In addition, compared to other market segments, the main board companies are more mature and stable, with clearer, more transparent, and more reliable information disclosure. Moreover, the main board-listed companies are typically larger in scale and have greater social influence.

ESGP data were obtained from the Huazheng (Sino-Securities) ESG Rating via the Wind database, while other data were sourced from the China Stock Market & Accounting Research (CSMAR) database, with the statistics cut-off date set at May 20, 2024. Due to the fact that the financial companies’ financial statements and market performance differ from those of companies in other industries, we excluded all financial companies [84,85]. To facilitate the empirical analysis, the study also excluded samples from data missing key variables, companies labeled with ST and PT stock codes, and firms that have suspended or terminated their listings or have been listed for less than a year. Furthermore, to reduce the impact of outliers on the empirical test results, a 1% Winsorize transformation was applied to the continuous variables. Ultimately, this study compiled 27,222 observations from 3819 listed companies.

3.2. Variables Measurements

The dependent variable ESGP:

Table 1. Variables and explanations.

Variable Type	Variable Name	Variable Abbreviation	Variable Definition	Source
Dependent Variable	ESG Performance	ESGP	Assign scores according to ESG ratings from low to high on a scale of 1 to 9	Huazheng ESG
Independent Variable	Board Independence	BI	The ratio of independent directors to the total number of board members	CSMAR
	Digital Transformation	DT	Natural logarithm of frequency share of words related to enterprise digitalization +1	CSMAR
Control Variables	CEO Duality	CEOD	A dummy variable that takes the value of 1 if the CEO also serves as the chairman of the board, and 0 if they hold separate roles	CSMAR
	State-owned Enterprise	SOE	1 if state-owned enterprise, otherwise 0	CSMAR
	Board Size	BS	The number of board directors	CSMAR
	Leverage	LEV	Total liabilities/Total assets	CSMAR
	Equity Concentration	EC	The sum of the squared shareholding percentages of the company’s top five largest shareholders	CSMAR
	Network Centrality of Independent Directors	NCID	$\mathrm{D}\mathrm{g}\mathrm{r}\mathrm{e}{\mathrm{e}}_{\mathrm{i}}={\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{n}}}{\displaystyle \frac{{\mathrm{P}}_{\mathrm{i}\mathrm{j}}}{\mathrm{n}-1}}$ pij indicates whether there is a relationship between company i and company j; if a board member of company i holds a board position in company j, then pij = 1. otherwise, pij = 0. n represents the number of companies that form the independent director network.	CSMAR

presents all the variables and their measurement methods. The ESGP is measured by an ESG score, a method previously utilized in several studies [86,87,88]. Although there are many versions of ESG ratings or scores for listed companies, including internationally renowned ESG rating or scoring databases such as MSCI and Bloomberg, as well as databases specifically targeting China, such as Huazheng ESG Rating, Wind, etc., this paper ultimately chooses the Huazheng ESG Rating database for the following reasons: (1) the Huazheng ESG Rating basically covers all A-share listed companies; (2) the Huazheng ESG Rating started relatively early, with some ESG rating data available since 2009; and (3) as a domestic rating agency in China, its evaluation indicators and methods are more suitable for Chinese companies and more in line with China’s national conditions, making the rating results relatively scientific and accurate. Moreover, the Huazheng ESG rating system aligns with the key principles of international ESG standards and reflects the core ideas of the Triple Bottom Line (TBL) theory. According to the TBL framework [89,90,91,92,93,94], evaluating an organization’s performance requires not only a focus on financial results but also consideration of its social and environmental impacts. The latest Huazheng ESG framework (see Appendix A,

Table A1. Huazheng ESG rating indicators.

3 Pillars	16 Themes	40+ Key Issues
Environment (E)	Climate Change	Greenhouse gas emissions, GHG emissions reduction roadmap, Response to climate change
	Resource Utilization	Water consumption, Land use and biodiversity, Material consumption
	Environmental Pollution	Industrial emissions, Electronic waste, Hazardous waste
	Environmentally Friendly	Renewable energy, Green buildings, Green factories
	Environmental Management	Sustainable certification, Environment penalty, Supply chain management—E
Social (S)	Human Capital	Employee health and safety, Employee inspiration and development, Employee relations
	Product Liability	Quality certification, Recall and complaints
	Supply Chain	Supplier risk and management, Supply chain relationship
	Community Investment	Inclusion, Community investment, Employment, Technology innovation
	Data Security and Privacy	Data Security and Privacy
Governance (G)	Shareholders’ Interest	Protection of shareholders’ interests
	Governance Structure	ESG, Risk control, Board structure, Executive turnover
	Information Disclosure Quality	ESG external assurance, Credibility of information disclosure
	Governance Risk	Major shareholder behavior, Solvency, Litigation, Tax transparency
	External Punishment	Various external punishments
	Business Ethics	Business ethics, Anti-corruption

) includes 3 primary pillar indicators, 16 secondary theme indicators, and 44 tertiary topic indicators. Among them, the five themes of climate change, resource utilization, environmental pollution, environmentally friendly, and environmental management directly correspond to the environmental impact in TBL theory; the five themes of human capital, product liability, supply chain, community investment, and data security and privacy directly correspond to social impact. Under the governance pillar, some indicators, such as protection of shareholder’s interests, risk control, credibility of information disclosure, and solvency, have a direct impact on the company’s economy. These indicators reflect the company’s performance in key areas such as capital acquisition, operational stability, financial health, and market confidence. By ensuring the company’s economic health, reducing financial risk, and enhancing transparency, these indicators help promote long-term sustainable development in the economic dimension of the TBL theory.

