Multiple Large Shareholders and ESG Performance: Evidence from Shareholder Friction

Abstract

Sustainable corporate governance increasingly influences corporate strategy considerations. Effective governance ensures organizational sustainability, with ESG being a crucial component. Large shareholders, as direct stakeholders, have a key role in developing and implementing corporate ESG strategies. Using data on Chinese listed firms over the 2011–2022 period, we find that multiple large shareholders (MLS) depress company ESG performance, suggesting that MLS may lead to friction and high coordination costs. Interestingly, stronger controlling shareholders mitigate this negative impact, particularly when they are state-owned. Our analysis shows that relatively equal power among MLS exacerbates friction, resulting in unstable executive teams and higher internal pay gaps, which lower governance (G) and social (S) scores. However, the presence of foreign and institutional investors among the large shareholders can alleviate these issues. The negative effect of MLS on ESG is significant in firms operating in clean industries, those with low analyst attention, or those not part of the “Stock Connect Scheme”. This study highlights the drawbacks of MLS in sustainable corporate governance from an ESG perspective.

1. Introduction

Over the past two decades, ESG has become an important concept in developing the modern economy. Stakeholder theory suggests that ESG investing shifts the corporate objective from maximizing corporate profits to integrating both economic and social values. By doing so, it more effectively aligns the interests of various stakeholders, including shareholders, managers, employees, suppliers, consumers, and society [1]. When making investment decisions, investors are increasingly evaluating companies based on their performance in ESG factors. Numerous researchers have investigated ESG behaviors at the corporate level, exploring the reasons behind and economic effects of ESG. Prior research has shown that companies meeting environmental and social goals operate more efficiently [2], are less vulnerable to managerial misconduct [3], face lower litigation risk [4], and generate higher firm value [5]. As a result, ESG investments can generate higher returns for investors [6].

A company’s ESG activities are fundamentally influenced by its corporate governance [7]. As a crucial component of corporate governance, ownership structure affects a company’s investment strategy and operations, therefore impacting its ESG activities. Previous research has investigated the impacts of different ownership structures on environmental and social responsibility performance [8,9,10]. These studies find that heterogeneity in ownership structure has diverse consequences on corporate ESG performance.

One notable feature of ownership structure in China is the presence of a dominant controlling shareholder (CS) [5,11]. This has drawn significant research attention to Type II agency problems [12,13]. Conflicts between controlling and non-controlling shareholders are intensified by the separation of control rights from cash flow rights [14]. MLS may be a solution to these problems. However, the existing literature presents an ambiguous view of the role of MLS. Some researchers argue that MLS can effectively mitigate agency problems and protect the interests of non-controlling shareholders. MLS can restrain CS encroachment by increasing dividends [15] and monitoring tunneling behaviors [16], therefore enhancing investment effectiveness [17] and increasing firm value [18]. However, other studies suggest that MLS may create problems. Collusion among MLS can harm the interests of other large shareholders [19,20]. Friction among MLS may lead to high coordination costs, hindering effective management oversight [21] and potentially decreasing firm value.

Firms engaging in ESG activities contribute to long-term sustainability but may do so at the expense of short-term interests. Conflicts regarding ESG, driven by large shareholders’ preferences or inconsistent interest objectives, can complicate the implementation of ESG strategies and impede organizational coherence. Whether MLS has sufficient incentives and capabilities to promote the implementation of corporate ESG strategies remains a critical question.

Current studies have generated conflicting conclusions on the relationship between MLS and corporate responsibility, considering both monitoring and collusion perspectives [22,23,24,25]. However, the existing studies have several limitations. First, they generally assume that CS has a negative influence, emphasizing that CS may harm a firm’s ESG performance through rent-seeking behavior and short-term profit maximization. Earlier research on stakeholders also argues that shareholder primacy theory conflicts with stakeholder theory, suggesting that shareholders act in their own interests at the expense of other stakeholders [26]. However, such studies tend to neglect the significance of CS as long-term investors, particularly in the Chinese context. In China, ESG implementation is often government-led and may benefit from government-related subsidies. To obtain long-term government support, CS with long-term shareholding can be a significant driver for ESG development. While the stakeholder perspective was initially seen as conflicting with the shareholder perspective, the prevailing view has evolved to emphasize mutual benefits [27], especially for long-term and active shareholders [28]. CS often embodies both of these characteristics. Therefore, the bias against CS needs correction. Second, existing research generally views the role of other large shareholders as positive, primarily because they can monitor CS. However, this perspective overlooks a key issue: the potential conflict of interest between CS and other large shareholders. While CS and other large shareholders are the most direct stakeholders in the company, these shareholders may pursue different strategic objectives and investment returns, leading to internal conflicts that negatively affect firms’ ESG performance. These research gaps need to be addressed to fully understand the roles and mechanisms of MLS in firms’ ESG performance.

To address these questions, we analyze a dataset comprising A-share listed companies in China from 2011 to 2022. We examine the impact of MLS on ESG performance, particularly from a friction perspective. Our analysis suggests that MLS impedes ESG performance, possibly due to the contestation of control rights and the high coordination costs associated with MLS.

We then conduct a moderating analysis and find that increased ownership of CS can mitigate the adverse effects of MLS on ESG performance. This finding suggests that a relative concentration of ownership reduces friction among large shareholders, thus reducing the adverse impact of MLS on ESG performance. Moreover, the positive influence is more pronounced when the CS is state-owned, indicating the government’s role in promoting ESG.

Mechanism analysis reveals that MLS friction reduces executive oversight, leading to decreased executive team stability and increased employee pay gaps. This, in turn, results in lower social and governance ratings, ultimately diminishing a firm’s overall ESG performance.

The heterogeneity analysis of large shareholders indicates that the adverse impact of MLS on ESG can be mitigated when foreign shareholders or institutional investors are part of the MLS. The cross-sectional analysis indicates that the negative effect of MLS on ESG is significant when a firm operates in a clean industry, receives low attention from analysts, or is not involved in the “Stock Connect Scheme”.

Our research contributions are as follows: First, our study augments existing research on the role of MLS from an ESG perspective. Previous studies generally suggest that coordination among MLS is too costly [21,29,30]. Our findings confirm that the divergence of interests and friction costs hinder ESG operations. Second, we provide new evidence supporting Edmans [27] that CS can play a leading role in promoting corporate ESG investments. Active shareholder engagement in corporate governance can improve a company’s ESG outcomes [31]. In Western countries, institutional investors such as mutual and insurance funds are major shareholders and hold significant ownership. Consequently, research on shareholder activism has primarily concentrated on the role of institutional investors. However, the China Securities Regulatory Commission (CSRC) imposes restrictions on the shareholding of institutional investors. Therefore, shareholder activism in China is less likely to be driven by institutional investors and more likely to be realized through CS. By analyzing data on the ownership structure of China, we confirm the positive effects of CS on ESG performance and shed new light on the role of ownership concentration. Third, although previous studies suggest that MLS contributes to enhancing ESG performance [24,32], our empirical results challenge this general perception. We explain this phenomenon as friction problems stemming from the divergence of interests among MLS. This reveals the problems of MLS and calls for further research to explore the full scope of the MLS governance mechanism.

The structure of the paper is as follows: Section 2 reviews the literature review and outlines the hypotheses. Section 3 details the research methodology. Section 4 reports the findings from the empirical analysis. Section 5 explores the mechanism underlying the results. Section 6 provides additional analysis. Section 7 concludes our research and provides policy recommendations.

2. Literature Review and Hypothesis Development

2.1. Multiple Large Shareholders

Current studies on MLS show both positive and negative impacts of corporate governance. Attig et al. [33] find that MLS plays a positive role in governance, as large shareholders can monitor each other and balance control rights to restrain tunneling or self-interested behaviors of CS [34]. MLS reduces the encroachment of minority shareholders by curbing excessive financial investment [35], improving earnings disclosure quality [36], and restricting dividend overpayment [15]. This improves corporate investment efficiency [17] and firm value [37]. Other large shareholders can also actively participate in governance using “voting-by-feet” or exiting threats against misconduct by CS [33,38,39].

However, some studies propose the view of “collusion” and argue that MLS may form alliances to pursue private benefits and jointly encroach on minority shareholders [16,18,39]. In addition, Jiang et al. [40] provide another reason for the ineffectiveness of MLS: the “cost-sharing” effect. They argue that the CS benefits from accrual earnings management but shifts the costs to other large shareholders when penalties are imposed. This mismatch between benefits and costs increases the likelihood of a firm engaging in earnings management.

Another study finds that MLS has high friction costs. Coordination and communication among large shareholders may overlook their supervisory role, allowing executives to gain private benefits [21]. Executives have an information advantage, enabling them to exploit potential conflicts and control disputes among MLS. They may intentionally provide misleading information to certain large shareholders and collude with the CS to gain greater managerial power and benefits [41]. For example, Zhang et al. [13] report a positive association between MLS and cost stickiness due to coordination problems among MLS.

In summary, the governance role of MLS remains inconclusive. The benefits and drawbacks of MLS in corporate governance require further investigation.

2.2. Shareholder and ESG Performance

Several studies discuss the differential impact of shareholders on corporate ESG performance. We review the relevant literature mainly from the perspectives of shareholder activism, long-term shareholders, and state-owned shareholders.

Shareholder activism suggests that shareholders are enthusiastic about ESG [42]. Shareholders actively pursue ESG projects and use their influence to ensure that firms fulfill their social responsibilities. ESG-minded shareholders align with market demands, enhancing a firm’s governance and operating performance [7,32]. Moreover, companies with poor ESG performance in the past can improve their ESG ratings through active shareholder engagement [31].

