1. Introduction
China has successfully transformed from one of the poorest countries in the world to the second largest economy in a 40-year period of rapid development. A positive trend in its economic growth and improvement in inhabitants’ lives has been observed [
1]. However, frequent safety accidents and livelihood problems reflect a severe phenomenon of regulatory capture in China. Regulatory capture means that interest groups control regulatory policy makers or regulators in various ways (e.g., lobbying and bribery) to seek favorable regulatory policies or implementation results [
2,
3,
4]. Regulatory capture often leads to market failure and damage to the public interest. Under such circumstances, the design and implementation of regulatory policies serve the regulated organizations. For example, the “Baby Milk” scandal shocked the whole country in 2008. The incidents of kidney stones in infants caused by poisoned milk powder were frequently exposed by the media before the situation deteriorated further, whereas due to the hidden interest transmission and regulatory capture, the local government did not take timely regulatory action after receiving public tip-offs, but deliberately concealed the facts and shielded enterprises’ illegal acts. Besides this, the phenomenon of regulatory capture also occurred in the Shuanghui meat additives incident in 2011 and the poisonous runway incident in 2016. Significantly in the expired vaccines incident in 2018, regulatory capture directly led to about 250,000 expired vaccines flowing to the market, triggering consumers’ great panic about vaccine safety in China. It can be seen that behind almost every security incident lies a serious problem of regulatory capture, which substantially threatens residents’ health and interests and leads to a critical crisis of government trust [
5].
Concerning the development of ecological civilization, the Chinese government has been steadily enhancing its efforts towards environmental regulations. Chinese enterprises are faced with more and more stringent constraints during the processes of establishment and operation, including environmental impact assessment (EIA), pollution discharge supervision, and environmental law enforcement and punishment, which have all effectively curbed those illegal business activities [
6]. However, the phenomenon of regulatory capture in the field of environmental protection in China is still profoundly grievous [
7]. As a matter of fact, we can all find the collusion between government and businessmen in the air pollution, water pollution, garbage disposal, and other environmental problems in China.
According to the classic principal—agent model [
8], the environmental regulatory authority (i.e., the principal) does not know the technologies, production, costs, and other operational information of the enterprise (i.e., the agent). As a result, it cannot accurately determine the intensity of environmental regulations. The commissioned regulatory agency (i.e., the supervisor) possesses the time, technologies, expertise, talents, and other dedicated resources to investigate the targeted enterprises in detail. Therefore, there is a first-layer principal—agent relationship in which the regulatory agency is entrusted to supervise businesses instead of the environmental regulatory authority. At the same time, the commissioned regulatory agency requires enterprises to truthfully report business information, which constitutes the second-layer principal–agent relationship. In the context of information asymmetry, enterprises are most motivated to capture the commissioned regulatory agency in order to conceal the true business information and evade regulations. Although the commissioned regulatory agency can enforce the policy impartially, it may accept the captured behavior implemented by the enterprise for its own interests, and conspire with the enterprise to deceive the environmental regulatory authority [
9]. Commissioned regulatory agencies thus become the main targets for capture [
10]. Of particular note, one basic assumption of the double-layer principal–agent model is that the supervisor is an independent third party. In the reality of environmental regulations in China, the commissioned regulatory agency is directly formed and managed by the environmental regulatory authority. In addition, the principal—agent model assumes that the principal is Congress as the highest organ of state power [
11]. However, China’s environmental regulatory authority, i.e., the Environmental Protection Bureau (EPB), is also established and supervised by the government, essentially forming a triple-layer principal—agent relationship (
Figure 1). Taken together, environmental regulation practices in China show obvious hierarchical and complex characteristics, and the resulting environmental regulation capture (ERC) involves mixed actors and their complicated relationships.
It is because there is a complex social network between actors in the case of ERC that a third party makes it difficult to distinguish the network subjects, their behavior and interactions among them. As a result, the existing research mainly discusses the institutional incentives and consequences of regulatory capture at the macrolevel [
12,
13], but pays less attention to the specific regulatory capture in the field of the environment. The lack of in-depth research on the behavioral interactions of all parties and the network structure underlying ERC cannot provide practical guidance for perfecting the environmental regulation system and curbing potential ERC problems [
14]. Accordingly, three key ERC issues require further elucidation. First, what are the structural characteristics and core actors of the ERC network? Second, what are the components and interest relationships in each sub-network? Third, how about the strength comparison and behavior motives among stakeholders? Mainly, this study conducts a thorough analysis of the ERC case in Bobai County’s EPB of Guangxi Province, China, and is intended to explore the hidden interactions between actors at different levels. Based on the triple-layer principal—agent model, the current research can make up for the omission of complexity in ERC problems in the past literature. The role analysis of network actors (e.g., initiator and intermediary) helps to identify the mechanism of ERC and provides valuable suggestions for preventing other types of regulatory capture.