Based on the above, this paper, referring to the previous research [59,95,96,97,98], chooses the Huazheng ESG Rating to measure ESGP. The scoring outcomes are categorized into nine levels ranging from leading (AAA, AA, A) to average (BBB, BB, B) and lagging (CCC, CC, C). Referring to the previous research [41], this study classifies the scores of Chinese listed companies from the ESG rating from low to high. For instance, in the ESG scoring, a CCC-rated company scores 1 point, while an AAA-rated company scores the highest at 9 points. The higher the ESG score, the better the ESGP.

The independent variables BI and DT: In line with prior research, we measure business BI by the percentage of independent directors serving on the board [15,16,17,49,99,100]. Regarding DT, this paper suggests that the vocabulary used in a company’s annual reports can indicate its strategic focus and future outlook, reflecting its business philosophy, development priorities, and significant future decisions. Typically, companies that place a high value on digital strategy tend to disclose more information related to “digitalization.” Measuring the extent of DT through the frequency of digitalization-related terms in annual reports has become a widely adopted approach [23,28,71,101,102,103,104,105]. The frequency of these keywords in annual reports for listed companies from 2013 to 2022 can be accessed through the CSMAR database. For the 2023 data, we supplemented them using the methods outlined by previous studies [28,34,41,53,104,105,106]. The main steps are as follows: First, the digitalization keyword library on CSMAR was updated with some new digitalization vocabularies that appeared in 2023, thereby establishing the keyword library. Second, Python was used to crawl the 2023 annual reports of listed companies (statistics deadline was 20 May 2024) and convert them into txt versions. Third, the jieba library was used to segment the words in the annual reports of listed companies. Finally, the frequency of DT keywords in each company’s annual report was counted and logarithmically transformed to measure the degree of DT of enterprises, defined as DT = ln (frequency proportion of enterprise digitalization-related vocabulary + 1).

Control variables: The research model incorporates several control variables commonly used in empirical studies. First, three key controls are included: Leverage (LEV), State-owned Enterprise (SOE), and Equity Concentration (EC). LEV, which indicates a firm’s risk level, is expected to influence ESGP, as higher leverage might motivate companies to enhance ESG practices as a strategy to mitigate risk. It is calculated as the ratio of total debt to total assets [70,100,107]. SOEs tend to allocate more resources to environmental and social responsibilities compared to non-SOEs, and therefore, SOE is included as a control variable. It is represented by a dummy variable, assigned a value of 1 if the firm is an SOE and 0 otherwise [108,109]. Additionally, EC is measured using the Herfindahl index, which sums the squares of the shareholding ratios of the top five shareholders, as research suggests that ownership concentration affects CSR decisions made by management [19,34]. In addition, three corporate governance variables are also controlled: Board Size (BS), CEO Duality (CEOD), and Network Centrality of Independent Directors (NCID). BS is included under the premise that larger boards may provide more resources and insights, positively impacting CSR integration into business practices [110,111,112]. BS is measured by the total number of board members [113]. CEO Duality, or the combination of CEO and board chair roles, can reduce agency costs, increase transparency, and improve sustainability performance. Thus, CEOD is a control variable, coded as 1 if the CEO is also the board chair, and 0 otherwise [107,114,115]. Independent directors with higher network centrality are more effective in enhancing governance, improving information disclosure, and reducing agency problems. Their centrality within social networks provides them with reputational capital and informational advantages, improving corporate ESGP [116,117]. Therefore, NCID is included as a control variable. The calculation method is detailed in Table 1.

3.3. Empirical Model

This study primarily explores two aspects: first, the impact of BI and the degree of DT on ESGP, and second, their interaction effects on ESGP. Following the previous research [115,118], we developed a two-way fixed effects model [13,70,119], accounting for both individual and temporal effects. By doing so, we mitigate potential biases in estimation that could arise from omitted variables, unobservable characteristics at the individual level, or time-related trends. This approach ensures a more accurate estimation of the causal relationships among the variables.

