Knowledge-Based Faultlines and Corporate Social Irresponsibility: Evidence from Chinese High-Polluting Companies

Abstract

Government requests and societal expectations have pressured high-polluting companies to focus on corporate social responsibility strategies. Using the upper echelons theory as a theoretical framework, we investigated how top management team (TMT) faultlines influence corporate social performance (CSP) based on data from 212 high-polluting companies. The results showed that CSP can be improved by reducing corporate social irresponsibility (CSiR), knowledge-based faultlines have a U-shaped effect on CSiR, and there is a knowledge-based faultline critical point. This implies that knowledge-based faultlines can improve CSiR before reaching this critical point. Additionally, medium-strength knowledge-based faultlines are more conducive to improving irresponsible behavior. CEO power plays a significant moderating role in the relationship between TMT faultlines and CSiR and slows the U-shaped effect of knowledge-based faultlines on CSiR. These findings could help enterprises optimize team structures, adjust corporate social responsibility strategies, and maintain sustainable development in high-polluting sectors.

1. Introduction

The current discussions on corporate social performance (CSP) (this research defines corporate social performance (CSP) as including both corporate responsible behavior (CSR) and corporate irresponsible behavior (CSiR)). focus on “doing good”. The main idea behind the study of CSP is to encourage enterprises to be more socially responsible, both in terms of what social responsibilities they should assume and how they should assume them. By “doing good”, a company can demonstrate its kindness and generosity. However, as the saying goes, “bad news travels fast”, which truly reflects the fact that “doing bad” can cause strong reactions among the public. Ocasio’s [1] study concluded that negative corporate information is more interesting to the public, who are not only the recipients of negative information but also become the disseminators of negative information, which in turn creates a great destructive force regarding the corporate image and reputation. Corporate “doing bad” can be translated as a lack of corporate social responsibility and corporate irresponsible behavior (CSiR). Armstrong [2] was one of the scholars to first focus on the issue of CSiR, and he noted that it is difficult to define CSR clearly, whereas it is much easier to define CSiR. He defines CSiR as a corporate action for individual benefit at the expense of overall benefits to society. However, research related to CSiR has greatly attracted academic attention in the thirty years since the concept was introduced [3]. Lange and Washburn [4] define CSiR as the antithesis of responsible behavior and as “acting in an irresponsible manner”. CSiR refers to corporate behaviors that cause obvious losses or harm. Lin-Hi and Müller [5] categorized CSiR behavior into intentional and unintentional behavior. According to them, CSiR can be defined as “corporate actions that result in (potential) disadvantages and/or harm to other actors”. Windsor [6] views CSR and CSiR as two ends of a continuum where managers can increase responsible behavior by reducing corporate irresponsibility. Keigg et al. [7], Yuan et al. [8], and Fu et al. [9] define CSR and CSiR as two different, intersecting concepts with different causes and consequences that can be present in a single firm at the same time. We define CSiR as follows: CSiR is an illegal, or legal but unethical, behavior that always manifests itself, ignoring the overall interests of society, for the benefit of a single stakeholder without considering the interests of other stakeholders. Typical corporate socially irresponsible behaviors include, but are not limited to, polluting the environment, treating employees and suppliers unfairly, engaging in bribery, using sweatshops, and selling poor-quality products to consumers [10,11].

As CSR and CSiR are gradually being differentiated, scholars have focused on the possible economic consequences of CSiR [12,13], and on the other hand, they have explored the factors that influence CSiR. Previous studies on the factors affecting CSiR have been conducted in the following three areas: the external environment level [14], the firm level [15], and the intra-enterprise member level [16,17]. Our focus on the influence of members within the firm on irresponsible behavior is due to the fact that regardless of how a company ultimately fulfills its social responsibility, it is ultimately decided by executives. Palmer and Ma-her [18] suggest that executives, as the corporate strategic decision-makers, have the most direct impact on CSiR. Since the proposal of upper echelons theory, scholars have verified that individual executive as well as top management team (TMT) characteristics have a direct impact on corporate decision-making and corporate behavior. Studies based on the impact of individual managers on irresponsible corporate behavior have found that CEOs’ needs for personal and social power, competence, tenure, and humility [8,19,20,21] are important factors that influence CSiR. Fu et al. [9] found that having a chief sustainability officer (CSO) position in a firm can reduce CSiR. Compared to individual executives, members of the TMT are more involved in corporate strategy development through information exchange, knowledge sharing, and synergy. The TMT are the developers and implementers of corporate strategy; they have a great deal of control over the firm, execute operational decisions, and ultimately bear the most significant responsibility for the organization’s output [22,23,24]. The impact of demographic characteristics of the TMT, such as gender, education, overseas background, and executive experience, on corporate irresponsible behavior has been studied [25,26]. Although some results have been presented on the impact of TMT characteristics on CSiR, most studies focus on a single characteristic and explore the correlation between the two at a shallow level. The executive team as a whole and its structural characteristics also affect CSiR, and how the characteristics of the executive team affect irresponsible behavior needs to be further explored.

According to Zadek [27], CSP is a continuous learning and improvement process in which managers need to continuously understand the needs of the company and its stakeholders and then make sound decisions that correspond to these needs. We focus on learning-related characteristics such as education, functional experience, etc. However, studying the impact of these variables on CSiR in isolation does not adequately capture the fact that executives are an aggregation of various characteristics. Top management teams (TMTs) are not always a perfect whole, and conflicts exist between executive teams’ subgroups. TMT faultlines provide a research basis for studying the multidimensional characteristics of management teams, which takes into account not only the multiplicity of individual characteristics and their clustering but is also more concerned with the arrangement of one or more attribute characteristics of group members [28]. This greatly compensates for the shortcomings of studies using single characteristics. A TMT knowledge-based faultline is a top management team structure in which sub-groups are divided based on the combination of knowledge characteristics such as experience and expertise. Additionally, it reflects not only the characteristics of knowledge integration but also the uniqueness and dependencies among executive team members. Studying the impact of knowledge-based faultlines on CSiR would present more useful results than studying the effect of a single characteristic of executives on TMT building.