The remainder of this paper is structured as follows. We introduce the theoretical background of regulatory capture theory in
Section 2, followed by a comprehensive account of relevant research in the field.
Section 3 elaborates the social network analysis (SNA) method and the details of a practical case selected in this paper. Then, in
Section 4, quantitative analysis results and discussions are presented around the three key issues. Finally,
Section 5 draws conclusions and offers policy suggestions, and points out new ideas for studies on regulatory capture.
4. Results
We define that if there is an interaction or interest transmission between two stakeholders in this ERC case, their relationship is recorded as 1, otherwise it is 0. Finally, a 20 × 20 relation matrix is sorted out, and matrix nodes represent the stakeholders in the case. Then UCINET 6 and NetDraw 2 (Borgatti, S.P., Everett, M.G. and Freeman, L.C.: Harvard, MA, USA) are applied to quantify and visualize the capture network.
4.1. Overall Network Analysis of ERC
The number of all stakeholders in the network is the scale of this whole network. It is generally accepted that the larger the scale of a network, the more complex the structure of the network, and the greater the influence on internal stakeholders. Although there are more than 20 actual stakeholders in this case, we choose only 20 main stakeholders as research subjects because some stakeholders have the same role. For example, there are 13 EIA experts involved. Therefore, the network scale of this incident is equal to 20, and is relatively dispersed [
41]. It is helpful for core stakeholders to obtain heterogeneous information and establish interest sub-networks.
UCINET 6 is applied to measure the density of this ERC network, and it is found that the network density is equal to 0.195, the average distance among internal stakeholders in this network is equal to 2.614, and the cohesion index based on the concept of “distance” is equal to 0.473. Wellman (1979) [
42] reports that if the network density lies in an interval of [0, 0.25], the linkages between nodes are sparse. It is demonstrated that the main stakeholders in this ERC network are not closely connected. The whole network is loosely established and the cohesion is relatively weak. One actor’s attitude and behavior are not enough to exert a great influence on other actors. However, such a network structure, in which there are closely linked insides and sparsely connected outsides, is conducive to the formation of multiple sub-networks. This is in line with the fact that interest subjects are capable of being divided by organizational units in the actual case. It elucidates that this ERC case presents a characteristic of collective collusion, and individual capture tends to develop into collective capture. For example, D1 first implements active rent-seeking activities towards A3, while A1, A2 and A3 are in the same sub-network, and EIA reports approved by A3 also need to be audited at A1 and A2. Therefore, D1 further implements ERC behavior towards A1 and A2. At the same time, as the office manager of the supervision section, A4 is responsible for the specific environmental regulation matters. As a result, A1, A2, and A3 further capture A4, eventually leading to the capture relationship spreading from A3 to the whole environmental regulatory authority.
4.2. Individual Network Analysis of ERC
4.2.1. Centrality
Table 2 depicts the measurement results of degree centrality, closeness centrality, and betweenness centrality for some dominant stakeholders in the ERC network. First, in terms of degree centrality, D1 has the highest degree centrality value, which indicates that D1 occupies the core position in the network and controls the greatest power to influence other stakeholders. Its ERC behaviors are the most frequent and serious. In addition, A3, B1, and A2 also have high degree centrality values. As illustrated in
Figure 3, together with D1, they establish the ERC supreme power circle. The four stakeholders are the main actors in this case. Second, in terms of closeness centrality, D1, A3, B1, and A2 have the highest closeness centrality, and are the least controlled by other stakeholders, indicating that they are at the core of this relationship network. Third, in terms of betweenness centrality, the characteristics and trends presented are almost consistent with those of the first two centrality indicators. The degree centrality of E7 and E3 is lower, but their betweenness centrality is higher; even higher than that of A2. This indicates that though they cannot dominate other stakeholders in the network, they act as an intermediary to facilitate the communication and interaction among other actors. They are in an important position in this ERC case. With the help of its partnership with D1, E3—the business manager of XY company—helps the poultry farmer E4 to handle the business of EIA report-writing from HB company. Thus, a linkage between D1 and E4 is established, and the latter gets involved in this ERC network. Similarly, E7, the boss of a metallurgical plant in Bobai country, learns from B3 about HB’s EIA report-writing business. E7 then actively seeks out D1 for business cooperation and thus becomes the intermediary among D1, B2 and B3. By contrast, as the director of Bobai EPB, A1 owns the supreme authority in the implementation of environmental regulations; but A1′s passive captured behavior mainly aims towards cooperation with A3′s active capture behavior. A1 does not have an impetus for implementing ERC in practices. As such, A1 cannot exert a dominant influence on other stakeholders’ actions. A1 contributes to this ERC incident by acquiescing in the illegal acts of A2 and A3.