To test the Hypothesis 1, we developed the following models:

ESGP_i,t = α₀ + α₁BI_i,t + α₂DT_i,t + α₃Controls_i,t + Year_t + Firm_i + ε

(1)

ESGP_i,t = β₀ + β₁BI_i,t + β₂DT_i,t + β₃ (BI_i,t×DT_i,t) + β₄Controls_i,t + Year_t + Firm_i + ε

(2)

In these two models, i and t represent the company and the year, respectively. The core explanatory variables are BI and DT, while ESGP is the dependent variable. The interaction term is denoted as BI_i,t × DT_i,t. The constants are represented by α₀ and β₀. The control variables are referred to as “Controls”, and ε represents the random error term. If the coefficients α₁ and α₂ are significant and positive, and β₃ is significant, the Hypothesis 1 is supported. Specifically, if β₃ is greater than 0, the complementarity hypothesis is supported; if β₃ is less than 0, the substitution hypothesis is supported. Additionally, this study focuses on the company level, taking into account the correlation between the error terms of the observations. For instance, financial data from the same company in different years may be influenced by inherent characteristics of the company, such as corporate culture and management style. Following the relevant literature, we clustered the standard errors at the firm level [115,120,121] to ensure the robustness of the regression results.

4. Empirical Analysis

4.1. Descriptive Statistics

Table 2. Descriptive statistics.

VarName	Obs	Mean	SD	Min	Median	Max
ESGP	27,222	4.133	0.929	1.000	4.000	8.000
BI	27,222	0.374	0.047	0.333	0.364	0.500
DT	27,222	1.508	1.382	0.000	1.386	5.088
CEOD	25,715	0.268	0.443	0.000	0.000	1.000
SOE	27,222	0.380	0.485	0.000	0.000	1.000
BS	27,222	8.371	1.441	5.000	9.000	12.000
LEV	27,222	0.430	0.199	0.064	0.423	0.887
EC	27,222	0.146	0.093	0.017	0.126	0.401
NCID	27,222	0.421	0.330	0.000	0.431	1.330

presents the descriptive statistics for the variables. The mean, maximum, minimum, and median values for ESGP are 4.133, 8.000, 1.000, and 4.000, respectively, indicating significant variability in the ESGP of Chinese listed companies, with generally low levels of emphasis on ESG. The mean, maximum, minimum, and median values for BI are 0.374, 0.50, 0.333, and 0.364, respectively. This suggests that most Chinese listed firms are merely meeting the minimum requirement set by the CSRC (China Securities Regulatory Commission), which mandates that independent directors should constitute at least one-third of the board. There is little evidence that these companies are making efforts to increase the proportion of independent directors beyond this minimum. For DT, the mean, maximum, minimum, and median values are 1.508, 5.088, 0.000, and 1.386, respectively, highlighting a wide disparity in the level of DT among companies, with considerable room for improvement in some cases.

4.2. Bivariate Analysis

Table 3. Correlation matrix.

	VIF	ESGP	BI	DT	CEOD	SOE	BS	LEV	EC	NCID
ESGP		1
BI	1.477	0.061 ***	1
DT	1.035	0.168 ***	0.057 ***	1
CEOD	1.129	0.002	0.115 ***	0.111 ***	1
SOE	1.296	0.058 ***	−0.083 ***	−0.144 ***	−0.311 ***	1
BS	1.572	0.015 **	−0.562 ***	−0.070 ***	−0.179 ***	0.252 ***	1
LEV	1.086	−0.098 ***	−0.023 ***	−0.062 ***	−0.115 ***	0.271 ***	0.118 ***	1
EC	1.057	0.093 ***	0.022 ***	−0.102 ***	−0.052 ***	0.214 ***	0.028 ***	0.049 ***	1
NCID	1.005	0.052 ***	0.024 ***	0.038 ***	0.023 ***	−0.050 ***	−0.028 ***	−0.004	0.012 **	1

Note: t statistics in parentheses, ** p < 0.05, *** p < 0.01.

displays the Spearman correlation matrix for independent and control variables to evaluate potential multicollinearity. With correlation coefficients ranging from 0.2% to 56.2%, the likelihood of significant multicollinearity affecting the regression results is low. Additionally, the Variance Inflation Factor (VIF) was employed for further analysis, with values ranging between 1.005 and 1.572, which are well below the threshold of 10, indicating no serious multicollinearity concerns. Thus, multicollinearity does not affect the validity of the analysis.

4.3. Regression Results

We first conducted the Hausman test, and all models had a p-value of 0. Therefore, the fixed effects model was more appropriate.

Table 4. Main regression results.