In previous studies, we found that power distance, CEO myopia, CEO overconfidence, etc., act as moderating variables to influence TMT characteristics and CSiR [29,30]. It can be seen that the CEO plays an important role in influencing the relationship between TMT and the behavior of the firm. However, the question of whether CEOs can influence the TMT faultlines and hence influence TMT decision-making has not been addressed in previous research. As a “good steward” of the company, the CEO’s goals are similar to those of the company’s executives, and it is easier for them to regulate the relationship between the executive team members [31]. CEO power is the ability of the CEO to use their authority and personal influence to deal with uncertainties and influence the formulation and implementation of business decisions [32]. Contrary to the traditional principal-agent theory, the management power theory posits that the CEO will use their power to benefit the business. A powerful CEO can play a bridging role to facilitate communication among TMT members and promote the integration of subgroup knowledge, which may affect the original relationship between knowledge-based faultlines and CSiR. Therefore, this study introduces CEO power as a moderating variable, examining the impact of CEO power on the relationship between faultlines and CSiR.

At present, there are few academic studies on the irresponsible behavior of high-polluting companies, mainly because some scholars believe that these companies should pay more attention to their own survival and development. However, in the era of the internet, every move of an enterprise will be quickly reflected to the public, and the negative impacts brought by CSiR will jeopardize the survival of the enterprise at any time. It is all the more important for high-polluting companies to curb and improve irresponsible corporate behavior. This requires executive teams to rely on their knowledge-related characteristics to influence choices on issues of CSiR. Therefore, how TMT knowledge-related faultlines affect the irresponsible behavior of high-polluting companies is an interesting research avenue. This research examines the relationship between TMT knowledge-based faultlines and CSiR and introduces CEO power as a moderating variable, which not only helps refine the study of the internal structure and decision-making operation process of TMTs but also expands the study of CSP based on upper echelons theory. There is a positive significance to studying the impact of knowledge-related faultlines on CSiR for the formation of executive teams, the timely detection and improvement of irresponsible behaviors in high-polluting enterprises, and the promotion of sustainable corporate development.

2. Theoretical and Hypotheses

2.1. Upper Echelons Theory

The upper echelons theory was proposed by Hambrick and Mason in 1984 and published in AMR. This theory states that because of managers’ limited rationality and insufficient information for decision-making, their own experience, characteristics, mind, and state all influence their decisions [33]. That is, compared to objective data, a firm’s strategic decisions are largely influenced by the personal characteristics of the executives throughout the strategic decision-making process.

The upper echelons theory is essentially an information-processing theory that attempts to explain and predict corporate behavior in terms of the characteristics (both demographic and psychological) of executives. CSP, as a specific strategic decision for enterprises, is bound to be influenced by the relevant characteristics of the executives [34]. We use upper echelons theory to infer the impact of knowledge-based faultlines on CSiR.

2.2. Knowledge-Based Faultlines and CSiR

Faultlines are demarcation lines that divide teams into homogeneous subgroups based on team member characteristics [35]. The faultline study is an extension of TMT diversity studies. While diversity studies typically consider only the composition of a particular characteristic of team members (such as gender diversity), faultline studies examine the consequences of the interaction of multiple characteristics of team members [36,37]. If team members are aligned on multiple attributes, they are more likely to be attracted to each other and thus form homogeneous subgroups. TMT faultlines can reflect the structural characteristics within executive teams more comprehensively.

Most studies of TMT faultlines divide them into task-based faultlines and relationship-based faultlines [38,39]. Task-based faultlines are usually TMT faultlines that are based on task-based characteristics (e.g., functional experiences, educational background, and tenure). Relationship-based faultlines are usually TMT faultlines based on relational characteristics (e.g., gender and race). Compared with task-based faultlines, relationship-based faultlines reflect the “stereotypes” that people form at the beginning of contact due to characteristics [40], whereas task-based faultlines are associated with knowledge transfer between team members [41,42]. TMT knowledge-based faultlines based on the features included in task faultlines consider other knowledge-related experiences, such as overseas backgrounds, to be taken into account. We define faultlines with other knowledge-related faultlines as knowledge faultlines, which inherit the characteristics of task faultlines and flexibly reflect the characteristics of the enterprise to be studied. TMT knowledge-based faultlines reflect the quantity and quality of knowledge exchanges between teams, which affect the efficiency and effectiveness of firms in improving irresponsible behavior [43].

It has been shown that there is a significant effect of TMT faultlines on CSP, but this effect varies in different research findings. There is a negative impact of TMT faultlines on CSP as they exacerbate internal conflicts and lead to a low level of information exchange [42,44]. However, a few studies have drawn positive conclusions that TMT faultlines are effective for avoiding the emergence of groupthink and can stimulate team creativity and motivation. These factors will prevent irresponsible behavior from accounting for irregularities [45]. With regard to the depth of the faultlines, scholars generally agree that there is a nonlinear relationship between the extent of the faultlines and their effects; it could be U-shaped or an inverted U-shape [46,47]. Liu et al. [48] took educational background and occupational background as the basis for dividing a faultline and found that the strength of this faultline showed an inverted U-shaped relationship with the level of technological innovation of the enterprise. Only a moderate degree of faultlines can better balance the diversity and synergy of a team so that the enterprise can communicate, learn, and collaborate well on the basis of abundant resources and maximize the positive efficacy of the team [49].

