Collaborative Governance of Tower Crane Safety in the Chinese Construction Industry: A Social Network Perspective

Abstract

Tower crane safety governance is an important issue related to the sustainable development of China’s construction industry. The complex collaborative relationship among stakeholders determines the efficiency of tower crane safety governance. From the perspective of social networks, this study constructs a collaborative governance structure model of tower crane safety from four dimensions, i.e., transaction, supervision, dependency, and communication, and analyzes the structural characteristics of tower crane safety collaborative governance and the mutual relationship among stakeholders. The results show that the tower crane safety governance process has a strong collaborative effect, but that collaboration in terms of supervision and communication among stakeholders is currently poor. The tower crane property owner occupies the core position, so their decisions have a great impact on tower crane safety. The power of the government is too large, and the power of supervision is too small, which affects the collaboration enthusiasm of other stakeholders, thus reducing the overall collaboration efficiency. The findings provide theoretical support for tower crane safety management in the construction industry in China. The social network perspective presented in this study can be applied to clarify relationships among stakeholders in other construction safety governance fields.

1. Introduction

According to the China Construction Machinery Industry Yearbook, the annual demand for tower cranes in China has been growing since 2016. By the end of 2020, the number of tower cranes in China was about 410,000. Tower cranes are considered vitally important lifting equipment in the construction sector [1,2], as they can effectively transfer construction elements and improve engineering construction efficiency [3,4]. However, tower crane operation involves dangerous processes such as manufacturing, mobilization, installation, hoisting, and dismantling, especially in high-intensity, high-altitude operation [5,6]. Numerous studies have found that accidents involving tower cranes occur frequently on construction sites [7,8,9], and are one of the leading causes of fatalities in the sector. According to rough statistics, there were at least 600 tower crane safety accidents in China from 2016–2020 (some common accidents are shown in

Table 1. Common tower crane accidents in China in recent years.

Year	Consequences of Accidents	Responsible Departments	Safety Sources from a Social Network Perspective
2017	7 people died, 2 people injured, direct economic loss of 8.4773 million yuan	main contractor, supervisor, tower crane property owner, tower crane installer, government authority	(1) Inadequate construction management system in the project department; (2) A false contract between tower crane property owner and tower crane installer; (3) Supervisor did not report the illegal operation of the tower crane; (4) Tower crane inspector did not register the tower crane.
2019	2 people died, direct economic loss of 2.266 million yuan.	construction investor, main contractor, supervisor, property owner, tower crane installer, government authority	(1) Construction permit and other certificates were not processed; (2) the supervisor had not fully entered the project department; (3) the supervisor ignored the illegal operation at the construction site; (4) the contract for tower crane lease and installation had not been signed; (5) the tower crane was not approved by the tower crane inspector; (6) maintenance was illegally contracted to unqualified individuals.
2019	5 people died, direct economic loss of 5.8 million yuan.	Construction investor, main contractor, supervisor, property owner, government authority	(1) Tower crane property owner forged dismantlement agreement; (2) Dismantle operation was not reported to main constructor and supervisor; (3) Main constructor and supervisor did not review tower crane disassembly documents.

). The large number of casualties and property loss accidents has seriously hindered the sustainable development of the construction industry [10]. Therefore, the implementation of policies to prevent tower crane accidents is urgent.

Most previous studies on tower crane safety management were based on accident analyses. Many researchers have used different methods involving case analysis [11] and expert consultation [12] to determine the causes of tower crane accidents, mainly using frequency statistical analysis [13] to identify high-frequency causes of tower crane accidents (such as human factors and mechanical failure). On this basis, some researchers considered near-misses [14], analyzed the possibility of a near-miss evolving into an accident [15], and evaluated the hazard level of various risk factors [16]. In addition, researchers considered the nonlinear comprehensive impact of risk factors and explored the effects between accidents and factors from a systematic perspective [17]. According to the above studies, some researchers have introduced intelligent technologies such as laser technology [18], machine vision [19], virtual reality [20], and wireless sensing [21,22] into tower crane safety management. Existing studies focus on the unsafe state of objects and unsafe human behavior associated with tower crane operations at construction sites [23,24], while ignoring the complex relationships between stakeholders related to tower crane safety, which are considered the main causes of tower crane accidents [25], as shown in Table 1.

The collaborative governance of tower crane safety involves many stakeholders and is a complex social technology system [17]. There are conflicting interests among different stakeholders, especially in terms of behavioral goals and information asymmetries that cause inefficient stakeholder collaboration. The government hopes to reduce the occurrence of engineering safety accidents by designing policies and norms to ensure maximal safety. While companies focus on maximizing their economic benefit, they may reduce the necessary safety investments, resulting in a large number of tower cranes being improperly used and maintained. Sometimes, in order to pursue more economic benefits, the supervision ignores laws and regulations [26], turns a blind eye to illegal companies, and even colludes with them, resulting in the poor safety supervision standards. For example, a tower crane safety accident that occurred in Yiyang City, Hunan Province, in 2019 involved a construction investor, main contractor, supervisor, property owner, tower crane installer and government authority. It may be argued that a bad contractual and regulatory relationship between these stakeholders indirectly caused the accident. As a result, poor collaboration among stakeholders can greatly reduce tower crane safety [5]. With the introduction of the policy of ‘integration’ management mode of construction lifting machinery in several provinces in China, the relationships among stakeholders involved in the collaborative governance of tower crane safety is worth exploring.