According to the centrality analysis, it is demonstrated that the ERC case forms a core power circle consisting of D1, A3, B1, and A2 (as illustrated in
Figure 2). The capture behavior and collusion interests mainly occur in the environmental regulatory authority and the agency. D1 is the general manager of HB company whose major business is EIA report-writing, and D1′s work is to canvass business orders and dock with the regulatory authority. He owns the broadest linkages in this incident. An interest pyramid of “total contract-subcontract” is formed by him. Therefore, we consider him to be the primary person responsible for this case, i.e., the initiator. Besides, B1, A2, and A3 are the principal leaders of the commissioned regulatory agency and environmental regulatory authority, respectively. They have the final say in the review and approval of EIA reports and the supervision of corporate operations. Meanwhile, they not only have the need for passive capture, but also have the motivation for initiative rent-seeking. For example, A3 contacts D1 on his own initiative after receiving the EIA report from DX hotel (writing by HB company), and implies that only by paying the “approval fee” will this report be approved. Therefore, we consider A3, B1, and A2 to be the main persons responsible for this ERC case.
Figure 3 also illustrates that D1, A3, B1, and A2 have a large number of relationships and extensive linkages, so they hold absolute control over the ERC network.
4.2.2. Cohesive Subgroups
Furthermore, we use the imported relation matrix as the basis for subgroup analysis. In detail, four cohesive subgroups based on fitness are obtained with the help of the “K-cores” function of NetDraw 2.
Figure 4 illustrates the stakeholder composition of four cohesive subgroups. The thickness of a line describes the degree of linkages between two stakeholders. It can be seen that D1, A3, B1, and A2 constitute a stable ERC circle. Their interactions are of great vitality and coherence. The capture behavior and interest transmission mainly occur among the four stakeholders. Besides, the capture behavior between D1 and A2 is calculated to be the most frequent. The twos dominate this ERC case.
Then, the “core—periphery” structure of this capture network is explored by using the “Continuous” function. After 1000 iterations, the correlation coefficient is up to 0.547, the mean of all stakeholders’ coreness values is equal to 0.179, and the standard deviation is equal to 0.134. From
Table 3, the coreness ranking of D1, A3, A2, and B1 is still in the top four. This is consistent with the results of the subgroups analysis based on fitness partition. Therefore, we believe that these fours are core structure subjects, while the remaining sixteens are on the periphery of this ERC network.
D1 is the first initiator of ERC behavior in reality. He owns the most frequent relationships and thus possesses great information advantages and rich social resources. On the one hand, with the growth of HB’s business volume, D1 is increasingly occupying the core position of this capture network. On the other hand, the interest transmissions among stakeholders are becoming more frequent, and former customers are constantly bringing new businesses to HB company. For example, like E7, E6—the boss of a wood processing factory in Bobai country—learns from B1 about HB’s EIA report-writing business and eventually purchases the business from D1. As a result, D1′s core role in the network is constantly consolidated, which further enhances the linkages with A3, B1, and A2, and promotes the establishment of a core power circle. As such, members of the core power circle have the motivation to implement ERC behavior proactively. D1 is the initiator in this case, while A3, B1, and A2 are the main responsible persons.
Other stakeholders are defined as periphery structure nodes in the network. They are dependent on the core structure subjects, and their motive is to provide help for the core subject’s capture behavior, and gain benefits. For example, although A1 has no direct linkage with the supervision bureau, HB company or enterprises, he has extensive linkages and resource mobilization capabilities within the environmental authority. He acquiesces in the bribery of A2, A3, and A4 with rent-drawing, thus helping D1 realize ERC and gaining huge benefits (A1 in this case ultimately gets the highest illegal income owing to his leadership). In addition, as essential partners of the core subjects, A4, B2, B3, and D2 either help to handle specific matters, selectively supervise the illegal emission of enterprises, or drum up deals for the EIA report-writing business. They provide auxiliary support for core subjects, and also obtain a small amount of benefits. Therefore, we believe that periphery structure subjects have no higher motivation to break or flee the network, and they are the indirect responsible persons in this case.