The relationship between knowledge-based faultlines and CSiR may not be a simple linear relationship. When there are no obvious sub-groups in TMTs, the diverse viewpoints provided by knowledge-based faultlines are relatively limited, the work opinions of team members tend to converge, and the task conflicts and roles of the overall team are not well manifested. At this point, decision-making is based on the original experience of the enterprise, and environmental changes cannot be identified quickly and resource allocation cannot be adjusted in a timely and accurate manner. Firms will then either maintain their original corporate social responsibility strategy or make minor improvements to their irresponsible behavior.

With the gradual formation of various sub-groups within the executive team, when the faultlines reach a certain level, the task conflict between the sub-groups will become more clear, and team members will actively participate in information sharing and discussions on heterogeneous viewpoints. The executive team at this point is capable of processing a range of information from the market environment, coordinating scattered activities, recognizing market adaptation needs, and allocating resources reasonably [50,51]. Due to the support and affirmation from the sub-groups to which they belong, team members can express their opinions freely. The overlapping team members act as a “communication bridge” between the sub-groups, ensuring the heterogeneous knowledge viewpoints can be fully shared and integrated and the role of knowledge-based faultlines can be reflected to the greatest extent [52]. At this time, TMT members can conduct a detailed analysis of the irresponsible behavior of the company and find the best solution to actively improve this irresponsible behavior.

However, as the knowledge-based faultlines become stronger, it becomes more difficult to build an information-sharing platform. At this time, although there is a large amount of heterogeneous information and knowledge in the team, it is difficult to achieve high-quality information and knowledge exchange. On the one hand, stronger knowledge-based faultlines imply higher homogenization within sub-groups and more frequent internal information exchange; higher communication barriers between subgroups reduce the frequency of information exchange, which is not conducive to obtaining the required accurate knowledge in a timely manner [53]. On the other hand, the stronger the knowledge-based faultlines of the team, the more it undermines cohesion within the team and the more difficult it is for the top management team members to achieve integration in their mental models, which affects team learning as well as the integration of perspectives in team decision-making [36,54]. The problem of executive “grouping” caused by the faultlines will change the decision-making power of the executive team from evenly dispersed to centralized and antagonistic, which will increase the motivation and opportunity for executives to seize personal benefits and infringe upon the interests of some stakeholders [44].

H1. 

There is a U-shaped relationship between knowledge-based faultlines and CSiR.

2.3. The Moderating Role of CEO Power

As the central manager of the company’s production and strategic decisions, the CEO has the most direct responsibility for the company’s business activities and plays an important role in information processing in the management team. Finkelstein [32] believes that the main responsibility of the CEO is to avoid the risks encountered in the company’s business management activities. Most of the risks within the enterprise come from the TMT, while the risks outside the enterprise come from the economic market and macropolicy environment. Internal and external risks to the company lead to the CEO holding more power. CEO power is the ability of the CEO to use their authority to handle matters and influence business management decisions. Based on the responsibilities of the CEO, general managers, presidents, and chief executive officers are all classified as CEOs in this study.

Finkelstein [32] classifies the power held by the CEO as CEO power and divides it into four dimensions: structural power, ownership power, expert power, and prestige power. Structural power is determined by the company’s organizational structure. The CEO, as the core of the company, is located at the top of the corporate governance structure, and the power that comes with this positional structure enables the CEO to deal with the uncertainties faced by the company. Ownership power is the equity owned by the CEO. When the CEO is the founder or holds stock, the CEO influences the choices of the board of directors and is more capable of controlling the direction of the business. Expert power refers to the expertise of the CEO, and Hambrick and Mason [33] argue that expert power is the power of the CEO to effectively manage the firm in an uncertain environment. When a CEO has held a title in a particular field, it indicates that they have the expertise and technical ability to deal with uncertainty in that area. Prestige power refers to a reputable CEO who can establish relationships between enterprises and obtain more resources. They can coordinate the relationship with other firms and reduce the external impact on the enterprise. When the CEO works part time in other companies, they can establish close relationships with the personnel of those companies, which can not only increase the breadth of the CEO’s own access to information and experience but can also enhance the social network of the company. The above four dimensions reflect both the CEO’s influence and his expertise.

One of the CEO’s responsibilities is to lead the board’s strategic decisions to reduce internal and external conflicts, and they are expected to facilitate communication among members of the executive team and integrate different perspectives. In contrast to the traditional view of principal-agent theory, management power theory assumes that the CEO will use their power to benefit the firm. The CEO uses their influence and appeal to have more control over the resources of the company and can reasonably control and deal with conflicts within the team to reduce internal conflicts. In the pursuit of their own dignity, beliefs, and intrinsic job satisfaction, CEOs are motivated to be “good stewards” of their companies [31,55]. CEO power slows down the rate at which knowledge-based faultlines reach the critical point, allowing them to better function as heterogeneous knowledge. A powerful CEO effectively disciplines other team members in the executive team, adequately moderating relational and process conflicts between opposing subgroups and avoiding further increases in the intensity of faultlines [56].

H2. 

CEO power moderates the relationship between knowledge-based faultlines and CSiR, and CEO power slows the U-shaped effect of knowledge-based faultlines on CSiR.

3. Measures

3.1. Sample and Data Collection

For this study, high-polluting companies listed in Chinese A-shares from 2015 to 2019 were selected as a research sample to investigate how TMT faultlines affect CSP. In 2014, China explicitly proposed strengthening CSP legislation and requiring enterprises to put social responsibility into practice. As a result of the lag in the implementation of the policy, in this study, the starting year for sample collection was set as 2015. We focus on five consecutive years of data because the members of the executive team remain relatively stable over a five-year period, which is more favorable for conducting a faultline study. We chose high-polluting firms as our study sample because high-polluting companies have a greater impact on the environment, are subject to stronger regulations and constraints, and are obliged to disclose corporate social responsibility behaviors.