Compared to short-term shareholders, long-term shareholders are more likely to choose responsible investment projects [4]. Kavadis and Thomsen [43] highlight that long-term shareholders positively influence corporate social responsibility and sustainability practices. Many studies provide evidence that long-term shareholders positively influence ESG performance. Kim et al. [28] report that active, long-term institutional investors significantly improve CSR performance. Pressure-resistant institutional investors focus on long-term interests and encourage corporate ESG activities, while institutional investors motivated by short-term gains may discourage ESG activities due to their pressure [44]. Studies on family ownership have shown that long-term horizons driven by concerns for socioemotional wealth and care for future generations lead to a greater preference for environmental investments [2,45].

China is a socialist country where the government significantly influences economic and social development, with an increasing emphasis on ESG. Undoubtedly, the government’s attitude significantly impacts enterprises, particularly state-owned enterprises (SOEs). Kavadis and Thomsen [43] find that SOEs have a long-term orientation that enhances ESG performance. The political motivation of governments to promote environmental and social responsibility encourages companies to actively pursue ESG goals [10]. Lin et al. [46] categorize Chinese listed SOEs as controlled by central and local governments. They find that centrally controlled SOEs emphasize social benefits more than locally controlled SOEs or non-SOEs. Additionally, these centrally controlled SOEs are more willing and able to engage in ESG activities.

In summary, shareholders can play an active role in promoting ESG performance, especially active shareholders with a high degree of control rights, which allows them to directly participate in the company’s ESG decisions and ultimately leads to a better corporate ESG rating.

2.3. Hypothesis Development

Current studies present conflicting conclusions between MLS and ESG performance from both monitoring and collusion perspectives. Advocates of the “monitoring effect” argue that other large shareholders can constrain the opportunistic behaviors of CS by encouraging green operations and innovation. This enhances corporate environmental performance [22], improves the quality of firms’ CSR reports [23], and results in better ESG performance [24]. Conversely, proponents of the “collusion effect” argue that MLS form alliances to maximize their wealth, which demotivates a company’s ESG efforts [25]. In practice, however, MLS often leads to friction and conflict due to the divergent interests and objectives of MLS. These disagreements may negatively affect corporate decision-making efficiency [47]. Chakraborty and Gantchev [48] report that in companies with MLS, no individual shareholder may have sufficient voting power to ensure the success of their favored agendas at shareholder meetings. Different large shareholders have diverse objectives, making it difficult to reach a consensus [49]. While MLS can exchange information and negotiate with each other, significant conflicts of interest and bargaining efforts increase coordination costs. Large shareholders may hesitate or disagree when making decisions on ESG projects with long investment cycles and low short-term returns, even if these projects are beneficial in the long term [43]. CS may lack sufficient votes to endorse favorable ESG programs and must consult other large shareholders. This prolongs the decision-making process and reduces the efficiency of ESG decisions.

Furthermore, the focus of MLS on resolving conflicts can reduce their oversight of managers, increasing the potential for opportunistic behavior by executives. Fang et al. [21] discover that firms with large shareholders possessing comparable voting power tend to offer higher excess compensation. This means that coordination friction among MLS diminishes the monitoring efficiency of large shareholders. Wang and Wang [50] indicate that MLS’s coordination costs may weaken the supervision of managers, potentially resulting in divergence from optimal capital structure dynamics. Poor executive oversight may cause a focus on short-term benefits rather than long-term sustainable development, therefore reducing a company’s ESG performance.

Therefore, our working hypothesis can be summarized as follows:

Hypothesis 1 (H1). 

MLS leads to a decrease in ESG performance due to coordination friction.

While frictions among MLS lead to lower ESG performance, CS may play an important role in moderating these frictions. Edmans [27] argues that shareholder value comprises both short-term returns and long-term benefits. The benefits of “shareholder-priority values” are not exclusive to other stakeholders, and there is no fundamental difference between ESG investments and other types of investments. CS has a decisive influence on a company’s governance structure and decision-making. Their attitude toward ESG significantly determines a company’s ESG performance. By leveraging their influence, CS can adjust the company’s strategic direction, operational mode, and investment decisions to achieve higher ESG standards. Kim et al. [28] and Liu et al. [44] confirm that CS is willing to abandon short-term gains to enhance a company’s long-term value. These studies propose that long-term investors are willing to sacrifice certain short-term benefits to enhance corporate social responsibility and build a reputation for long-term returns. Consequently, CS can benefit from corporate ESG investments in the future, even if such investments reduce current profitability [43].

Therefore, we pose the following hypothesis:

Hypothesis 2 (H2). 

CS can mitigate friction among MLS and thus reduce its negative impact on ESG performance.

The Chinese government is vital for promoting ESG performance for SOEs. The government encourages SOEs to fulfill social responsibilities through policy implementation and target setting. SOEs have more substantial economic resources than non-SOEs. This financial advantage allows SOEs to invest more heavily in ESG activities. The government’s evaluation mechanisms incentivize SOEs to achieve both financial and social objectives, mitigating the negative impacts of complex corporate structures on ESG performance. Additionally, as SOEs are controlled by the state, they can effectively promote ESG objectives even with MLS. Unified policy guidance, strong regulatory mechanisms, and sufficient resource support ensure that the interests of all parties are aligned in terms of ESG, reducing conflicts and friction.

Therefore, we pose the following hypothesis:

Hypothesis 3 (H3). 

State-controlled shareholders more significantly reduce MLS friction.

3. Data and Methodology

3.1. Data and Sample

Our dataset comprises Chinese A-share listed companies spanning the years 2011 to 2022. We obtain ownership structure and financial data from the CSMAR database. Firms from the financial industry are excluded due to their unique operational features and regulatory requirements. We exclude firms with “special treatment” status, firms without MLS (i.e., all shareholders hold less than 10% of the shares), and firms with missing data in the database. Our final dataset includes 2717 companies with 25,767 firm–year observations. To address the impact of potential outliers, we apply Winsorization to all continuous variables at the 1st and 99th percentiles.

3.2. Variable Definition

3.2.1. Dependent Variable: ESG Performance

We use the ESG index provided by Sino-Securities Index Information Service (Shanghai) Co., Ltd. (Shanghai, China) to gauge corporate ESG performance. This index encompasses all Chinese listed firms and includes a broad range of indicators over multiple years. Scores on the ESG index range from 0 to 100, with higher values representing better ESG performance. The index is well-regarded within both industry and academic circles [46]. To facilitate the interpretability of the regression coefficients, we normalize the original ESG scores by dividing them by 10. Rescaling the data does not impact the statistical significance of the regression results. Consequently, the corporate ESG scores range from 0 to 10.

3.2.2. Independent Variable: Multiple Large Shareholders

Following prior studies [16,18,40], we define a major shareholder as an entity owning at least 10% of a company’s total shares. According to the Company Law of China, shareholders possessing over 10% have the right to call for an extraordinary general meeting. We merge affiliated entities into a single shareholder by aggregating their shareholdings based on related-party information. We define a binary variable, Multi, which equals 1 if the firm has two or more large shareholders, and 0 otherwise. In robustness tests, following the approach of Jiang et al. [35], we set the threshold for a large shareholder to more than 5% of total shares.

3.2.3. Control Variables

According to existing research [7,51], we select control variables that potentially affect corporate ESG performance. Specifically, Size is the natural logarithm of total assets at the fiscal year-end, Roa is the net income divided by total assets at the fiscal year-end, Lev is the ratio of total debt to total assets at the fiscal year-end, Age is the natural logarithm of the firm’s age since listing, Growth is the annual sales growth rate and Big4 is a dummy variable equals to 1 if a firm is audited by a “Big 4” accounting firm and 0 otherwise.

Prior research indicates that ownership structure and governance mechanisms significantly influence a firm’s ESG performance. Thus, we control for these variables as follows: Board is the natural logarithm of the number of board members, Indrt is the ratio of independent directors to the total number of board members, Dual is a dummy variable that equals 1 if the CEO also serves as the chairperson of the board and 0 otherwise, and HHI is calculated by squaring the ownership proportions of the top ten shareholders and summing these squared values to measure ownership concentration levels.

We also control for several variables related to environmental and social responsibility. CSR is a dummy variable that equals 1 if a firm discloses a CSR report in the year and 0 otherwise. Ggov refers to the level of corporate green management, constructed using five indicators as defined by Zhao et al. [52]. To assess this, we check whether a company holds either the ISO14001 or ISO9001 certifications listed in the CSMAR environmental database. Additionally, we verify whether the firm has established an eco-friendly management system, conducted eco-friendly education and training, and undertaken eco-friendly special actions. An aggregated score is then calculated by summing these indicators to serve as a proxy for Ggov. Ginv indicates the degree of corporate green innovation, defined as the ratio of a firm’s granted green patents to all patent applications in the year, following the approach of Amore & Bennedsen [53].

Table 1. Variable definition and measurement.