	(1)	(2)
VARIABLES	Model 1	Model 2
BI	0.591 ***	0.901 ***
	(2.60)	(3.04)
DT	0.041 ***	0.118 ***
	(4.28)	(2.59)
BI × DT		−0.207 *
		(−1.74)
CEOD	0.028	0.028
	(1.25)	(1.23)
SOE	0.029	0.028
	(0.51)	(0.51)
BS	0.007	0.007
	(0.83)	(0.84)
LEV	−0.655 ***	−0.653 ***
	(−8.93)	(−8.90)
EC	0.238	0.236
	(1.27)	(1.26)
NCID	−0.053 **	−0.053 **
	(−2.13)	(−2.13)
Year	Yes	Yes
Firm	Yes	Yes
Constant	4.014 ***	3.897 ***
	(27.15)	(23.62)
Observations	25,317	25,317
R-squared	0.527	0.527
F test	14.13	12.78
r2_a	0.456	0.456
F	14.13	12.78

Note: Fixed effects regression models. Results are clustered by firms. T-statistics are in parentheses. Definitions of the variables are in Table 1. * p < 0.1. ** p < 0.05. *** p < 0.01.

presents the findings from the complementarity and substitution tests. In Model 1, BI and DT are included as the primary variables. And the results indicate that BI has a significant positive impact on ESGP (β = 0.591, p < 0.01). These findings support the research conclusions of Belen Lozano and Martinez Ferrero [17], Menicucci and Paolucci [16], Dicuonzo et al. [15], demonstrating that BI can lead to higher ESGP for companies. Similarly, the results of Model 1 confirm the findings of Cai et al. [24], Chen et al. [25], Hou et al. [26], Zhang [19], Zhuo et al. [27], and Luo et al. [18], who found that DT positively impacts corporate ESGP (β = 0.041, p < 0.01). Model 2 in Table 4 tests the complementary and substitutive relationship between BI and DT in promoting ESGP. The results show that both BI and DT are significantly positively correlated with corporate ESGP. However, the interaction term BI × DT is significantly negatively correlated with ESGP at the 10% statistical level (β = −0.207, p < 0.1). Therefore, the Hypothesis 1 that the interactive effect of BI and DT on ESG is significant is established, and the results support the substitution effect. The simple slope test in the panel (a) of Figure 2 also indicates that when DT is low, the relationship between BI and ESGP is significant (simple slope = 0.875, p < 0.01), but when DT is high, the relationship between BI and ESGP is not significant (simple slope = 0.303, n.s.). These results suggest that when DT reaches a higher level and digital technologies for supervision are well developed, additional supervision from independent boards does not increase the marginal returns on corporate ESGP. Similarly, the simple slope test in the panel (b) of Figure 2 shows that when BI is low, the relationship between DT and ESGP is significant (simple slope = 0.051, p < 0.01). However, when BI is high, while the relationship between DT and ESGP remains significant, the slope significantly decreases (simple slope = 0.031, p < 0.01). This indicates that as the BI increases, its collective supervisory effort becomes more effective, and further strengthening supervision through digital technologies yields diminishing marginal returns for corporate ESGP. In summary, the impact of DT as a new supervisory mechanism on the ESGP of listed companies can substitute for and penetrate traditional supervisory mechanisms like BI. This result supports substitution and rejects complementarity. Moreover, we also find that the substitutive effect of DT on BI is stronger than the effect of BI on DT, as the former results in a greater slope reduction, even leading to the relationship between the independent and dependent variables becoming insignificant.

4.4. Additional Analysis

4.4.1. Robustness Test

To assess the robustness of our findings, we performed several additional analyses. First, considering the potential issue of endogeneity, previous research suggests that DT may be endogenous variables [34,60,102,122]. To address endogeneity, particularly reverse causality, we employed the two-stage least squares (2SLS) regression method with instrumental variables. Instrumental variables were created for DT and the interaction term (BI × DT). Because DT decisions may be influenced by industry peer effects, yet the industry average DT level remains independent of a firm’s direct ESGP, we utilized the annual industry average (excluding the focal firm) (DTmean) as an instrumental variable for DT [33,34,71,123,124]. Since DT may be endogenous, the interaction term should also be endogenous [125]. Therefore, we used the interaction term of BI and DTmean (BI × DT_IV) as the instrumental variable for BI × DT. The first two columns of

Table 5. Endogeneity test.

	2SLS		GMM
	(1)	(2)	(3)	(4)
VARIABLES	Model 1	Model 2	Model 1	Model 2
L.ESGP			0.397 ***	0.381 ***
			(23.32)	(18.78)
BI	0.635 ***	1.218 ***	1.372 ***	2.610 ***
	(2.74)	(3.03)	(5.56)	(4.47)
DT	0.141 *	0.289 **	0.054 ***	0.356 ***
	(1.69)	(2.51)	(3.70)	(3.12)
BI × DT		−0.387 *		−0.796 ***
		(−1.91)		(−2.62)
Control	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
firm	Yes	Yes	Yes	Yes
Constant			2.381 ***	2.253 ***
			(12.26)	(5.17)
Observations	25,269	25,269	21,710	21,710
R-squared	0.002	0.001
Kleibergen–Paap rk LM statistic	87.218	86.489
p-Value	0	0
Kleibergen–Paap rk Wald F statistic	133.037	65.762
Stock–Yogo test at 10%	16.38	7.03
AR(1) p-Value			0	0
AR(2) p-Value			0.285	0.304
Hansen test p-Value	0	0	0.109	0.108

Note: 2SLS and GMM models. Results are clustered by firms. T-statistics are in parentheses. Definition of the variables is in Table 1. * p < 0.1. ** p < 0.05. *** p < 0.01.

report the 2SLS estimation results. Model 1 shows that at the 5% level, BI is significantly positively correlated with ESGP (β = 0.635, p < 0.01), and at the 10% level, DT is positively correlated with ESGP (β = 0.141, p < 0.1). Model 2 again verifies the substitution hypothesis: The interaction term between BI and DT (BI × DT) is negatively correlated with ESGP (β = −0.387, p < 0.1).