Companies were selected as follows: (1) companies with risk warnings on their stocks were removed; (2) companies in the financial sector category were excluded; (3) companies with executive teams containing less than four members were removed; (4) companies with changes to more than one-third of their executive teams were removed. A total of 1060 observations were finally obtained from 212 high-polluting companies over the course of 5 years. We filled in the missing data by searching for annual reports and relevant web reports. CSP scores were obtained via reports from Hexun.com. In this study, R was used to calculate knowledge-related faultlines, and sample data were empirically analyzed using Stata 15.1.

3.2. Variables

3.2.1. Dependent Variable: CSiR

Most studies use the data exposed by KLD to measure CSP. The KLD database analyzes the responsible and irresponsible behaviors of the enterprises in seven dimensions, all indicator items of “strength” are considered CSR, and those of “concern” are considered CSiR. China measures CSP via their own metrics; the Hexun CSP Score is one of them. Hexun has fewer missing data points and detailed scores for each measured dimension, which show how well a given social responsibility is being fulfilled and which social responsibility is not being fulfilled. Our study, which measures irresponsible corporate behavior, uses Hexun’s data. Hexun categorizes CSP into five dimensions: shareholder responsibility, employee responsibility, supplier responsibility, customer and consumer rights and interests responsibility, environmental responsibility, and social responsibility. Additionally, under each dimension, there are secondary and tertiary indicators to provide a comprehensive evaluation of social responsibility. The third level of indicators corresponds to the score of each detailed CSP indicator. At this level, when the score is greater than 0, it is recorded as a responsible behavior; when the score is less than 0, it is recorded as an irresponsible behavior; and when the score is 0, a judgment is made according to whether or not the enterprise discloses the indicator; if it does not disclose it, it is not counted as an irresponsible behavior; and if it should disclose it but the score is 0, it is recorded as an irresponsible behavior. We use the calculation method of Di Giuli and Kostovetsky [57], where the number of irresponsible behaviors in the third level of indicators is summed to get the final result, to obtain the CSiR. This study also refers to the CSMAR database for corporate violation data to ensure the accuracy of the 0 value in the CSiR.

3.2.2. Independent Variable: Knowledge-Based Faultlines

In the study, the Fau measure developed by Thatcher and colleagues [58] is used to calculate knowledge-based faultlines. Knowledge-based faultlines have four types of characteristics: academic qualifications, tenure, overseas experience, and functional backgrounds. These characteristics represent the knowledge composition of the members: education is the basic knowledge that managers have from past professions; tenure is the knowledge that managers have acquired from growing with the company, including knowledge of corporate preferences; overseas experience is the knowledge that managers have acquired from overseas; and functional background is the knowledge that managers possess related to their functions. Scholarly qualifications are divided into five categories, with values 1 through 5, ranging from high school and lower up to PhD level. Managers with overseas experience are recorded as 1, while those without are recorded as 0. Functional backgrounds are divided into three categories: production functional backgrounds are assigned as 1 and include production, management, and accounting; output functional backgrounds are assigned as 2 and include marketing, sales, and R&D; and peripheral functional backgrounds are assigned as 3 and include legal affairs, investment, and human resources.

The calculation formula is as follows:

$${\mathrm{F}\mathrm{a}\mathrm{u}}_{\mathrm{g}}=(\frac{{\sum }_{\mathrm{j}=1}^{\mathrm{p}}{\sum }_{\mathrm{k}=1}^2{\mathrm{n}}_{\mathrm{k}}^{\mathrm{g}}{({\overline{\mathrm{X}}}_{\mathrm{j}\mathrm{k}}-{\overline{\mathrm{X}}}_{\mathrm{j}})}^2}{{\sum }_{\mathrm{j}=1}^{\mathrm{p}}{\sum }_{\mathrm{k}=1}^2{\sum }_{\mathrm{i}=1}^{{\mathrm{n}}_{\mathrm{k}}^{\mathrm{g}}}{({\overline{\mathrm{X}}}_{\mathrm{i}\mathrm{j}\mathrm{k}}-{\overline{\mathrm{X}}}_{\mathrm{j}})}^2})\hspace{1em}\mathrm{g}=1,2,3\dots \dots \mathrm{s}$$

where n represents the number of TMT members, p represents the total number of features examined, and g represents the categorization. The Fau_g value is measured using R version 3.6.2 software. The value range of Fau_g is [0, 1]. The larger the value of Fau_g, the greater the strength of the faultlines.

3.2.3. Moderator Variables: CEO Power

The literature divides CEO power into structural power, ownership power, expert power, and prestige power. Structural power is the management power conferred by the organization [32], which is measured by whether CEOs have duality. If a CEO had duality, we recorded it as 1; otherwise, it was 0. Ownership power is whether a CEO owns equity; a value of 1 is assigned if they do, otherwise it is 0. Expert power refers to CEOs’ professional skills. CEOs who have professional titles are assigned 1, and otherwise 0. A reputable CEO builds connections between companies. Reputation rights are measured by whether a CEO works part time in other firms, which we mark as 1 if they do and 0 if they do not. The average of the four powers is CEO power.

3.2.4. Control Variables

In this research, we refer to work by Deng et al. [59] and Georgakakis [54] to categorize the control variables into firm control variables and team control variables. Firm control variables involve the following variables: firm age, which is the time from the establishment of the enterprise to the time of our measurement and is taken as a logarithm; firm performance, which is measured by ROA; and a firm’s gearing ratio, which is used to measure a firm’s level of risk. We assign a value of 1 to state-owned firms and 0 to non-state-owned firms depending on the nature of the ownership.

Team control variables involve the following variables: TMT size is calculated by the number of managers, which may affect the strength of TMT faultlines. Executives with political connections are more sensitive to environmental policies, which is measured by the number of TMT members with government experience. We also selected TMT age, TMT degree, and TMT gender as control variables. TMT age is the logarithm of the average age of team members; TMT degree is the average education of team members; and TMT gender is the ratio of female managers to the total number of team members. Similarly, we also account for CEO characteristics such as CEO age, CEO degree, and CEO tenure.