Variable	Definition and Measurement
ESG	ESG performance, taking the ESG rating scores developed by Sino-Securities database
Multi	A dummy variable that equals 1 if the firm has two or more large shareholders and 0 otherwise
Size	The natural logarithm of total assets at the fiscal year-end
Roa	The ratio of net income to total assets
Lev	The ratio of total debt to total assets
Age	The natural logarithm of the firm’s age since listing
Growth	The annual growth rate of sales
Big4	A dummy variable that equals 1 if the firm is audited by a “Big 4” accounting firm and 0 otherwise
Board	The size of the Board of Directors, represented by the natural logarithm of the number of directors
Indrt	Board independence, defined as the ratio of independent directors to the total number of directors
Dual	A dummy variable that equals 1 if the CEO is also the chairperson of the Board and 0 otherwise
HHI	Ownership concentration, defined as the sum of the squared shareholdings of the top 10 shareholders of the firm
CSR	A dummy variable that equals 1 if the firm discloses a CSR report in the year and 0 otherwise
Ggov	The level of corporate green management, measured by five indicators as defined by Zhao et al. (2015) [52]
Ginv	Green innovation, defined as the ratio of granted green patents to all patent applications in the year

contains detailed descriptions of the variables.

3.3. Research Design

We develop a baseline regression model to examine the link between MLS and ESG performance:

$$ES{G}_{i,t}={\alpha }_0+{\alpha }_{1}Mult{i}_{i,t}+\alpha Contro{l}_{i,t}+{\displaystyle \sum Year}+{\displaystyle \sum Ind}+{\epsilon }_{i,t}$$

(1)

where ESG represents the ESG performance of firm i in year t and Multi denotes whether firm i has at least two large shareholders in year t. We control for industry-fixed effects (Ind) and year-fixed effects (Year). We cluster standard errors at the firm level to alleviate the potential cross-sectional dependence effect.

For H1, the coefficient of Multi is of interest in Model 1. Accordingly, we posit that companies with MLS will experience more interest contestations and coordination frictions among large shareholders, which may lead to lower ESG rating scores. We expect a negative coefficient of Multi in Model 1.

To test H2 and H3, we construct Model 2 to analyze the moderating role of CS, which is as follows:

$$ES{G}_{i,t}={\alpha }_0+{\alpha }_1{\mathrm{CS}}_{i,t}\times Mult{i}_{i,t}+{\alpha }_{2}C{S}_{i,t}+{\alpha }_{3}Mult{i}_{i,t}+\alpha Contro{l}_{i,t}+{\displaystyle \sum Year}+{\displaystyle \sum Ind}+{\epsilon }_{i,t}$$

(2)

where CS denotes the percentage of shares held by CS and is used to measure the power of CS. Based on H2, we anticipate the coefficient on the interaction term to be significantly positive. Based on H3, we also group the samples to examine the significance of the interaction term’s coefficient for SOEs and non-SOEs. According to H3, we anticipate that the interaction term’s coefficient will be significantly positive for SOEs.

4. Empirical Results

4.1. Descriptive Statistics

Table 2. Descriptive statistics.

VarName	Obs	Mean	Median	SD	Min	Max
ESG	25,767	7.318	7.345	0.532	3.662	9.093
Multi	25,767	0.247	0.000	0.431	0.000	1.000
Size	25,767	22.283	22.099	1.285	19.908	26.298
Roa	25,767	0.038	0.037	0.060	−0.261	0.195
Lev	25,767	0.421	0.414	0.198	0.056	0.893
Age	25,767	2.207	2.303	0.737	0.693	3.367
Growth	25,767	0.346	0.129	0.890	−0.703	6.551
Big4	25,767	0.054	0.000	0.227	0.000	1.000
Board	25,767	2.236	2.303	0.175	1.792	2.773
Indrt	25,767	0.376	0.364	0.053	0.333	0.571
Dual	25,767	0.285	0.000	0.451	0.000	1.000
HHI	25,767	0.158	0.129	0.111	0.015	0.555
CSR	25,767	0.295	0.000	0.456	0.000	1.000
Ggov	25,767	1.050	1.000	1.266	0.000	5.000
Ginv	25,767	0.067	0.000	0.130	0.000	0.667

presents the descriptive statistics for the main variables. In our sample, 24.7% of the firm–year observations have more than one major shareholder. The average (median) ESG rating score is 7.318 (7.345) with a standard deviation of 0.532, indicating variation in ESG performance across sample companies. The minimum and maximum ESG scores are 3.662 and 9.093, implying a wide range of differences in ESG performance, even though the average scores are relatively high. Our sample statistics are consistent with those of previous studies using Chinese data [3,13,52,54].

4.2. Univariate Analysis

Table 3. Results of univariate analysis.

Varname	Multi = 0		Multi = 1		Mean-Diff	t
	Obs	Mean	Obs	Mean
ESG	19,412	7.322	6355	7.306	0.015 **	2.004
Size	19,412	22.247	6355	22.393	−0.146 ***	−7.879
Roa	19,412	0.037	6355	0.039	−0.001	−1.487
Lev	19,412	0.424	6355	0.414	0.009 ***	3.247
Age	19,412	2.231	6355	2.131	0.100 ***	9.407
Growth	19,412	0.357	6355	0.311	0.046 ***	3.563
Big4	19,412	0.038	6355	0.105	−0.067 ***	−20.742
Board	19,412	2.229	6355	2.259	−0.030 ***	−11.869
Indrt	19,412	0.377	6355	0.373	0.005 ***	6.174
Dual	19,412	0.289	6355	0.270	0.020 ***	3.021
HHI	19,412	0.161	6355	0.148	0.012 ***	7.778
CSR	19,412	0.287	6355	0.321	−0.033 ***	−5.083
Ggov	19,412	1.028	6355	1.115	−0.087 ***	−4.754
Ginv	19,412	0.065	6355	0.071	−0.006 ***	−3.112

** or *** indicates a significance level at 5% and 1%, respectively.

summarizes the test results for differences in ESG score means between firms with MLS and those with a single large shareholder (SLS). The mean ESG score is 7.306 for MLS and 7.322 for SLS, with a mean difference of 0.015. The mean ESG score for MLS is significantly lower than that for SLS at the 5% level.

Table 3 also contrasts the characteristics of firms with MLS and those with SLS. Due to substantial variance between the two subgroups, we will conduct multivariate analysis, propensity score matching, and a firm-fixed effect test to control for the impact of different firm characteristics and address endogeneity concerns.

4.3. Correlation Analysis

Table 4. Correlation analysis.

	ESG	Multi	Size	Roa	Lev	Age	Growth	Big4	Board	Indrt	Dual	HHI	CSR	Ggov	Ginv
ESG	1
Multi	−0.012 **	1
Size	0.229 ***	0.049 ***	1
Roa	0.216 ***	0.009	0.011 *	1
Lev	−0.039 ***	−0.020 ***	0.523 ***	−0.340 ***	1
Age	−0.016 **	−0.059 ***	0.458 ***	−0.145 ***	0.344 ***	1
Growth	0.020 ***	−0.022 ***	0.029 ***	−0.017 ***	0.099 ***	0.057 ***	1
Big4	0.100 ***	0.128 ***	0.329 ***	0.039 ***	0.108 ***	0.097 ***	−0.014 **	1
Board	0.053 ***	0.074 ***	0.258 ***	0.034 ***	0.130 ***	0.151 ***	−0.025 ***	0.102 ***	1
Indrt	0.063 ***	−0.038 ***	−0.013 **	−0.019 ***	−0.016 **	−0.038 ***	0.031 ***	0.008	−0.563 ***	1
Dual	−0.032 ***	−0.019 ***	−0.177 ***	0.020 ***	−0.121 ***	−0.229 ***	−0.022 ***	−0.069 ***	−0.184 ***	0.132 ***	1
HHI	0.097 ***	−0.048 ***	0.201 ***	0.145 ***	0.046 ***	−0.091 ***	0.020 ***	0.162 ***	0.050 ***	0.024 ***	−0.053 ***	1
CSR	0.352 ***	0.032 ***	0.480 ***	0.067 ***	0.174 ***	0.283 ***	−0.022 ***	0.208 ***	0.172 ***	−0.007	−0.117 ***	0.110 ***	1
Ggov	0.262 ***	0.030 ***	0.227 ***	0.068 ***	0.062 ***	0.081 ***	−0.081 ***	0.118 ***	0.089 ***	−0.021 ***	−0.046 ***	0.041 ***	0.466 ***	1
Ginv	0.047 ***	0.019 ***	0.101 ***	−0.033 ***	0.091 ***	0.018 ***	0.018 ***	0.017 ***	0.025 ***	−0.015 **	−0.025 ***	−0.031 ***	0.037 ***	0.046 ***	1

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

displays the Pearson correlation coefficients for the primary variables, highlighting that ESG performance is significantly negatively correlated with Multi at the 5% level. Combined with the univariate analysis, the results initially support H1.

In addition, we conduct a VIF test for the main variables. The results show that the VIFs for all main variables are below 3, suggesting the absence of significant multicollinearity issues.

4.4. Multivariate Analysis

Table 5. Results of baseline regressions.