Second, overlooking dynamic relationships could lead to significant endogeneity concerns, often addressed by employing the system Generalized Method of Moments (GMM) to manage unobserved heterogeneity [114,121,126,127,128]. To account for the dynamic interactions between BI, DT, and ESGP, we applied the system GMM method [129]. The dynamic system GMM estimation (last two columns of Table 5) shows that ESGP is auto-correlated with its previous year’s value, and we used this lagged variable (L.ESGP) as an explanatory variable. The GMM Models 1 and 2 yield results consistent with our main model and the 2SLS model, once again demonstrating the substitutive relationship between BI and DT in promoting ESGP. This consistency suggests that our main results are not overly influenced by reverse causality or dynamic relationship biases.

Third, when studying ESG, it is essential to evaluate these components separately in addition to the overall assessment [130]. Therefore, to explore how the substitutive effect of BI and DT manifests across the different pillars of ESG, and as a further robustness check, we conducted regressions using each ESG pillar as the dependent variable [79,100,118] to analyze the contribution of each pillar to the substitutive effect of BI and DT.

Table 6. Comparing the three pillars of ESG.

	Environmental Pillar		Social Pillar		Governance Pillar
	(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
BI	−0.166	−0.148	0.180	0.198	1.601 ***	2.252 ***
	(−0.54)	(−0.41)	(0.45)	(0.39)	(5.09)	(5.18)
DT	0.044 ***	0.048	0.091 ***	0.095	0.016	0.179 ***
	(3.69)	(0.77)	(5.58)	(1.20)	(1.18)	(2.73)
BI × DT		−0.012		−0.012		−0.435 **
		(−0.07)		(−0.06)		(−2.54)
Control	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes	Yes	Yes
Constant	2.111 ***	2.104 ***	4.360 ***	4.353 ***	5.198 ***	4.953 ***
	(10.88)	(10.09)	(16.53)	(15.12)	(25.92)	(21.24)
Observations	25,316	25,316	25,316	25,316	25,316	25,316
R-squared	0.519	0.519	0.539	0.539	0.505	0.505

Note: Fixed effects models with the E, S, and G pillars as the dependent variables. Results are clustered by firms. T-statistics are in parentheses. Definition of the variables is in Table 1. ** p < 0.05. *** p < 0.01.

reports the results, showing the impact of the substitutive effect on individual ESG pillars. Columns 1 and 2 present the estimates for environmental performance, while columns 3 and 4 and columns 5 and 6 represent estimates for social performance and governance performance, respectively. The results indicate that, among the three pillars of ESG, only under the G pillar is the interaction term BI × DT significantly negative (β = −0.435, p < 0.05). This shows that the substitutive effect of DT on BI is mainly reflected in the G pillar.

Fourth, in the latest version of the TBL theory, Elkington pointed out that the standard for measuring the success of sustainability goals should focus on the overall well-being of humanity and the planet [131]. Therefore, he emphasized that social and environmental outcomes are more important than economic outcomes. Additionally, in the long term, social and environmental outcomes are closely related to financial performance [121,132]. Based on this, we added several control variables related to social and environmental impacts, including the support for poverty student population (SPSP) refers to the number of students from low-income families sponsored by listed companies, the educational resources investment (ERI) refers to the amount of investment in educational resources in impoverished areas, and greenhouse gas emissions (GHGE). These data were obtained from the CSMAR database. Research has shown that, in the unique context of China, SOEs generally disclose information more broadly and extensively than non-SOEs [133]. Our data also confirmed this finding. Since most of the data came from SOEs, we removed the SOE variable. The final results are shown in

Table 7. Changing the control variables.

	(1)	(2)
VARIABLES	Model 1	Model 2
BI	33.103 ***	42.983 ***
	(6.85)	(6.99)
DT	2.786 ***	4.186 ***
	(7.13)	(5.76)
BI × DT		−2.948 *
		(−2.01)
Control	Yes	Yes
GHGE	0.000 ***	0.000 ***
	(4.30)	(4.27)
SPSP	0.001 ***	0.001 ***
	(3.31)	(4.23)
ERI	0.000 ***	0.000 ***
	(5.04)	(5.89)
Year	Yes	Yes
Firm	Yes	Yes
Constant	−29.080 ***	−34.368 ***
	(−6.13)	(−6.88)
Observations	33	33
R-squared	0.991	0.993

Note: Fixed effects models with modified control variables. Results are clustered by firms. T-statistics are in parentheses. Definition of the variables is in Table 1. * p < 0.1. *** p < 0.01.