4. Results

4.1. Descriptive Statistics and Correlations

Table 1. CSiR descriptive statistics and correlations.

	Mean	S.D	1	2	3	4	5	6	7	8	9	10	11	12	13	14
CSiR	17.694	7.030	1
TMT Faultline	0.611	0.192	0.050	1
Firm age	1.985	0.827	−0071 **	0.066 **	1
performance	0.049	0.075	−0.188 ***	−0.026	−0.119 ***	1
Lev	0.388	0.188	0.026	0.008	0.264 ***	−0.236 ***	1
Ownership	0.349	0.478	−0.051 *	0.036	0.535 ***	−0.100 ***	0.193 ***	1
TMT size	6.974	2.176	−0.208 ***	−0.295 ***	0.158 ***	−0.014	0.137 ***	0.056 *	1
TMT political connection	0.629	0.872	−0.054 *	−0.012	−0.047	−0.006	0.065 **	−0.161 ***	0.138 ***	1
TMT age	1.674	0.045	−0.104 ***	0.050	0.263 ***	−0.003	0.103 ***	0.196 ***	0.063 **	−0.020	1
TMT degree	3.223	0.589	−0.174 ***	−0.035	0.197 ***	0.039	0.043	0.114 ***	−0.003	−0.013	0.353 ***	1
TMTgender	0.149	0.153	0.020	−0.033	−0.182 ***	0.087 ***	−0.216 ***	−0.149 ***	−0.122 ***	0.019	−0.141 ***	−0.076 **	1
CEO age	1.691	0.058	−0.046	−0.033	0.262 ***	−0.079 **	0.119 ***	0.041	0.084 ***	0.031	0.113 ***	0.032	−0.130 ***	1
CEO degree	3.420	0.851	−0.172 ***	−0.079 ***	0.137 ***	0.007	0.108 ***	0.046	0.200 ***	−0.015	0.022	0.326 ***	−0.095 ***	−0.083 **	1
CEO tenure	1.774	0.296	0.033	−0.040	0.083 ***	0.024	0.026	−0.053 *	−0.040	0.022	0.017	0.032	0.022	0.141 ***	0.072 **	1

***, **, and * indicate significance at 1%, 5%, and 10% levels.

shows the descriptive statistics of the variables and the correlation between them. The VIF for the entire regression model was 1.79, and since it is less than 10, it can be assumed that there is no serious multicollinearity problem.

The mean value of CSiR is 17.794 and the standard deviation is 7.030, which indicates the existence of irresponsibility and even seriousness among the companies. The mean value of TMT faultlines is 0.611, which indicates that there are relatively significant knowledge-based faultlines in the selected companies. As can be seen from the correlation, there is no significant linear relationship between knowledge-related faultlines and CSiR. In the correlation analysis, we also found that the control variables of firm age, performance, executives with political connections, TMT age, and TMT degree negatively affect CSiR. As companies become more established, their focus shifts towards improving CSiR; firms that perform well and firms with government-connected executive teams are more focused on alleviating CSiR, whereas the older the average age and the higher the average education, the greater the concern for improving irresponsible behavior. The correlation only considers a two-by-two relationship between the variables. It is also necessary to include control variables and run regressions to obtain rigorous conclusions.

4.2. Regression Results

Model 2 in

Table 2. Regression: TMT faultlines on CSiR.

CSiR	Model 1	Model 2	Model 3	Model 4
Firm age	−0.083 * (0.048)	−0.081 * (0.048)	−0.078 (0.048)	−0.086 * (0.050)
performance	−1.955 *** (0.403)	−1.957 *** (0.403)	−1.948 *** (0.403)	−1.964 *** (0.403)
Lev	0.496 *** (0.162)	0.496 *** (0.162)	0.490 *** (0.162)	0.510 *** (0.163)
Ownership	−0.055 (0.074)	−0.056 (0.074)	−0.070 (0.0740	−0.066 (0.074)
TMT size	−0.086 *** (0.014)	−0.087 *** (0.014)	−0.089 *** (0.014)	−0.092 *** (0.015)
TMT political connection	−0.045 (0.033)	−0.045 (0.033)	−0.051 (0.033)	−0.050 (0.034)
TMT age	−1.709 ** (0.711)	−1.693 ** (0.713)	−1.607 ** (0.714)	−1.595 ** (0.714)
TMT degree	−0.188 *** (0.055)	−0.189 *** (0.055)	−0.199 *** (0.056)	−0.203 *** (0.056)
TMTgender	−0.135 (0.193)	−0.138 (0.193)	−0.134 (0.193)	−0.138 (0.194)
CEO age	−0.906 * (0.531)	−0.913 * (0.532)	−0.844 (0.533)	−0.746 (0.552)
CEO degree	−0.118 *** (0.037)	−0.118 *** (0.037)	−0.119 *** (0.037)	−0.112 *** (0.039)
CEO tenure	−0.086 (0.104)	−0.087 (0.104)	−0.092 (0.104)	−0.086 (0.106)
TMT Faultline		−0.043 (0.157)	−2.053* (1.074)	−2.413** (1.084)
TMT Faultline2			1.449 * (0.766)	1.688 ** (0.072)
CEO powers				−0.088 (0.134)
TMT Faultline× CEO powers				8.157 ** (4.142)
TMT Faultline2× CEO powers				−5.312 * (2.965)
Year	Control	Control	Control	Control
F	14.60	13.74	13.21	11.61
R2	0.185	0.185	0.187	0.192
N	1060	1060	1060	1060

Standard deviations are in parentheses; ***, **, and * indicate significance at 1%, 5%, and 10% levels.

is the multiple linear regression result of faultlines and CSiR, and it can be found that the result is not significant. This indicates that there is no simple linear relationship between knowledge-based faultlines and CSiR; thus, to determine whether there are other relationships between them, we carried out further verification. Model 3 shows the regression results of the nonlinear relationship between faultlines and CSiR, demonstrating there is a U-shaped relationship between them (β = 0.766, p < 0.1). This result indicates that as the knowledge-based faultlines increase, they reduce the irresponsible behavior of the firm before reaching the critical point, and after reaching the critical point, a further increase in the knowledge-based faultlines will no longer reduce the irresponsible behavior of the firm. Thus, H1 is verified. We also determine the knowledge-based faultline critical point as 0.5. This validates the scholars’ previous observation that medium-strength knowledge-based faultlines can better balance the diversity and synergy of the team, thus enabling the company to communicate, learn, and collaborate [48,49].