	(1)	(2)	(3)
	ESG	ESG	ESG
Multi	−0.0376 ***	−0.0372 ***	−0.0426 ***
	(−2.94)	(−2.93)	(−3.83)
Size	0.1564 ***	0.1489 ***	0.0878 ***
	(22.50)	(20.88)	(13.18)
Roa	1.3172 ***	1.3181 ***	1.1650 ***
	(13.87)	(14.03)	(13.57)
Lev	−0.4553 ***	−0.4498 ***	−0.4060 ***
	(−11.59)	(−11.50)	(−11.56)
Age	−0.0794 ***	−0.0812 ***	−0.0995 ***
	(−7.95)	(−7.97)	(−11.03)
Growth	−0.0074	−0.0080	−0.0040
	(−1.44)	(−1.56)	(−0.86)
Big4	0.0318	0.0251	−0.0205
	(1.14)	(0.92)	(−0.87)
Board		0.1769 ***	0.1142 ***
		(4.17)	(3.10)
Indrt		0.9501 ***	0.8516 ***
		(7.72)	(7.86)
Dual		−0.0245 **	−0.0153
		(−2.01)	(−1.40)
HHI		0.0322	0.0026
		(0.52)	(0.05)
CSR			0.2905 ***
			(21.09)
Ggov			0.0584 ***
			(14.35)
Ginv			0.1545 ***
			(4.29)
_cons	4.0505 ***	3.4654 ***	4.8521 ***
	(26.53)	(19.16)	(29.84)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
R-Square	0.185	0.191	0.277
Adj. R-Square	0.183	0.189	0.275
N	25,767	25,767	25,767

** or *** indicates a significance level at 5% and 1%, respectively.

shows the results from Model 1, which includes industry- and year-fixed effects. Column 1 shows that the results contain firm characteristic variables. Column 2 summarizes the results, including corporate governance and ownership characteristics. Column 3 reports the results with the addition of control variables related to environmental and social responsibility. The coefficient of Multi is noteworthy, as it is negative and significant at the 1% level across all columns, suggesting that MLS significantly hinders corporate ESG performance.

Table 5 demonstrates that larger Size and higher Roa result in better corporate ESG ratings. This suggests that larger and more profitable companies are more inclined to participate in ESG activities to enhance their corporate reputation and social image. Conversely, Lev is negatively related to ESG performance, indicating that firms with greater financing constraints are more hesitant to pursue ESG activities. Regarding the governance-related control variables, the better the level of governance is, the higher the ESG score. Exercising social responsibility and environmental protection strongly influence corporate ESG performance, as the related control variables are significantly and positively associated with ESG rating.

Overall, the results in Table 5 confirm our expectation that MLS leads to lower corporate ESG performance.

4.5. Moderating Analysis

To further investigate the power of the CS, we introduce the variable CS, which represents the shareholding ratio of the CS. A higher CS indicates a stronger incentive and ability for the CS to participate in corporate governance. Based on H2, we argue that CS can better mitigate the costs of MLS friction. We analyze the effect of the active participation of CS on ESG by interpreting the coefficients of the interaction term Multi × CS. Column 1 in

Table 6. Regression results for controlling shareholder engagement.

	(1)	(2)	(3)
	Full Sample	SOEs	Non-SOEs
	ESG	ESG	ESG
Multi	−0.1229 ***	−0.1287 ***	−0.0839 **
	(−4.13)	(−2.68)	(−2.29)
Top1	0.0452	−0.0819	0.0638
	(0.71)	(−0.61)	(0.88)
Multi × CS	0.2684 ***	0.3544 ***	0.1256
	(3.20)	(2.74)	(1.19)
Size	0.0868 ***	0.0828 ***	0.0929 ***
	(12.84)	(8.33)	(10.57)
Roa	1.1125 ***	1.1483 ***	1.0818 ***
	(12.69)	(6.51)	(11.07)
Lev	−0.4096 ***	−0.2961 ***	−0.4816 ***
	(−11.51)	(−5.17)	(−11.08)
Age	−0.0981 ***	−0.0408 **	−0.1497 ***
	(−10.23)	(−2.44)	(−11.96)
Growth	−0.0031	−0.0101	0.0031
	(−0.66)	(−1.51)	(0.46)
Big4	−0.0307	−0.0245	−0.0393
	(−1.31)	(−0.81)	(−1.07)
Board	0.1123 ***	0.0787	0.0581
	(3.00)	(1.48)	(1.12)
Indrt	0.8652 ***	1.0551 ***	0.6833 ***
	(7.77)	(6.59)	(4.69)
Dual	−0.0161	−0.0405 *	0.0041
	(−1.45)	(−1.75)	(0.33)
HHI	−0.0902	−0.0273	−0.1988 *
	(−1.00)	(−0.17)	(−1.76)
CSR	0.2915 ***	0.2705 ***	0.2952 ***
	(20.77)	(13.88)	(15.67)
Ggov	0.0580 ***	0.0562 ***	0.0580 ***
	(14.04)	(9.30)	(10.89)
Ginv	0.1535 ***	0.1751 ***	0.1256 ***
	(4.18)	(2.86)	(2.81)
_cons	4.8671 ***	4.7698 ***	5.0596 ***
	(29.34)	(20.57)	(21.83)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
R-Square	0.276	0.358	0.256
Adj. R-Square	0.274	0.353	0.252
N	24,836	8451	16,339

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

summarizes the regression results for the full sample. The coefficient of Multi × CS is significantly positive at the 1% level, indicating that an increase in ownership by CS helps mitigate the negative impact of MLS on ESG. This finding indirectly confirms that the active engagement of CS not only mitigates the friction caused by MLS but also improves a firm’s ESG performance, thus supporting H2.

Existing literature indicates that state-owned and non-state-owned shareholders have different objectives and interests. For example, SOEs are more concerned with ESG, leading to a higher prevalence of ESG reports among SOEs than among non-SOEs [55]. Li and Zhang [56] find a positive relationship between government ownership and the ESG activities of Chinese SOEs, suggesting that SOEs’ responses to ESG are more politically driven. Jia et al. [57] show that institutional investors have a stronger impact on improving ESG performance in state-owned enterprises. Therefore, we split the sample into two subgroups, SOEs and non-SOEs, and conducted separate regressions for each group. The findings are shown in columns 2 and 3 of Table 6. The coefficient of Mult × CS is significantly positive at the 1% level in the SOE group and significantly different between groups. Since the CS of SOEs is usually the government, we confirm that the CS of SOEs plays a greater role in promoting ESG performance. Thus, Hypothesis 3 is supported.

4.6. Robustness Tests

We employ the Heckman treatment effect model, staggered DID, instrumental variable estimation, regressions with firm–year-fixed effects, PSM-OLS, and other robustness tests to robust our results.

4.6.1. Heckman Treatment Effect Model

To address omitted variable bias and sample self-selection bias, we develop instrumental variables (IV) for MLS and apply the Heckman treatment effect model. In the first stage, we estimate the likelihood of having MLS using a probit model. In the second stage, we incorporate the inverse Mills ratio (IMR) estimated from the first stage into Model 1 to adjust for self-selection bias.

We construct two IVs, L.zzhs and Trading. L.zzhs indicates whether the firm was included in the CSI 300 or CSI 500 index in the previous year [58], as inclusion in these indices affects investor portfolios and the number of large shareholders but does not directly impact ESG performance. Trading represents the firm’s annual stock turnover rate. Typically, a higher turnover rate indicates a higher proportion of outstanding shares, making it easier for shareholders to reduce their holdings and exit. Consequently, the shareholding ratio of large shareholders is less likely to reach 10%. However, the stock turnover rate does not have a direct effect on ESG performance.

As shown in

Table 7. Results of Heckman treatment test.

	(1)	(2)	(3)	(4)	(5)	(6)
	First Stage	Second Stage	Second Stage	First Stage	Second Stage	Second Stage
L.zzhs	−0.1246 **
	(−2.38)
Trading				−0.0007 ***
				(−12.85)
Multi		−0.0420 ***	−0.1133 ***		−0.0468 ***	−0.1183 ***
		(−3.57)	(−3.47)		(−4.20)	(−3.80)
CS			0.1189			0.1211
			(1.32)			(1.43)
Multi × CS			0.2474 ***			0.2452 ***
			(2.67)			(2.81)
Size	0.1141 ***	0.0579 ***	0.0583 ***	0.0310	0.0801 ***	0.0799 ***
	(3.89)	(4.99)	(4.94)	(1.17)	(11.44)	(11.22)
Roa	−0.5171 *	1.3612 ***	1.3039 ***	−0.5914 **	1.2043 ***	1.1472 ***
	(−1.92)	(13.23)	(12.37)	(−2.23)	(13.79)	(12.82)
Lev	−0.4319 ***	−0.2718 ***	−0.2798 ***	−0.2401 *	−0.3716 ***	−0.3759 ***
	(−3.19)	(−4.96)	(−5.02)	(−1.85)	(−10.19)	(−10.18)
Age	−0.2468 ***	−0.0029	−0.0042	−0.2531 ***	−0.0762 ***	−0.0752 ***
	(−6.35)	(−0.10)	(−0.14)	(−7.49)	(−7.19)	(−6.94)
Growth	−0.0087	−0.0009	0.0001	−0.0019	−0.0037	−0.0028
	(−0.50)	(−0.19)	(0.03)	(−0.11)	(−0.80)	(−0.59)
Big4	0.6492 ***	−0.2467 ***	−0.2525 ***	0.6490 ***	−0.0722 ***	−0.0828 ***
	(6.61)	(−3.51)	(−3.55)	(6.63)	(−2.69)	(−3.06)
Board	0.4777 ***	−0.0574	−0.0574	0.4268 ***	0.0763 **	0.0743 *
	(3.10)	(−0.92)	(−0.91)	(2.90)	(2.02)	(1.94)
Indrt	0.1376	0.8330 ***	0.8502 ***	0.0157	0.8528 ***	0.8674 ***
	(0.30)	(7.29)	(7.22)	(0.04)	(7.86)	(7.78)
Dual	−0.0652	0.0056	0.0043	−0.0642	−0.0097	−0.0103
	(−1.56)	(0.41)	(0.31)	(−1.60)	(−0.88)	(−0.93)
HHI	−1.4812 ***	0.6066 ***	0.4115 *	−1.9468 ***	0.1342 **	−0.0494
	(−6.71)	(3.46)	(1.96)	(−8.92)	(2.17)	(−0.42)
CSR	0.0222	0.2845 ***	0.2855 ***	0.0057	0.2899 ***	0.2909 ***
	(0.40)	(19.82)	(19.53)	(0.11)	(21.02)	(20.71)
Ggov	0.0173	0.0531 ***	0.0528 ***	0.0185	0.0569 ***	0.0566 ***
	(1.13)	(11.41)	(11.19)	(1.26)	(13.90)	(13.62)
Ginv	0.0696	0.1192 ***	0.1172 ***	0.0907	0.1497 ***	0.1476 ***
	(0.51)	(3.10)	(2.98)	(0.70)	(4.15)	(4.02)
IMR		−0.5034 ***	−0.4928 ***		−0.1139 ***	−0.1149 ***
		(−3.52)	(−3.39)		(−4.20)	(−4.18)
_cons	−3.5749 ***	6.3209 ***	6.2841 ***	−1.3315 **	5.1891 ***	5.1794 ***
	(−4.93)	(13.76)	(13.41)	(−1.98)	(28.36)	(27.60)
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
R-Square		0.280	0.279		0.278	0.277
Adj. R-Square		0.278	0.277		0.276	0.275
N	23,050	23,050	22,180	25,761	25,761	24,830