, and our conclusions remain robust.

Lastly, we conducted two additional robustness tests. First, considering that the correlation between different sample clusters may affect the regression results, we extended the clustering to the industry level [119,134]. Appendix A 

Table A2. Additional robustness test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
VARIABLES	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
BI	0.591 **	0.901 **	0.509 **	0.850 ***	0.578 ***	0.980 ***	0.471 **	0.897 ***
	(2.17)	(2.55)	(2.26)	(2.90)	(2.65)	(3.48)	(2.18)	(3.22)
DT	0.041 ***	0.118 ***	0.043 ***	0.127 ***	0.033 ***	0.133 ***	0.037 ***	0.142 ***
	(4.50)	(3.04)	(4.48)	(2.82)	(3.66)	(3.10)	(4.05)	(3.32)
BI × DT		−0.207 **		−0.227 *		−0.266 **		−0.281 **
		(−2.00)		(−1.93)		(−2.38)		(−2.52)
CEOD	0.028	0.028	0.030	0.029	0.020	0.019	0.021	0.021
	(1.28)	(1.25)	(1.32)	(1.30)	(0.92)	(0.90)	(1.01)	(0.98)
SOE	0.029	0.028	0.024	0.023	0.001	−0.000	0.008	0.007
	(0.44)	(0.43)	(0.42)	(0.41)	(0.02)	(−0.00)	(0.15)	(0.13)
BS	0.007	0.007	0.006	0.007	0.007	0.007	0.005	0.006
	(0.69)	(0.70)	(0.75)	(0.76)	(0.80)	(0.83)	(0.63)	(0.67)
LEV	−0.655 ***	−0.653 ***	−0.685 ***	−0.683 ***	−0.721 ***	−0.719 ***	−0.740 ***	−0.739 ***
	(−6.45)	(−6.42)	(−9.45)	(−9.43)	(−10.28)	(−10.27)	(−10.70)	(−10.69)
EC	0.238	0.236	0.295	0.292	0.301 *	0.299	0.335 *	0.333 *
	(1.10)	(1.09)	(1.58)	(1.57)	(1.65)	(1.64)	(1.83)	(1.82)
NCID	−0.053 ***	−0.053 ***	−0.041 *	−0.042 *	−0.047 *	−0.047 **	−0.035	−0.036
	(−2.70)	(−2.70)	(−1.67)	(−1.68)	(−1.95)	(−1.96)	(−1.49)	(−1.50)
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year × Province			Yes	Yes			Yes	Yes
Year × Industry					Yes	Yes	Yes	Yes
Constant	4.014 ***	3.897 ***	4.049 ***	3.922 ***	4.064 ***	3.912 ***	4.107 ***	3.947 ***
	(19.44)	(16.76)	(27.65)	(24.06)	(28.76)	(24.93)	(29.28)	(25.36)
Observations	25,317	25,317	25,311	25,311	25,268	25,268	25,262	25,262
R-squared	0.527	0.527	0.541	0.541	0.579	0.579	0.592	0.592

Note: Fixed effects models. Results are clustered by industry in columns (1) and (2), while others are clustered by firms. T-statistics are in parentheses. Definitions of the variables are in Table 1. * p < 0.1. ** p < 0.05. *** p < 0.01.

reports the results, and our conclusions remain robust even when clustering at the industry level. Second, to address potential bias in coefficient estimates caused by unobserved regional or industry-level factors, we implemented several measures to enhance the robustness of our analysis. Specifically, we incorporated fixed effects for industry-year and region-year variables, and clustered standard errors at the firm level to account for hidden characteristics [23,124]. The results, presented in Appendix A Table A2, indicate that even with the inclusion of these high-dimensional fixed effects, our findings remain robust and consistent.

4.4.2. Heterogeneity Analysis

In this section, we further examine the differences in the substitutive effect of BI and DT in promoting ESGP under different circumstances, specifically focusing on corporate ownership heterogeneity and industry heterogeneity. To test for corporate ownership heterogeneity, we divided the firms in our sample into SOEs and non-SOEs and then performed regression analysis on these subsamples. The results are shown in columns 1–4 of

Table 8. Heterogeneity analysis.