4.2.3. The Structural Hole
Table 4 depicts the measurement value of the structural hole index of the main stakeholders in the ERC network. D1 and A3 have the largest effective scale value and efficiency value among all stakeholders. This indicates their capture behaviors are unrestricted and efficient. Besides, their constraint values are both the smallest. Thus, D1 and A3 are located in the position of structural holes and can easily apply the control advantage to influence other actors’ behaviors. Surprisingly, A3 has a higher hierarchy value than D1. It is demonstrated that A3 is located at the core of this core power circle. In fact, A3, as a spokesperson of the environmental regulatory authority, not only captures superior leaders to engage them in collusion, but also opens up the channel for enterprises to communicate with the supervision bureau. In addition, A3 buys off the EIA experts and manages the jury meeting. He plays the role of “bridge” to promote exchanges and cooperation, thus possesses the most prominent centrality.
Although B1, the head of the supervision bureau, and A2, deputy director of the Bobai EPB, are also at the core of this network, they are not capable of initiating ERC actively. As shown in
Table 5 by the honest broker measurement, B1 and A2 have the largest number of intermediary sizes and pairs except D1 and A3, thus illustrating that the main purpose of their behavior is to cooperate with interactions between D1 and A3. Therefore, even though B1 and A2 have high effective scale values and efficiency values, and they are also located in the core power circle, they cannot control the exchanges of information or interests substantially like D1 or A3. B1 and A2 are essential intermediaries in the network. They make full use of their extensive relationships to play the role of intermediary and profit from it. For example, taking advantage of his leadership role, B1 requires personnel at the supervision bureau to selectively supervise the illegal emission of polluting enterprises, thereby helping D1 capture environmental regulations implemented by the supervision bureau. However, the cost is that D1 must pay a “supervision fee” of CNY 1000 per EIA report.
5. Discussion
Drawing on the regulatory capture and principal—agent theories, this study explored actor roles, interest transmission relationships, behavior motives, and mechanisms of ERC. A case study based on multisource matched data revealed that ERC has the characteristics of concealment and complexity. The environmental regulatory authority (e.g., A2 and A3), commissioned regulatory agency (e.g., B1), and agency of enterprises (e.g., D1) form the core power circle of the ERC network, in which the first two play the role of intermediary and the latter acts as an initiator. In addition, they occupy the core position of the ERC network and have the motivation to proactively implement ERC behavior. Periphery structure stakeholders are dependent on the core structure actors, and their motive is to provide assistance to the core actors’ capture behavior and thus gain benefits.
5.1. Theoretical Contributions
The current research makes several crucial theoretical contributions to the relevant literature. First, we apply the regulatory capture theory to the field of environment-related research, making up for the lack of past literature focusing on regulatory capture in the economic field [
19,
43]. The regulatory capture incident concerning residents’ livelihood (e.g., the environment) is widespread [
44,
45], but the economic development often comes first, especially in emerging economies or in the face of major crises [
46]. As a result, we rarely explore regulatory capture outside the economic sphere. The discussion of ERC in this article contributes to an improved framework of regulatory capture research. Meanwhile, different from the macrolevel perspective commonly adopted in the existing research, we conduct a thorough analysis of the typical ERC case to illustrate the mechanism of the ERC incident. Our research reveals the network roles of stakeholders and their motives for participating in ERC, thus providing targeted implications for designing practical preventive measures and valuable suggestions for evading other types of regulatory capture.
Second, the study combines the practices of China’s environmental regulations to extract a triple-layer principal—agent model, which adds to a growing understanding of actor roles and behavior interactions in ERC. Although the classic principal—agent model explains the potential causes and mechanisms by which the commissioned regulatory agency captures the environmental regulatory authority [
8], the concealment and complexity characteristics of ERC behavior are omitted. Especially in grassroots authority, to increase their own interests, the environmental regulatory authority has the will to actively seek rent and deceive the superior government [
47]. The agency of enterprises naturally possesses the motive of active capture. Therefore, the interest transmission network has been constantly strengthened, and core actors’ roles and positions have also been continuously consolidated. Based on a typical ERC case, the study identifies the interactions among multiple stakeholders and illustrates the complex and hidden relations in ERC. Our research thus develops the principal—agent model.