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

, Columns 1 and 4 report the first-stage results. The coefficients on L.zzhs and Trading are significantly negative, satisfying the inclusion requirement. We regress ESG performance on IMR from the first stage, including Multi and all control variables. Columns 2 and 4 report a significantly negative coefficient for Multi, indicating that our findings remain robust. Columns 3 and 6 report significantly positive coefficients for Multi × CS, indicating that H2 is robust.

4.6.2. Staggered DID Analysis

To better reflect the causal relation between MLS and ESG performance, we use staggered DID. We compare ESG performance changes around a firm’s transition from SLS (MLS) to MLS (SLS) with those of firms that remain at SLS (MLS). Firms transitioning from SLS (MLS) to MLS (SLS) form the treatment group, while firms continuously maintaining SLS (MLS) serve as the control group. Subsequently, we construct the analysis model as follows:

$$ES{G}_{i,t}={\gamma }_0+{\gamma }_{1}Trea{t}_{i,t}\times Afte{r}_{i,t}+\rho Contro{l}_{i,t}{\displaystyle \sum Year}+{\displaystyle \sum Firm}+{ϵ}_{i,t}$$

(3)

In Model 3, Treat × After is a dummy variable indicating whether firm i transitions from SLS (MLS) to MLS (SLS) in year t. The results in Column 1 of

Table 8. Results of staggered DID.

	SLS to MLS			MLS to SLS
	(1)	(2)	(3)	(4)	(5)	(6)
	Full Sample	SOEs	Non-SOEs	Full Sample	SOEs	Non-SOEs
Treat × After	−0.0938 ***	0.0195	−0.1481 ***	0.0332	0.0595	0.0340
	(−3.63)	(0.49)	(−4.23)	(1.44)	(1.05)	(1.30)
Size	0.1277 ***	0.0929 **	0.1376 ***	0.1296 ***	0.1327 **	0.1199 ***
	(6.30)	(2.50)	(5.05)	(5.59)	(2.22)	(4.63)
Roa	0.8059 ***	0.7076 *	0.7002 ***	0.2481	0.9436 *	0.1411
	(4.10)	(1.68)	(3.18)	(1.54)	(1.91)	(0.80)
Lev	−0.2915 ***	0.0496	−0.3389 ***	−0.4829 ***	−0.3775 *	−0.3791 ***
	(−3.24)	(0.32)	(−2.92)	(−5.93)	(−1.79)	(−4.09)
Age	−0.0984 **	0.1702	−0.0570	−0.1086 **	−0.0558	−0.1163 **
	(−2.54)	(1.55)	(−1.12)	(−2.57)	(−0.61)	(−2.30)
Growth	−0.0260 **	−0.0548 ***	−0.0166	0.0068	−0.0152	0.0092
	(−2.06)	(−2.91)	(−1.03)	(0.64)	(−1.03)	(0.68)
Big4	0.0270	−0.0550	0.0204	−0.1189 *	−0.2260 *	−0.0645
	(0.43)	(−0.57)	(0.20)	(−1.93)	(−1.88)	(−0.82)
Board	0.2147 *	−0.2557	0.3989 ***	0.1734	−0.1304	0.1927
	(1.96)	(−1.24)	(2.69)	(1.53)	(−0.64)	(1.41)
Indrt	0.6838 **	−0.3419	0.9379 **	0.0211	0.4497	−0.1151
	(2.35)	(−0.66)	(2.49)	(0.07)	(0.88)	(−0.31)
Dual	−0.0283	−0.0460	−0.0284	−0.0152	−0.0844	−0.0204
	(−1.05)	(−0.86)	(−0.89)	(−0.65)	(−1.32)	(−0.78)
HHI	−0.0242	−0.3997	0.3569	0.0512	−0.2709	−0.0666
	(−0.14)	(−1.49)	(1.55)	(0.25)	(−0.87)	(−0.25)
CSR	0.1258 ***	0.1715 ***	0.1590 ***	0.1291 ***	0.2064 **	0.1260 ***
	(3.26)	(2.94)	(2.99)	(3.30)	(2.56)	(2.69)
Ggov	0.0405 ***	0.0366 ***	0.0364 ***	0.0241 ***	0.0230	0.0215 **
	(4.32)	(3.08)	(2.60)	(2.64)	(1.14)	(2.05)
Ginv	0.2671 ***	0.1108	0.3590 ***	0.1586 **	0.1926	0.1451
	(3.20)	(0.85)	(3.43)	(2.02)	(1.34)	(1.45)
_cons	3.9777 ***	5.3932 ***	3.1361 ***	4.3250 ***	4.6530 ***	4.5403 ***
	(8.05)	(5.44)	(4.58)	(7.19)	(3.34)	(6.58)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes	Yes	Yes
R-Square	0.603	0.682	0.614	0.559	0.607	0.562
Adj. R-Square	0.552	0.627	0.550	0.501	0.539	0.498
N	2683	965	1648	3387	673	2618

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

show that ESG scores significantly decline with the transition from SLS to MLS. Conversely, the results reported in Column 4 indicate that ESG scores increase but not significantly with the transition from MLS to SLS. This suggests that MLS demotivates a firm’s ESG effort and outcomes.

The subsample regression reveals that ESG performance in non-SOEs significantly decreases when transitioning from SLS to MLS. This finding suggests that MLS in non-SOEs is more likely to lead to a decline in ESG performance, therefore indirectly supporting H3.

4.6.3. Instrumental Variable Estimation

To further address potential endogeneity between MLS and ESG performance, we design instrumental variable estimation for regression analysis. Following Jiang et al. [17], we use the previous year’s industry average of Multi for other firms as the instrumental variable (Multi_IV). We anticipate that firms are more likely to have MLS when it is a common ownership structure among peer firms, expecting a positive relationship between Multi and Multi_IV. In addition, Multi_IV is not directly correlated with a firm’s ESG performance. We apply two-stage least squares (2SLS) regression and generalized method of moments (GMM) regression.

The results are shown in

Table 9. Instrumental variable method.

	(1)	(2)	(3)
	First Stage	Second Stage (2sls)	Second Stage (GMM)
Multi_IV	0.5904 ***
	(9.98)
Multi		−0.3491 ***	−0.3491 ***
		(−3.44)	(−3.44)
Size	0.0256 ***	0.0983 ***	0.0983 ***
	(3.16)	(20.67)	(20.67)
Roa	−0.1644 **	1.1125 ***	1.1125 ***
	(−1.99)	(15.83)	(15.83)
Lev	−0.1129 ***	−0.4560 ***	−0.4560 ***
	(−2.81)	(−17.72)	(−17.72)
Age	−0.0735 ***	−0.1223 ***	−0.1223 ***
	(−6.54)	(−12.82)	(−12.82)
Growth	−0.0015	−0.0048	−0.0048
	(−0.31)	(−1.23)	(−1.23)
Big4	0.2254 ***	0.0537 **	0.0537 **
	(6.27)	(1.97)	(1.97)
Board	0.1389 ***	0.1623 ***	0.1623 ***
	(2.98)	(5.80)	(5.80)
Indrt	0.0413	0.8785 ***	0.8785 ***
	(0.31)	(11.86)	(11.86)
Dual	−0.0189	−0.0251 ***	−0.0251 ***
	(−1.54)	(−3.26)	(−3.26)
HHI	−0.4042 ***	−0.1000 *	−0.1000 *
	(−7.10)	(−1.90)	(−1.90)
CSR	0.0032	0.2897 ***	0.2897 ***
	(0.19)	(33.22)	(33.22)
Ggov	0.0049	0.0609 ***	0.0609 ***
	(1.08)	(20.04)	(20.04)
Ginv	0.0205	0.1545 ***	0.1545 ***
	(0.48)	(5.72)	(5.72)
_cons	−0.5528 ***	4.6116	4.6116
	(−2.74)	(40.70)	(40.70)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
R-Square	0.060	0.223	0.223
Adj. R-Square	0.057	0.221	0.221
N	23,050	23,050	23,050

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

. The regression coefficient of Multi_IV in the first stage is significantly positive at the 1% level, and the F-values exceed the threshold of 10. This allows us to confirm the correlation and exogeneity of Multi_IV. The results from the second-stage estimation reveal a significantly negative coefficient for Multi in both the 2SLS and GMM models. This suggests that MLS-decreased ESG performance remains robust, even after correcting for endogeneity bias.