To test H2, Model 4 shows the result of the moderating effect of CEO power, and it is found that CEO power moderates the U-shaped relationship between knowledge-based faultlines and CSiR (β = −5.312, p < 0.1). Before the knowledge-based faultlines reach the critical point, a higher CEO power can promote the integration of heterogeneous knowledge and facilitate the positive influence of the knowledge-based faultlines on the irresponsible behaviors of the company, so that the company can identify potential risks in a timely manner. When the knowledge-based faultlines reach a critical point, the CEO’s power slows down the negative impact of the knowledge-based faultlines on irresponsible behavior. However, as CEO power increases, it affects the original group’s willingness to integrate knowledge, which makes the amelioration of CSiR behavior by knowledge-based faultlines less obvious. In Table 1 and Table 2, one can observe that the existence of knowledge-based faultlines in high-polluting companies is not simply a linear relationship; knowledge-based faultlines can enrich the knowledge of the team within a certain range and play a positive role in improving the irresponsible behavior of the company. The CEO can play a role in mitigating the conflict of the executive team, so a strong CEO has a positive effect on the communication of the executive team members.

4.3. Further Study

There are 138 private enterprises and 74 state-owned enterprises among the 212 enterprises. Compared to private enterprises, state-owned enterprises have more advantages regarding access to government resources.

State-owned enterprises are pillar enterprises that have a bearing on the country’s economy and people’s livelihoods. State-owned enterprises undertake multiple objectives, such as obtaining profits, maintaining social stability, and safeguarding employment. As a special type of enterprise organization, state-owned enterprises have both economic and social attributes. Compared with other forms of enterprises, state-owned enterprises take on more social responsibility. Although the state supports state-owned enterprises in actively fulfilling their social responsibilities, irresponsible behavior still exists in state-owned enterprises. To determine whether knowledge-based faultlines improve the irresponsible behavior of state-owned enterprises, we conducted the following study.

It can be seen from

Table 3. Regression: TMT faultlines on CSiR in state-owned enterprises.

CSiR	Model 5	Model 6	Model 7	Model 8
Firm age	−0.192 * (0.110)	−0.190 * (0.114)	−0.193 * (0.115)	−0.196 (0.120)
performance	−1.313 (1.056)	−1.312 (1.058)	−1.257 (1.049)	−1.126 (1.041)
Lev	0.550 * (0.307)	0.550 * (0.307)	0.591 * (0.308)	0.606 ** (0.307)
TMT size	−0.118 *** (0.025)	−0.118 *** (0.026)	−0.112 *** (0.026)	−0.115 *** (0.026)
TMT political connection	0.029 (0.080)	0.030 (0.081)	0.031 (0.081)	0.033 (0.083)
TMT age	0.193 (1.413)	0.219 (1.489)	0.163 (1.508)	0.418 (1.549)
TMT degree	−0.473 *** (0.132)	−0.475 *** (0.133)	−0.475 *** (0.134)	−0.446 *** (0.137)
TMTgender	−0.081 (0.358)	−0.080 (0.361)	0.042 (0.361)	0.094 (0.367)
CEO age	−2.805 * (1.559)	−2.804 * (1.561)	−2.560 (1.564)	−2.309 (1.552)
CEO degree	−0.036 (0.088)	−0.035 (0.088)	−0.047 (0.088)	−0.036 (0.089)
CEO tenure	−0.196 (0.181)	−0.197 (0.183)	−0.178 (0.184)	−0.183 (0.192)
TMT Faultline		−0.020 (0.337)	−1.871 ** (0.893)	−3.433 *** (1.133)
TMT Faultline2			1.635 * (0.842)	2.837 *** (1.002)
CEO powers				−0.187 (0.296)
TMT Faultline× CEO powers				18.127 ** (7.869)
TMT Faultline2× CEO powers				−12.706 ** (6.366)
Year	Control	Control	Control	Control
F	6.22	5.85	6.13	5.14
R2	0.200	0.200	0.205	0.223
N	370	370	370	370

Standard deviations are in parentheses; ***, **, and * indicate significance at 1%, 5%, and 10% levels.

that the linear relationship between knowledge-based faultlines and CSiR does not hold in Model 5, and Model 6 presents a non-linear relationship between knowledge-based faultlines and CSiR, which is consistent with the previous hypothesis that there is a U-shaped relationship between knowledge-based faultlines and CSiR in state-owned enterprises (β = 1.653, p < 0.1). There is a critical point in the knowledge-based faultlines; before the critical point, the faultlines can provide a wide range of information, while after the critical point, rich knowledge does not lead to continuous improvement in irresponsible behavior. In addition, state-owned enterprises need to control the faultlines to a certain degree. Model 8 shows the results of CEO power as a moderating variable, showing that CEO power moderates the relationship between knowledge-based faultlines and CSiR in state-owned enterprises (β = −12.706, p < 0.05), and, consistent with the previous hypothesis, CEO power moderates the U-shaped relationship and enables the knowledge-based faultlines to have more time to play a positive role in corporate irresponsible behavior. State-owned enterprises, backed by the government, are less likely to cater to the government than private enterprises, but they are also under social and reputational pressure to actively improve irresponsible behavior. Improving irresponsible behavior may not result in a significant improvement in social reputation, but failing to improve irresponsible behavior and allowing it to grow into a larger problem will inevitably harm the sustainability of the company. For example, the environmental pollution problem caused by Zijin Mining has not only brought losses to the company but also tens of millions of USD in economic losses to the local area.