Third, although the literature has investigated the regulatory capture issue using different quantitative methods, such as field studies and modeling stimulation [
34,
48], our research is an initial attempt to integrate the quantitative method (i.e., the SNA approach) and qualitative analysis (i.e., the case study) to uncover the process by which an individual’s interest transmission behavior evolves into the collective ERC. ERC involves multiple stakeholders. The interests of different sectors are intertwined. Interactions between individuals are hidden behind formal business dealings. It is difficult to elaborate on the real motives and deep-seated causes by analyzing stakeholders’ behavior from the theoretical perspective. Our analysis of a specific ERC case explains the interactions between stakeholders, restoring the evolution process of ERC in real situations, and enhancing the credibility of research conclusions. Besides this, we also use the SNA approach to quantify the network role of each actor and its structural position, making the relationships of ERC visible. The effort of integrating qualitative and quantitative methods provides a methodological reference for future research.
5.2. Practical Implications
The findings also have important implications for management in environmental regulations. First, the ERC presents an obvious characteristic of collective collusion. Individual capture often evolves into collective capture. Network members will gradually lose the motivations and abilities to break or flee the relationship network, along with the constant consolidation of interest networks and frequent interactions among stakeholders. Thus, the principle of flattening must be considered when constructing an organizational structure, in order to realize the decentralization and balance of power, thereby weakening a single subject’s control of collective actions. It is necessary to promote the transformation of the environmental regulation principal–agent mode from a peer or superior government (as the single supervisory entity) to multiple subjects. The command-and-control regulatory mode should be supplemented by the synergetic effects of the public, news media, industry associations, and other multiple subjects so as to establish a long-term mechanism of ERC governance [
49].
Second, the SNA results show that different stakeholders are in different positions and hold different degrees of control in the network. Both the enterprises (the role represented by D1) and the environmental regulatory authority (the role represented by A3) are at the core of this capture network, and occupy the position of structural holes. They are the initiator and intermediary for ERC incidents, respectively. Thus, it is suggested that quantitative methods, such as the SNA approach, should be applied to clarify the accountability of the primary responsible persons, the main responsible persons, and the indirect responsible persons for quantifying stakeholders’ responsibility in the final punishment. The peer or superior government departments should intensify the fight against collusion and interest transmission between officials and businesses, so that the cost of ERC greatly exceeds its benefits. The huge penalties for accident enterprises and administrative accountability for regulatory authorities have been demonstrated to be valid in reducing the willingness of peripheral structure subjects to participate in the ERC [
31].
Third, the ERC is dominated by core structure subjects (e.g., D1, the agency of enterprises), but they still need intermediaries (e.g., E3, the business manager of XY company) to play the role of “bridge”. An intermediary is dependent on core structure subjects, so the core structure subject can effectively achieve its capture purpose by the help of intermediaries’ extensive relations. Therefore, breaking the interest community is key in order to evade ERC. It is necessary to separate the rights of jurisdiction and supervision in order to encourage more stakeholders to participate in environmental regulations and report illegal acts. The government can establish an exchange system for environmental-regulatory personnel to carry out regulations across regions. Especially when the ERC happens, local environmental regulators must not participate in the investigation process. Instead, environmental regulators should be dispatched from other regions, so as to prevent regulators from establishing an interest transmission relationship with the regulated enterprises during long-term contacts.
Finally, there is not necessarily a direct relationship between interest subjects in the peripheral position, but once they play the role of intermediary, they will draw more conspirators over to this network, thereby expanding the network scale and profiting from it. Therefore, a blacklist system can be established in the field of environmental protection to regulate the professional and social morality of all parties and induce a fear of ERC among enterprises and authorities. The labor union system has been proven to effectively restrain individual behavior all over the world [
50]. As a result, the government should encourage the establishment of the labor union within the enterprise so as to increase employees’ recognition of the collective and enhance their enthusiasm in participating in production and supervision. In addition, the whistleblower system can encourage employees to report illegal business activities, thus forming a bottom-up, internal supervision mechanism.
5.3. Limitations and Future Directions
Selecting a typical incident in the field of environmental regulations for case analysis, this study deeply analyzes the micro-mechanism for ERC, clarifies the main stakeholders and interest exchanges in this network, and bridges the gap between macro-phenomena analysis and micro-interaction analysis. Despite the implications of curbing regulatory capture problems, the current research has some limitations, which suggest meaningful future research directions. First, we focus on the ERC behavior from the perspective of interest transmission, but as social persons, stakeholders are affected by irrational facts. Thus, estimating the impact of individual psychological factors on ERC behavior may be one area where we can extend this study. Second, the actors and structure of the ERC network are constantly changing. Will the ERC network structure change significantly and regularly? Do the core structure subjects always occupy the structural hole position? Therefore, future researchers may explore the dynamic evolution of the ERC network. Additionally, we select a typical case to uncover the ERC process, which limits the generalizability of the research conclusions to some extent. It is recommended that future research conduct a comparative analysis of multiple cases to improve the credibility and validity of our conclusions.