4.6.4. Firm-Fixed Effect Test

We rerun the regressions by controlling for both time-invariant and firm-specific characteristics, as shown in Column 1 of

Table 10. Results of firm-fixed effect and PSM-OLS.

	(1)	(2)
	Firm-Fixed Effect	PSM-OLS
Multi	−0.0265 ***	−0.0334 ***
	(−2.95)	(−2.67)
Size	0.1015 ***	0.0872 ***
	(13.20)	(10.38)
Roa	0.4610 ***	1.0934 ***
	(6.99)	(8.97)
Lev	−0.4131 ***	−0.4219 ***
	(−13.34)	(−8.77)
Age	−0.0884 ***	−0.0979 ***
	(−6.89)	(−8.47)
Growth	−0.0022	−0.0000
	(−0.62)	(−0.01)
Big4	0.0131	−0.0263
	(0.57)	(−0.87)
Board	0.0783 **	0.1092 **
	(2.26)	(2.36)
Indrt	0.6349 ***	0.7892 ***
	(6.76)	(5.56)
Dual	−0.0101	0.0019
	(−1.14)	(0.13)
HHI	0.0673	0.1657 **
	(1.08)	(2.27)
CSR	0.1420 ***	0.2963 ***
	(11.14)	(16.62)
Ggov	0.0313 ***	0.0621 ***
	(10.29)	(11.84)
Ginv	0.0405	0.1364 ***
	(1.48)	(2.84)
_cons	4.9146 ***	4.8277 ***
	(26.36)	(22.64)
Firm	Yes
Ind		Yes
Year	Yes	Yes
R-Square	0.577	0.296
Adj. R-Square	0.526	0.291
N	25,767	9801

** or *** indicates a significance level at 5% and 1%, respectively.

. The independent variables retain the predicted signs and remain statistically significant at the 1% level.

4.6.5. PSM-OLS Test

We use propensity score matching to adjust for differences in firm characteristics between MLS and SLS firms, helping to reduce selection bias from observable variables. First, we regress Multi with the control variables from Model 1 to estimate the propensity scores for having MLS. These predicted propensity scores are then used to match SLS firms on a 1:1 nearest-neighbor basis.

Column 2 in Table 10 presents the re-estimated results using the matched sample. These results are consistent with our findings shown in Table 5.

4.6.6. Other Robustness Tests

We perform three additional robustness tests. First, following prior research [35], we redefine “large shareholder” by applying a 5% shareholding threshold and present the regression results in Column 1 of

Table 11. Results of other robustness tests.

	(1)	(2)	(3)
	Alternative Measurement of MLS	Lagged Variable One Period	Excluding Absolute Control Sample
Multi_5per	−0.0447 ***
	(−4.57)
Multi		−0.0424 ***	−0.051 ***
		(−3.67)	(−4.15)
Size	0.0877 ***	0.0755 ***	0.0956 ***
	(13.18)	(10.87)	(12.19)
Roa	1.1625 ***	2.0162 ***	1.1426 ***
	(13.57)	(22.09)	(12.01)
Lev	−0.4052 ***	−0.3294 ***	−0.4104 ***
	(−11.53)	(−9.01)	(−10.03)
Age	−0.1022 ***	−0.0789 ***	−0.1055 ***
	(−11.25)	(−8.67)	(−9.52)
Growth	−0.0041	0.0101 **	−0.0017
	(−0.88)	(2.06)	(−0.29)
Big4	−0.0209	−0.0176	−0.0232
	(−0.89)	(−0.72)	(−0.78)
Board	0.1148 ***	0.1011 ***	0.1400 ***
	(3.12)	(2.62)	(3.17)
Indrt	0.8467 ***	0.7677 ***	0.8254 ***
	(7.83)	(6.74)	(6.28)
Dual	−0.0148	−0.0194 *	−0.0173
	(−1.36)	(−1.71)	(−1.38)
HHI	−0.0200	−0.0187	−0.0188
	(−0.37)	(−0.33)	(−0.18)
CSR	0.2908 ***	0.3247 ***	0.2832 ***
	(21.08)	(22.30)	(17.50)
Ggov	0.0585 ***	0.0646 ***	0.0573 ***
	(14.38)	(14.94)	(11.93)
Ginv	0.1565 ***	0.1566 ***	0.1534 ***
	(4.34)	(4.22)	(3.63)
_cons	4.8685 ***	5.0766 ***	4.6895 ***
	(29.97)	(30.16)	(24.13)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
R-Square	0.278	0.312	0.274
Adj. R-Square	0.276	0.310	0.271
N	25,767	23,050	17,852

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

. Second, considering the potential delayed influence of MLS on ESG performance, we lag the main variables by one period and rerun the regression, with the results presented in Column 2 of Table 11. Third, we exclude firms where the CS holds more than 50% of the total equity shares and show the results in Column 3 of Table 11. All these robustness tests consistently support our main findings.

4.7. Summary of Findings against Hypotheses

We confirmed the research hypotheses through regression analysis and robustness tests. To better present these hypotheses and findings, we have summarized them in

Table 12. Summary of findings against hypotheses.

Hypothesis	Content	Findings	Supported
H1	MLS leads to a decrease in ESG performance due to coordination friction	MLS significantly reduces ESG performance due to coordination friction	Supported
H2	CS can mitigate friction among MLS and thus reduce its negative impact on ESG performance	Strong controlling shareholder power mitigates the negative impact of MLS	Supported
H3	State-controlled shareholders more significantly reduce MLS friction	State-controlled shareholders significantly reduce MLS friction and its negative impact on ESG performance	Supported

. Overall, the empirical results support the three hypotheses. Specifically, MLS is associated with lower ESG performance (H1). This negative effect is mitigated by controlling shareholders (H2) and is more significant in SOEs (H3).

5. Mechanism Analysis

5.1. Impact of Friction among MLS

The more shares held by other large shareholders, the more power they can exert on the CS. However, this may cause friction among large shareholders, negatively affecting ESG performance. Chakraborty & Gantchev [48] use shareholder concentration to reflect the level of coordination friction based on shareholder voting rights. Following Fang et al. [21], we use the relative control rights of other large shareholders to represent contestation or friction among large shareholders.

We define five variables to measure the friction of MLS. The first variable, Top2, is the voting right of the second-largest shareholder. The second variable, S2to1, is the ratio of the second-largest shareholder’s shareholding divided by that of the largest shareholder. The third variable, S2345, represents the cumulative shareholding of the second to the fifth largest shareholders. The fourth variable, S2345to1, is the sum of the shareholdings of the second to fifth largest shareholders divided by the largest shareholder. The fifth variable, Num, denotes the number of shareholders whose shareholdings exceed 10%, excluding the largest shareholder. Columns 1–5 of

Table 13. Regression on the friction of multiple large shareholders.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	ESG	ESG	ESG	ESG	ESG	ESG	ESG
Top2	−0.0025 ***
	(−3.12)
S2to1		−0.0926 ***
		(−3.95)
S2345			−0.0024 ***
			(−4.31)
S2345to1				−0.0617 ***
				(−4.39)
Num					−0.0371 ***
					(−4.05)
Homo						−0.0350
						(−1.61)
Family							−0.0028
							(−0.12)
Size	0.0880 ***	0.0879 ***	0.0892 ***	0.0885 ***	0.0878 ***	0.0873 ***	0.0886 ***
	(13.20)	(13.22)	(13.38)	(13.29)	(13.17)	(7.52)	(7.67)
Roa	1.1641 ***	1.1552 ***	1.1697 ***	1.1571 ***	1.1668 ***	0.9890 ***	0.9860 ***
	(13.57)	(13.47)	(13.67)	(13.51)	(13.61)	(6.62)	(6.59)
Lev	−0.4047 ***	−0.4062 ***	−0.4078 ***	−0.4085 ***	−0.4064 ***	−0.4288 ***	−0.4248 ***
	(−11.52)	(−11.58)	(−11.62)	(−11.64)	(−11.57)	(−6.36)	(−6.26)
Age	−0.1002 ***	−0.0983 ***	−0.1040 ***	−0.0996 ***	−0.0997 ***	−0.0882 ***	−0.0859 ***
	(−11.00)	(−10.95)	(−11.28)	(−11.07)	(−11.05)	(−5.37)	(−5.25)
Growth	−0.0042	−0.0042	−0.0039	−0.0039	−0.0039	0.0088	0.0094
	(−0.90)	(−0.90)	(−0.83)	(−0.84)	(−0.84)	(1.03)	(1.09)
Big4	−0.0193	−0.0196	−0.0172	−0.0205	−0.0208	−0.0076	−0.0036
	(−0.81)	(−0.83)	(−0.72)	(−0.87)	(−0.88)	(−0.18)	(−0.09)
Board	0.1145 ***	0.1178 ***	0.1171 ***	0.1189 ***	0.1139 ***	0.1447 **	0.1582 **
	(3.11)	(3.20)	(3.18)	(3.23)	(3.09)	(2.18)	(2.41)
Indrt	0.8511 ***	0.8543 ***	0.8487 ***	0.8542 ***	0.8501 ***	0.8217 ***	0.8298 ***
	(7.86)	(7.89)	(7.86)	(7.90)	(7.85)	(3.97)	(4.00)
Dual	−0.0149	−0.0145	−0.0144	−0.0139	−0.0153	−0.0171	−0.0189
	(−1.36)	(−1.33)	(−1.33)	(−1.28)	(−1.40)	(−0.83)	(−0.92)
HHI	0.0032	−0.0435	−0.0302	−0.0766	0.0000	0.3418 ***	0.3422 ***
	(0.06)	(−0.78)	(−0.55)	(−1.33)	(0.00)	(2.87)	(2.87)
CSR	0.2912 ***	0.2910 ***	0.2912 ***	0.2909 ***	0.2906 ***	0.2929 ***	0.2937 ***
	(21.11)	(21.12)	(21.12)	(21.10)	(21.09)	(11.76)	(11.76)
Ggov	0.0586 ***	0.0585 ***	0.0583 ***	0.0582 ***	0.0584 ***	0.0638 ***	0.0644 ***
	(14.38)	(14.37)	(14.34)	(14.30)	(14.34)	(8.95)	(9.07)
Ginv	0.1547 ***	0.1556 ***	0.1546 ***	0.1557 ***	0.1538 ***	0.1371 **	0.1436 **
	(4.29)	(4.32)	(4.29)	(4.34)	(4.26)	(2.12)	(2.19)
_cons	4.8533 ***	4.9477 ***	4.8473 ***	4.8523 ***	4.8541 ***	4.6836 ***	4.6040 ***
	(29.85)	(30.24)	(29.86)	(29.97)	(29.84)	(15.81)	(15.72)
Ind	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R-Square	0.277	0.277	0.278	0.278	0.277	0.322	0.321
Adj. R-Square	0.275	0.275	0.276	0.276	0.275	0.315	0.314
N	25,767	25,767	25,767	25,767	25,767	6355	6355