4.4. Robustness Testing

We also performed robustness tests on the results. The size of the enterprise population has been considered a control variable in most previous studies, and in order to avoid omitting important variables, we tested the robustness of the size of the enterprise population, which was calculated by taking the logarithm of the total number of employees in the firm. The regression results in

Table 4. Robustness: Firm size.

	Model 9	Model 10
Firm age	−0.042 (0.045)	−0.050 (0.451)
Logemployee	−0.309 *** (0.091)	−0.238 ** (0.093)
performance	−3.984 *** (0.760)	−4.065 *** (−0.755)
Lev	0.289 (0.190)	0.293 (0.191)
Ownership	−0.051 (0.078)	−0.301 (0.077)
TMT size	−0.037 * (0.019)	−0.047 ** (0.019)
TMT political connection	−0.066 * (0.038)	−0.057 (0.037)
TMT age	−1.688 ** (0.769)	−1.926 ** (0.771)
TMT degree	−0.141 ** (0.068)	−0.126 ** (0.056)
TMTgender	−0.120 (0.185)	−0.165 (0.188)
CEO age	−0.761 (0.530)	−0.772 (−0.526)
CEO degree	−0.155 *** (0.038)	−0.160 *** (0.038)
CEO tenure	0.054 (0.142)	0.028 (0.141)
TMT Faultline	−2.189 * (1.192)	−7.262 *** (2.693)
TMT Faultline2	1.564 * (0.852)	4.758 ** (1.891)
CEO powers		−3.753 ** (1.670)
TMT Faultline× CEO powers		10.597 ** (5.087)
TMT Faultline2× CEO powers		−6.825 * (3.565)
Year	Control	Control
F	10.09	8.87
R2	0.189	0.188
N	1060	1060

Standard deviations are in parentheses; ***, **, and * indicate significance at 1%, 5%, and 10% levels.

are consistent with the previous section, so the robustness of this study is considered good.

A common calculation method for TMT faultlines is Thatcher’s [60] method, but this method only divides the executive team into two subgroups by default, and this artificial division of the executive team into two subgroups has certain drawbacks. Subsequently, scholars discovered the ASW algorithm through continuous research [61]. This algorithm can accurately divide TMT into several subgroups and determine the strength of the faultlines. Therefore, we used the ASW method to calculate the faultlines for robustness testing; the findings in

Table 5. Robustness: Asw and CSiR_t+1.

	Asw		CSiRt+1
Firm age	−0.047 (0.049)	−0.044 (0.051)	0.019 (0.050)	−0.025 (0.045)
performance	−0.050 * (0.028)	−0.053 * (0.027)	−4.577 *** (0.838)	−3.427 *** (0.725)
Lev	0.478 ** (0.196)	0.439 ** (0.196)	0.036 (0.210)	0.273 (0.184)
Ownership	−0.067 (0.080)	−0.068 (0.080)	−0.159 ** (0.078)	−0.114 (0.072)
TMT size	−0.074 *** (0.018)	−0.074 *** (0.018)	−0.044 ** (0.018)	−0.053 ** (0.015)
TMT political connection	−0.055 (0.037)	−0.062 (0.038)	−0.020 (0.035)	−0.004 (0.033)
TMT age	−1.255 *** (0.359)	−1.227 *** (0.351)	−2.017 ** (0.871)	−1.692 ** (0.690)
TMT degree	0.874 (0.668)	0.701 (0.701)	−0.157 ** (0.066)	−0.110 ** (0.052)
TMTgender	−0.634 *** (0.226)	−0.573 ** (0.242)	−0.419 ** (0.198)	−0.267 (0.181)
CEO age	−0.441 (0.516)	−0.451 (0.531)	−0.827 (0.506)	−0.941 (0.454)
CEO degree	0.151 (0.123)	0.150 (0.121)	−0.122 *** (0.037)	−0.105 ** (0.036)
CEO tenure	−0.103 (0.142)	−0.131 (0.147)	−0.133 (0.1327	0.004 (0.114)
TMT Faultline	−1.307 * (0.682)	−2.092 *** (0.692)	−2.684 ** (1.211)	−4.068 ** (1.931)
TMT Faultline2	0.956 * (0.579)	1.517 ** (0.597)	1.927 ** (0.861)	3.258 ** (1.545)
CEO powers		−0.016 (0.150)		−2.366 ** (1.034)
TMT Faultline× CEO powers		8.771 ** (3.948)		8.362 ** (3.403)
TMT Faultline2× CEO powers		−5.364 * (3.177)		−6.670 ** (2.725)
Year	Control	Control	Control	Control
F	9.48	8.30	8.70	8.38
R2	0.159	0.173	0.176	0.179
N	1060	1060	1060	1060

Standard deviations are in parentheses; ***, **, and * indicate significance at 1%, 5%, and 10% levels.

indicate that the results of this study are robust.

In some previous articles [43], it was often argued that executive team decisions in the current period have an impact on CSP in the next period, so we chose to lag CSiR by one period in the robustness test. The results obtained are still robust. The regression results in Table 5 are consistent with our hypothesis.