	State-Owned		Non-State-Owned		Manufacturing		Non-Manufacturing
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
VARIABLES	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
BI	0.723 **	1.294 ***	0.379	0.468	0.373	0.533	0.889 **	1.545 ***
	(2.19)	(3.08)	(1.21)	(1.08)	(1.33)	(1.49)	(2.30)	(2.85)
DT	0.051 ***	0.216 ***	0.029 **	0.049	0.040 ***	0.085	0.033 *	0.173 **
	(3.33)	(2.82)	(2.42)	(0.82)	(3.50)	(1.35)	(1.93)	(2.42)
BI × DT		−0.446 **		−0.052		−0.119		−0.372 **
		(−2.20)		(−0.34)		(−0.72)		(−2.02)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	4.087 ***	3.876 ***	4.054 ***	4.019 ***	4.137 ***	4.075 ***	3.987 ***	3.753 ***
	(18.09)	(15.70)	(20.70)	(17.54)	(22.35)	(19.83)	(16.02)	(13.15)
Observations	9610	9610	15,639	15,639	16,486	16,486	8808	8808
R-squared	0.548	0.548	0.533	0.533	0.528	0.528	0.554	0.555

Note: Fixed effects models. Results are clustered by firms. T-statistics are in parentheses. Definitions of the variables are in Table 1. * p < 0.1. ** p < 0.05. *** p < 0.01.

. Our findings indicate that the substitutive effect of BI and DT on promoting ESGP is significant for SOEs, while the effect is not significant for non-SOEs.

Furthermore, to test for industry heterogeneity, we divided the firms in our sample into MEs and non-MEs and then performed regression analysis on these subsamples. The findings are presented in columns 5–8 of Table 8. The results indicate that the substitutive effect of BI and DT in promoting ESGP is not significant for MEs, while the effect is significant for non-MEs.

5. Discussion

This study provides an in-depth analysis of the combined impact of BI and DT on corporate ESGP, exploring the complementary and substitutive relationships between the two in enhancing corporate ESGP.

First, our findings support the substitution effect hypothesis. Specifically, the two supervisory mechanisms act as substitutes in promoting ESGP, which is contrary to the findings of Lu et al. [41] and Meng et al. [34]. When the level of DT is high, increasing BI is not necessary for improving ESGP. Similarly, when BI is high, increasing DT does not necessarily encourage companies to invest more in ESG practices. Although both mechanisms independently serve as supervisory mechanisms to reduce agency costs [13,53] and can individually promote ESGP [50,103], the adoption of multiple supervisory mechanisms can increase costs, and, beyond a certain point, additional governance measures may lead to diminishing marginal returns [40]. The mechanism behind the substitution effect may reflect the scarcity of governance resources and the effectiveness of supervisory tools. For example, based on resource dependence theory, companies tend to prioritize governance tools that maximize resource utilization. In some cases, DT can provide real-time and efficient supervisory functions, thereby reducing the reliance on traditional governance mechanisms such as independent directors. Another explanation is social embeddedness theory, which focuses on the role of social networks and relationships in corporate governance. According to this theory, the role of independent directors is primarily realized through their network of relationships with external stakeholders [116]. They often serve as important bridges connecting the company with the external environment, providing market information, policy advice, and strategic recommendations. However, with the advancement of DT, especially as companies interact directly with stakeholders through social media, online platforms, and big data technologies, the role of independent directors in these areas may gradually diminish.

Second, through further analysis of the individual pillars of ESG, we found that the substitution effect of DT on BI is primarily reflected in the G pillar. This finding indicates that the substitution effect of DT and BI in improving ESGP mainly stems from their overlapping roles in corporate governance (supervision) mechanisms. Specifically, DT can enhance the quality of internal controls, information transparency, and governance efficiency through real-time monitoring and automated processes, thereby reducing reliance on traditional supervision mechanisms. In the realm of corporate governance, DT can replace certain supervisory functions of independent directors, particularly in areas involving information transparency and compliance. For instance, blockchain technology can create immutable transaction records and contracts, which not only simplify audit processes but also allow external regulators and stakeholders to more easily access and verify information, reducing the need for manual reviews by independent directors. This result suggests that companies aiming to enhance performance in the G pillar can rely more on DT, rather than being overly dependent on independent directors for supervision.

Third, digital supervision mechanisms may be more effective than traditional independent director supervision in corporate governance. Our explanation for this is that digital supervision can monitor and analyze large amounts of data in real time and automate processes, reducing human bias and errors, thereby improving efficiency and fairness. This form of supervision is not influenced by personal emotions or subjective judgments, ensuring consistency and transparency in the execution of standards. By contrast, traditional human supervision systems, such as independent board supervision, are limited by human attention, capability, and responsibility. In practice, principals are unable to completely monitor the actions and efforts of agents in terms of their thoroughness, timeliness, and coordination [135]. Therefore, compared to “human governance”, “digital governance” is more efficient, objective, and fair.

Furthermore, this substitution effect is more pronounced in SOEs and non-MEs. Perhaps this is primarily because, in non-SOEs, independent directors are often less independent due to concentrated ownership, where controlling shareholders significantly influence board appointments [136]. As shown in Table 8 (columns 3 and 4), this weakens the supervisory role of independent directors, resulting in a reduced impact on ESGP, making the substitutive effect of BI on DT less significant. By contrast, MEs, which face stricter regulatory requirements, especially in environmental protection [21], often prioritize environmental strategies. As a result, their focus on governance is reduced, leading to a diminished role for independent directors in improving ESGP [137]. However, as discussed earlier in the analysis of individual ESG pillars, the substitutive effect of BI and DT on ESGP is primarily reflected in the G pillar. The data in Table 8 (columns 5 and 6) support this, showing an insignificant relationship between BI and ESGP. In summary, in non-SOEs, weak BI limits the supervisory role of independent directors, reducing their substitutive effect on DT in improving ESGP. In MEs, the emphasis on environmental management over governance similarly weakens this effect.