** or *** indicates a significance level at 5% and 1%, respectively.

report the regression results based on these measurements of relative power among MLS. The coefficients for these five measurements are all negative and statistically significant at the 1% level. These results strongly support H1, indicating that the potential contestation of control among MLS negatively impacts ESG performance. This suggests that coordination friction among MLS leads to poorer ESG rating scores.

Maury and Pajuste [16] find that the likelihood of collusion between MLS increases when large shareholders are of the same type (state-owned or non-state-owned). Collusion is also more likely if the large shareholders are all family members. To exclude the explanation of MLS collusion on ESG performance, we construct two dummy variables. The first variable, Homo, equals 1 if all large shareholders have the same type of ownership (all state-owned or all privately owned) and 0 otherwise. The second variable, Family, equals 1 if all large shareholders are family members and 0 otherwise. We rerun the regressions and report the results in Columns 6 and 7 of Table 13. The coefficient of Multi is not significant, suggesting no evidence that the collusion of MLS leads to lower ESG rating scores.

5.2. Subcategories of ESG

We further explore how MLS affects the subcategories of ESG, ultimately impacting firms’ overall ESG ratings. Specifically, we run regressions for environmental (E), social (S), and corporate governance (G) scores separately. The results, presented in

Table 14. Regressions on the ESG subcategory.

	(1)	(2)	(3)
	E	S	G
Multi	−0.0116	−0.0491 **	−0.0513 ***
	(−0.68)	(−2.32)	(−3.60)
Size	0.0974 ***	0.1142 ***	0.0638 ***
	(9.56)	(8.94)	(7.52)
Roa	−0.1225	1.3403 ***	1.8217 ***
	(−1.02)	(8.72)	(14.58)
Lev	0.1222 **	0.1501 **	−1.0271 ***
	(2.30)	(2.28)	(−21.40)
Age	−0.0247 *	−0.2377 ***	−0.0506 ***
	(−1.76)	(−13.38)	(−4.60)
Growth	−0.0002	−0.0077	−0.0031
	(−0.03)	(−0.93)	(−0.49)
Big4	−0.0836 **	−0.0808 *	0.0632 **
	(−2.08)	(−1.73)	(2.21)
Board	−0.0176	0.1469 **	0.1679 ***
	(−0.30)	(2.05)	(3.88)
Indrt	−0.1652	0.1027	1.8720 ***
	(−0.94)	(0.49)	(14.68)
Dual	−0.0385 **	0.0231	−0.0234 *
	(−2.21)	(1.12)	(−1.71)
HHI	−0.1503 *	−0.4914 ***	0.3795 ***
	(−1.70)	(−4.72)	(5.66)
CSR	0.3728 ***	0.3101 ***	0.2264 ***
	(16.35)	(11.48)	(12.72)
Ggov	0.1164 ***	0.0645 ***	0.0254 ***
	(17.98)	(8.32)	(4.96)
Ginv	0.6178 ***	0.1361 **	−0.0479
	(10.42)	(2.05)	(−1.03)
_cons	4.0648 ***	3.7993 ***	6.3605 ***
	(16.72)	(11.93)	(29.51)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
R-Square	0.302	0.286	0.255
Adj. R-Square	0.300	0.284	0.253
N	25,767	25,767	25,767

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

, indicate that MLS primarily affects ESG performance through social and corporate governance aspects.

5.3. Behaviors of Executives

Theoretical analyses and empirical results show that MLS causes frictions that lead to lower social and governance scores, ultimately resulting in poor ESG performance. What role do executives play in this process? MLS frictions can weaken the monitoring of executives, allowing them to engage in opportunistic behavior, such as receiving excessive compensation [21]. We argue that MLS frictions lead to executive misconduct, which affects a firm’s social and governance scores and ultimately lowers ESG performance.

To measure executive misconduct, we select two variables. The first variable, STMT, indicates executive team stability. We predict that MLS frictions create uncertainty within the executive team, potentially increasing executive turnover and decreasing stability. This instability can prevent executives from effectively implementing long-term ESG projects, causing them to be myopic and less focused on ESG, thus reducing ESG performance from a governance perspective.

The second variable, FPG, indicates the employee pay gap. We predict that MLS frictions result in inadequate shareholder oversight of executives, leading to increased executive compensation to satisfy their own interests. This widens the pay gap within the firm and reduces fairness, therefore lowering ESG ratings. This affects a company’s ESG performance from a societal perspective.

Table 15. MLS and executive behaviors.

	(1)	(2)
	STMT	FPG
Multi	−0.0155 ***	0.3148 ***
	(−6.47)	(2.91)
Size	0.0012	1.0181 ***
	(0.96)	(13.19)
Roa	0.1490 ***	8.0449 ***
	(7.23)	(12.70)
Lev	−0.0204 ***	0.1622
	(−2.91)	(0.60)
Age	−0.0129 ***	−0.2560 ***
	(−7.50)	(−2.84)
Growth	−0.0029 *	−0.1807 ***
	(−1.78)	(−4.52)
Big4	−0.0027	0.8169 **
	(−0.63)	(2.12)
Board	0.0241 ***	−0.1412
	(3.04)	(−0.37)
Indrt	0.0194	−1.3673
	(0.85)	(−1.26)
Dual	0.0093 ***	0.0042
	(4.03)	(0.05)
HHI	−0.0041	−2.4352 ***
	(−0.41)	(−4.17)
CSR	0.0010	0.1510
	(0.37)	(1.10)
Ggov	0.0010	0.0643 *
	(1.12)	(1.94)
Ginv	0.0066	−0.4678 *
	(0.82)	(−1.82)
_cons	0.7279 ***	−14.8276 ***
	(22.65)	(−8.72)
Ind	Yes	Yes
Year	Yes	Yes
R-Square	0.063	0.213
Adj. R-Square	0.060	0.211
N	25,767	25,767

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

presents the regression results, indicating that firms with MLS exhibit significantly lower executive team stability and a significantly larger employee pay gap. These findings suggest that coordination frictions among MLS reduce the effectiveness of monitoring by large shareholders, therefore exacerbating the agency problem between shareholders and executives, which ultimately leads to poor ESG performance.

6. Further Analysis

6.1. Analysis of the Heterogeneity of MLS

Prior research indicates that foreign and institutional shareholders influence corporate governance and firm policy-making [49,59]. We analyze the impact of institutional and foreign shareholders on ESG performance by examining shareholder heterogeneity in MLS.

The concept of ESG, initially introduced by the United Nations, has continuously evolved on the international stage. Foreign investors generally favor companies that pursue environmental or social responsibility projects. They actively invest in companies with better ESG initiatives [60] or promote a firm’s ESG performance by directly engaging in corporate activities [31]. Therefore, we assume that foreign shareholders can alleviate the friction among MLS. We define a dummy variable, Foreign, which equals 1 if there is a large foreign shareholder in the firm and 0 otherwise. Column 1 of

Table 16. Results of the heterogeneity of MLS analysis.