5. Discussion

5.1. Conclusions

Previous research has treated CSP as a whole, when, in fact, the root of improving CSP is reducing or improving irresponsible corporate behavior. High-polluting companies cannot be excluded from fulfilling CSP and ensuring sustainability. Although scholars have focused on the impact of certain TMT characteristics on CSiR, the effect of TMT faultlines on CSiR has not been verified. This study uses upper echelons theory as a theoretical foundation to investigate the impact of knowledge-based faultlines on CSiR in high-polluting companies. It was found that:

(1)

There is a U-shaped curve relationship between knowledge-based faultlines and CSiR, and there is a knowledge-based faultline critical point, which is 0.5. Knowledge-based faultlines can improve CSiR before reaching this critical point. Medium-strength knowledge-based faultlines can better leverage the executive team’s role in integrating knowledge and facilitating the executive team’s strategy to improve irresponsible behaviors for the continued health of the business;

(2)

CEO power can moderate the U-shaped relationship between knowledge-based faultlines and CSiR. With a powerful CEO, communication is facilitated among members of the executive team, the negative impact of faultlines is reduced, and the executive team’s decisions regarding CSR are conducive to the sustainable development of the company;

(3)

State-owned enterprises are considered to be more socially responsible and need to provide employment opportunities for society, improve the treatment of their employees, and actively engage in environmental protection and charity work [62]. There is a U-shaped relationship between TMT knowledge-based faultlines and corporate irresponsible behavior in state-owned enterprises, and the knowledge-based faultlines of the executive team should be controlled to within a certain range. A powerful CEO of state-owned enterprises can reduce the negative effects of faultlines, thus enabling these companies to better use the knowledge to improve their irresponsible behavior.

5.2. Contributions

5.2.1. Theoretical Contributions

We have developed a new definition of the characteristics of executives included in the knowledge faultlines based on the types of firms studied. In the past, faultlines were mostly divided into relationship-based faultlines and task-related faultlines. In this study, overseas background was added to the task-related faultlines, and faultlines formed by education, tenure, functional background, and overseas background were defined as knowledge-based faultlines. Studying the impact of knowledge-related faultlines on CSiR helps us study the influence of TMT knowledge diversity on the formulation and execution of corporate strategy.

Applying knowledge-based faultlines to study the impact on CSiR is a new application of upper echelons theory in the field of CSP, which can better explain how TMT structure specifically affects CSP. Existing research has focused on the impact of TMT background characteristics or individual executive roles on CSiR, and while sub-groups influence corporate decision-making in upper echelons theory [63], this has not been fully validated in the field of CSiR. We introduce TMT faultlines into the CSiR study. Our study concludes that the impact of knowledge-based faultlines on CSiR is not a simple linear relationship but a U-shaped relationship. Knowledge-based faultlines can provide a wealth of knowledge and play a positive role as long as they are kept within a reasonable limit. From the perspective of executive teams, studying the influence of knowledge-based faultlines on CSP is a response to the lack of research on the influence mechanism of corporate irresponsible behavior in recent CSP studies and is conducive to improving corporate social responsibility by improving the composition of team members.

We introduce CEO power as a moderating variable in our study as an application and validation of CEO influence on TMT structure. In previous studies, it was determined that the CEO can act as a bridge for communication [54], while an overconfident CEO may affect corporate integration [64]. In our study, CEO power was found to moderate the conflict among members brought about by knowledge-based faultlines and ensure that faultlines function for a longer period. This verifies that a powerful CEO can promote interaction among members and ensure the smooth execution of corporate strategies. This study enriches the literature on CEO as a moderating variable in the relationship between TMT characteristics and CSP.

5.2.2. Managerial Contributions

The findings of the knowledge-based faultlines study on CSiR have important implications for management practices in high-polluting companies.

High-polluting companies face pressure from the environment as well as government regulation. The activities of businesses are subject to scrutiny, especially in contemporary society. If high-polluting companies are to develop and be accepted by society members, they must reduce their “short-sighted” behavior in production and operation decision-making processes, stimulate their endogenous motivation, and improve their irresponsible behavior.

The CSP strategy involves stakeholder rights and interests and is directly linked to the company’s financial performance; it is a strategy that requires the collective wisdom of TMT members. This study shows that knowledge-based faultlines have a U-shaped relationship with CSiR and that the presence of knowledge-based faultlines not only separates executive teams but also leads to heterogeneous knowledge that brings a wealth of information and different insights. When selecting the members of a TMT, the impact on TMT faultlines should be considered. A new executive brings the knowledge that the organization needs while keeping the knowledge-based faultlines at an appropriate level of strength. Furthermore, enterprises should encourage more communication and cooperation among executive members to mitigate the negative impact of knowledge-based faultlines and improve decision-making efficiency.

The CEO is an important leader of a company. They should not only be clear about the goals of the company but also actively adjust their behavior. A good CEO can avoid further enhancement of faultlines and promote the effective use of information flow and cognitive resources within the team. A CEO should work to create a common strategic goal for the team to enhance member acceptance. Companies need a CEO who can coordinate activities across subgroups so that team members can focus on the mission rather than differences between subgroups. Furthermore, firms should carefully consider a change of CEO to avoid the negative impact of the CEO leaving and enhance the faultlines of the team. When promoting the successor to the CEO, the relevant resume of the CEO needs to be taken into account, and the candidate’s working situation with other team members should be considered on the basis of assessing the personal ability of the CEO.

5.3. Limitations

Firstly, the sample we chose for the study is made up of high-polluting companies in China, so whether the obtained conclusions hold true for other countries will need to be further verified in future studies. This consideration is necessary due to the different characteristics of regulatory, legislative, and economic systems in different regions.

Secondly, the data we selected were up to 2019, which is due to the COVID-19 outbreak in 2020, by which all firms in all industries were affected. Thus, to avoid extreme data, we have not considered firm data after 2020. However, a separate study of post-2020 companies or a comparison with corporate social responsibility strategies from previous years would lead to more interesting conclusions.

Thirdly, we did not examine the path between knowledge-based faultlines and CSiR; in fact, such paths are possible and need to be explored in future studies.

Finally, CEO power was the moderating variable we chose, but there may be other moderating variables that can act as a bridge for executive team members to communicate.