It is worth noting that although we found a substitution effect between BI and DT, this does not imply that their combination is ineffective in all contexts. On the contrary, specific contexts—such as industry characteristics and ownership structure—may moderate the strength of this substitution effect. In certain cases, BI and DT may function simultaneously and exhibit complementarity. These differences suggest that companies must flexibly adjust their supervisory mechanisms based on specific factors such as industry characteristics and ownership structure to optimize ESGP. Therefore, future research could further explore the contexts in which the complementarity between BI and DT may be more pronounced, providing companies with more precise governance strategy recommendations.

6. Conclusions and Implications

Our research findings indicate that BI and DT act as substitute governance mechanisms in their impact on ESGP. Further analysis reveals that this substitution effect is most evident within the governance pillar of ESG, with DT having a stronger substitutive role compared to BI. Additionally, the heterogeneity analysis reveals that this effect is more prominent in SOEs and non-MEs.

Based on the empirical findings, we propose several theoretical and managerial implications. In terms of theoretical implications, first, this study expands the existing literature on the impact of corporate governance and DT on ESGP by revealing the substitution effect between BI and DT in enhancing ESGP. This finding deepens our understanding of corporate governance mechanisms, particularly the interaction between traditional oversight mechanisms (such as BI) and emerging technology-driven mechanisms (such as DT). The research shows that employing multiple oversight mechanisms may not always be the optimal strategy, as it can lead to increased governance costs and diminishing marginal returns [75]. Through an analysis of the different dimensions of ESG, this study finds that the substitution effect is most prominent in the G dimension, highlighting the importance of DT in modern corporate governance. This suggests that companies can improve governance efficiency by optimizing digital oversight mechanisms to partially replace traditional governance methods. Additionally, the study finds that the substitution effect exhibits particular characteristics in SOEs and non-MEs. Future research could further explore the generalizability of this effect across different countries and industries. While this research focuses on the Chinese market, the theoretical logic of the substitution effect is applicable internationally, especially in the context of rapid globalization and digitalization. We suggest that future cross-national studies verify the applicability of this mechanism under various legal, cultural, and economic systems to enhance its theoretical generalizability. This study provides new directions for future research, recommending continued exploration of the substitution effect of different oversight mechanisms in diverse contexts and how to achieve optimal configuration of oversight mechanisms within companies.

From a managerial perspective, the findings have important implications for corporate managers and policymakers. Companies need to gain a deeper understanding of how various governance mechanisms interact to improve corporate ESGP. To achieve this, they must perform strategic cost–benefit analyses, as modifying or adopting governance practices incurs costs [76]. For example, when advancing DT, companies need to carefully consider its impact on the existing governance structure. For firms that already have a significant proportion of independent directors, additional digital supervisory mechanisms may reduce the marginal benefits of additional supervision [77]. Therefore, companies should appropriately adjust and optimize their board structures during DT to achieve the best allocation of resources. Especially for SOEs and non-MEs, special attention should be paid to the substitutive effects between DT and BI when promoting ESG practices. For SOEs, where BI typically plays a stronger supervisory role, it is important to balance the relationship between traditional and emerging digital supervision mechanisms during the DT process, aiming to create synergies when implementing supervisory mechanisms to maximize ESGP. For non-MEs, which tend to invest more in corporate governance, it is equally important to find an optimal balance between “human governance” and “digital governance” to ensure that supervisory mechanisms function efficiently overall [78].

While this study uncovers the substitution effect between BI and DT in improving ESGP, it has a few limitations. First, since the analysis primarily focuses on Chinese-listed companies, this may limit the broader applicability of the findings to different economic, regulatory, and cultural contexts. While the insights gained from China’s rapidly developing ESG landscape and DT efforts are valuable, future research should consider extending the sample to include companies from diverse regions. This would help assess whether the conclusions hold across various international contexts challenges, thereby increasing the generalizability and practical relevance of the results. Second, this study only evaluates BI and DT as governance mechanisms. Future studies could investigate other governance mechanisms, such as board diversity or ESG committees, to gain a more complete understanding of how corporate governance impacts ESG outcomes. Third, as this study utilizes the Huazheng ESG rating system to measure ESGP, the robustness of the conclusions is constrained by this specific standard. If data from other rating agencies were employed, the reliability of the findings could potentially differ. To address the variation between rating systems, future research could apply the “Sustainable Value Added” (SVA) approach, proposed by Figge and Hahn [138], to assess ESGP. The SVA method integrates economic, environmental, and social value creation to evaluate a company’s contributions to sustainable development, which could provide valuable insights in future studies [139,140].