	(1)	(2)
	ESG	ESG
Multi	−0.0640 ***	−0.0622 ***
	(−5.15)	(−5.12)
Foreign	−0.0722 **
	(−2.12)
Multi × Foreign	0.1437 ***
	(3.62)
Inst		−0.0143
		(−0.29)
Multi × Inst		0.0973 *
		(1.84)
Size	0.0857 ***	0.0851 ***
	(12.86)	(12.69)
Roa	1.1690 ***	1.1697 ***
	(13.65)	(13.65)
Lev	−0.4052 ***	−0.4028 ***
	(−11.56)	(−11.47)
Age	−0.0984 ***	−0.0988 ***
	(−10.91)	(−10.97)
Growth	−0.0038	−0.0038
	(−0.81)	(−0.81)
Big4	−0.0343	−0.0350
	(−1.45)	(−1.49)
Board	0.1160 ***	0.1146 ***
	(3.17)	(3.10)
Indrt	0.8506 ***	0.8425 ***
	(7.87)	(7.79)
Dual	−0.0158	−0.0158
	(−1.45)	(−1.45)
HHI	0.0081	0.0003
	(0.15)	(0.00)
CSR	0.2891 ***	0.2898 ***
	(20.95)	(21.04)
Ggov	0.0580 ***	0.0580 ***
	(14.29)	(14.25)
Ginv	0.1542 ***	0.1544 ***
	(4.31)	(4.30)
_cons	4.8901 ***	4.9120 ***
	(30.08)	(30.08)
Ind	Yes	Yes
Year	Yes	Yes
R-Square	0.278	0.278
Adj. R-Square	0.276	0.276
N	25,767	25,767

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

shows that foreign blockholders alleviate the friction of MLS and improve ESG performance, as demonstrated by the significantly positive coefficient of Multi × Foreign.

Many studies argue that institutional investors positively impact corporate governance [6,31]. More institutional investors are adopting ESG concepts in their investment activities [4,54,57]. We assume that institutional investors mitigate the negative effect of MLS on ESG performance if they are large shareholders. We define a dummy varible, Inst, which equals 1 if there is an institutional large shareholder in the firm, and 0 otherwise. Column 2 of Table 16 shows that institutional blockholders alleviate the negative impact of MLS on ESG performance, as the coefficient on Multi × Inst is significantly positive.

6.2. Cross-Sectional Analysis

Furthermore, we explore MLS and ESG performance considering a firm’s environmental risk, analyst attention, and involvement in the “Stock Connect Scheme”.

First, we classify the full sample into polluting and clean industry subgroups and separately run regressions. According to the 2012 CSRC Industry Classification, we define companies in the 19 industries of B07, B08, B09, B10, C20, C22, C25, C26, C28, C30, C31, C32, C33, C43, D44, D45, E47, E48 and E50 as high-pollution enterprises [61], while others are classified as low-pollution enterprises. We find that MLS has a greater negative effect on ESG performance in clean industries. One possible explanation is that ESG-related regulations are more stringent in polluting industries, prompting large shareholders in these sectors to engage more in ESG activities to mitigate non-compliance risks and costs.

Second, existing research suggests that analysts serve as information intermediaries, supervisors of governance, and facilitators of price discovery. Analysts utilize their information advantages and professional expertise to evaluate companies, and their reports can expedite price adjustments. We posit that in scenarios with low analyst attention, the market information environment is poorer. Therefore, we measure analyst attention by the number of analysts following a company, categorizing samples above the median as the “high-attention” group and the others as the “low-attention” group. The regression results show that MLS has a stronger negative impact on low-attention companies. This indicates that in poor market information environments, MLS is more likely to lead to lower ESG ratings.

Finally, we classify the sample based on whether the companies are involved in the “Stock Connect Scheme”. The Chinese stock markets (Shanghai and Shenzhen Stock Exchanges) were initially designed for domestic investors, with limited participation from outside the mainland through arrangements such as Qualified Foreign Institutional Investors (QFII). The Hong Kong Stock Exchange has long been an internationalized market. In 2014 and 2016, the Chinese government announced the “Stock Connect Scheme”, which connects the mainland (Shanghai and Shenzhen) and Hong Kong stock markets. This allows institutional and individual investors in both regions to trade listed stocks qualified by market regulators, increasing market liquidity and trading volume. We find that the negative effect of MLS on ESG performance is more significant for firms not participating in the scheme. A possible reason is that the participants in the Hong Kong stock market are predominantly international investors. This finding aligns with the shareholder heterogeneity analysis, further confirming the positive role of foreign investors in promoting ESG performance (

Table 17. Results of other heterogeneity analyses.

	(1)	(2)	(3)	(4)	(5)	(6)
	Polluting Industry	Clean Industry	High Attention	Low Attention	Stock Connect Scheme	Not Stock Connect Scheme
Multi	−0.0264	−0.0448 ***	−0.0154	−0.0595 ***	−0.0026	−0.0588 ***
	(−1.01)	(−3.67)	(−1.08)	(−4.18)	(−0.13)	(−4.79)
Size	0.0918 ***	0.0870 ***	0.0711 ***	0.0725 ***	0.0896 ***	0.0825 ***
	(6.42)	(11.64)	(7.78)	(7.37)	(6.94)	(10.48)
Roa	0.8046 ***	1.2722 ***	0.6552 ***	1.2085 ***	0.8917 ***	1.2548 ***
	(3.73)	(13.68)	(5.09)	(11.22)	(5.13)	(13.35)
Lev	−0.5296 ***	−0.3766 ***	−0.4055 ***	−0.3876 ***	−0.3989 ***	−0.3975 ***
	(−7.23)	(−9.49)	(−8.42)	(−8.81)	(−5.37)	(−10.48)
Age	−0.1303 ***	−0.0885 ***	−0.0643 ***	−0.1121 ***	−0.0668 ***	−0.1063 ***
	(−6.48)	(−8.94)	(−5.12)	(−9.71)	(−3.64)	(−10.98)
Growth	−0.0092	−0.0036	0.0009	−0.0067	−0.0043	−0.0043
	(−0.74)	(−0.72)	(0.13)	(−1.18)	(−0.36)	(−0.86)
Big4	0.0014	−0.0261	−0.0286	−0.0082	0.0028	−0.0542 **
	(0.03)	(−0.96)	(−1.02)	(−0.23)	(0.08)	(−2.06)
Board	0.2106 **	0.0862 **	0.0996 **	0.1268 **	0.1666 **	0.0912 **
	(2.54)	(2.10)	(2.21)	(2.47)	(2.47)	(2.19)
Indrt	1.0533 ***	0.7983 ***	0.8443 ***	0.8657 ***	1.0920 ***	0.7163 ***
	(4.29)	(6.66)	(6.33)	(5.67)	(5.84)	(5.71)
Dual	−0.0249	−0.0143	−0.0236 *	−0.0137	−0.0255	−0.0146
	(−0.97)	(−1.20)	(−1.66)	(−0.97)	(−1.17)	(−1.24)
HHI	0.0461	0.0019	0.0670	−0.0112	0.2330 **	−0.0728
	(0.38)	(0.03)	(0.95)	(−0.16)	(2.34)	(−1.26)
CSR	0.3597 ***	0.2687 ***	0.2874 ***	0.2906 ***	0.2755 ***	0.2895 ***
	(12.48)	(17.19)	(16.74)	(15.94)	(11.56)	(18.62)
Ggov	0.0371 ***	0.0666 ***	0.0544 ***	0.0608 ***	0.0613 ***	0.0563 ***
	(4.56)	(14.30)	(10.67)	(10.87)	(9.17)	(11.80)
Ginv	0.0108	0.1995 ***	0.1232 ***	0.1802 ***	0.1629 **	0.1565 ***
	(0.17)	(4.71)	(2.58)	(3.94)	(2.25)	(3.90)
_cons	4.4601 ***	4.9206 ***	5.2052 ***	5.1864 ***	4.4683 ***	5.0988 ***
	(12.08)	(27.44)	(22.75)	(22.84)	(14.63)	(26.77)
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
R-Square	0.271	0.287	0.272	0.237	0.293	0.258
Adj. R-Square	0.267	0.285	0.268	0.233	0.285	0.256
N	5657	20,110	12,009	13,758	5886	19,881

*, ** or *** indicates a significance level at 10%, 5% and 1%, respectively.

).

7. Conclusions

Using a sample of Chinese listed companies over the 2011–2022 period, we investigate the relationship between MLS and ESG performance. After controlling for possible endogeneity issues, we obtain the following findings. First, MLS significantly reduces ESG performance due to conflicts of interest and coordination friction. Second, CS mitigates the negative impact of MLS on ESG, providing new evidence to support the positive role of CS in corporate governance. This impact is more significant when the CS is state-owned. Third, a heterogeneity analysis of MLS reveals that poor ESG performance can be improved when foreign or institutional investors are among MLS. Fourth, cross-sectional analysis reveals that the negative effect of MLS on ESG is more significant in firms operating in clean industries, with low analyst attention, or not participating in the “Stock Connect Scheme”.

The policy implications of our study are as follows: First, firms should address the decline in decision-making efficiency and management monitoring caused by shareholder friction and enhance cooperation and coordination among shareholders to improve social responsibility performance. Although multiple large shareholders can provide supervision and balances of power, they should also collaborate and mitigate their weaknesses. Second, when external monitoring is insufficient, moderate ownership concentration can be a rational choice for firms, particularly in improving ESG performance. While over-concentration of ownership may lead to Type II agency problems, moderate concentration can facilitate efficient decision-making and promote corporate social responsibility. Third, the capital market should improve external governance mechanisms. This can be achieved by utilizing the supervisory role of information intermediaries such as foreign investors, institutional investors, and professional analysts. Fourth, to maximize stakeholder value, companies should not rely solely on the spontaneous growth of stakeholders but should establish a network of relationships that includes both shareholders and other stakeholders to enhance their decision-making system.

This paper offers new insights into the adverse effects of MLS on corporate ESG performance, providing valuable guidance for firms and policymakers on enhancing corporate ESG outcomes.