Existing studies have mentioned the importance of stakeholder relationships. Brugha evaluated stakeholders from an organizational perspective and determined their relevance to projects by analyzing their interests, influence, position, and other characteristics [27]. Xue emphasized that collaborative alliances of stakeholders can improve project performance [28], and that common goals among stakeholders can help improve project quality [29]. With the gradual advancement of stakeholder research, social network analysis has been widely used in related research [30]. SNA provides the ability to assess the quantitative results of stakeholder importance, with some researchers using social network attributes to analyze interdependence [31], communication [32], collaboration, and alliance [33] among stakeholders.

Considering that existing studies mostly discuss tower crane safety management from the perspective of accident analysis, stakeholder relationships are rarely considered. Therefore, this study constructs a collaborative governance structure of tower crane safety from a social network perspective, clarifies the relationship among the stakeholders involved, and aims to fill the gap of social perspective in tower crane safety management. This study contributes to a better understanding of the collaborative mechanisms of stakeholders and provides valuable policy insights for government authorities.

The remaining part of the study is structured as follows: In Section 2, the research framework is proposed, the stakeholders in tower crane safety collaborative governance and their relationships are determined, and the data collection and SNA model are introduced in detail. Section 3 discusses the implications of the research results. Finally, conclusions are drawn in Section 4.

2. Materials and Methodology

2.1. Research Framework

Collaborative governance of tower crane safety is a complex social system, involving interactions among many stakeholders. These characteristics are consistent with those of a social network structure [34]. In order to accurately determine the structure of tower crane safety collaborative governance and analyze the relationships among stakeholders, this study constructed a research framework and four-dimensional network based on the previous research results of SNA and stakeholders. The research framework is shown in Figure 1.

First, cases and literature were collected and analyzed in order to identify the stakeholders of tower crane safety and the relationships among them. Then, the index weights for measuring the stakeholder relationships were determined by combining expert recommendations. On this basis, a questionnaire for the study of tower crane safety collaborative governance was developed After conducting the questionnaire for data collection, the results of stakeholder identification, relationship descriptions, and relationship strengths were integrated into a database in the form of an extensive excel spreadsheet. This database contains all relevant information for the study and was used as input for the construction of composition and four-dimensional networks. Finally, an SNA model for collaborative governance of tower crane safety was constructed using nodes to represent stakeholders and connecting lines among them to represent the relationships among stakeholders [35]. The social network analysis method was used to explore the overall structure of tower crane safety collaborative governance and the individual attributes of each stakeholder in order to study the multi-stakeholder collaborative governance of tower crane safety.

2.2. Identification of Stakeholders and Relationships among Them

After secondary design and assembly at complex construction sites, tower cranes are commonly used to lift large quantities of construction materials. Their height typically increases with the height of the building [36,37]. Multi-stage operations usually involve many stakeholders, such as the governments, installers, and users. Their interests and behaviors are difficult to evaluate; therefore, tower crane safety is closely related to each stakeholder [38,39].

Based on ‘Special Equipment Safety Law of the People’s Republic of China and Rules for the Supervision’, ‘Administration of Construction Hoisting Machinery Safety’ and other relevant laws and regulations, this study analyzed nearly 50 investigation reports on tower crane fatalities and drew the following conclusions: During the life cycle of a tower crane, the manufacturer designs and manufactures the equipment according to the relevant regulations. Usually, the property owner purchases from the manufacturer and registers with the relevant government authority. As the user of a tower crane, the main contractor and some subcontractors lease the tower crane from the property owner. Then, the installer is responsible for the installation of the crane. Upon installation, the crane is checked and accepted by a qualified inspector, who also reports it to the government authority for registration before use. During tower crane operation, a maintenance company is responsible for regular overhauls and maintenance. Finally, the crane is disassembled by the installer. The entire process is supervised by a supervisor [40]. Therefore, the normal use of tower crane requires all stakeholders to complete their tasks effectively in order to achieve safe working conditions. The stakeholders involved in the collaborative governance of tower crane safety are sorted and described in

Table 2. Stakeholders in the collaborative governance of tower crane safety.

Stakeholder	Code	Note
Government Authority	SC1	To supervise and manage all operations in construction projects
Construction Investor	SC2	To invest in construction projects
Main Contractor	SC3	To contract the entire project and take full responsibility for project safety
Subcontractor	SC4	To undertake part of the construction tasks of the construction project from the main contractor (when a tower crane is involved)
Supervisor	SC5	To provide professional technical service for the construction investor and to guide and supervise the main contractor
Tower Crane Manufacturer	SC6	To design and manufacture the tower crane
Tower Crane Property Owner	SC7	To provide tower crane and operators for the engineering project
Tower Crane Installer	SC8	To be responsible for the installation, daily operation and dismantling of the tower crane
Tower Crane Inspector	SC9	To carry out safety inspections on the installed tower crane, operators and related system documents
Tower Crane Maintenance Company	SC10	To provide technical repair and maintenance services
Insurance Company	SC11	To insure against the losses caused by various safety accidents during tower crane operation

.

In this research, it was key to identify the interrelationships among stakeholders and quantify relationship strengths [41]. The network of tower crane safety collaborative governance has large number of participants and involves social governance relationships, such as the obligation of the tower crane property owner to the tower crane manufacturer to buy the equipment or to entrust the tower crane installer with installation and dismantling operation services, which is a transactional relationship among stakeholders [42,43]. The government relies on the tower inspection unit to obtain information about machinery and equipment; this describes the dependency relationship between them. To describe the collaborative governance structure for tower crane safety more comprehensively, it is necessary to combine multiple dimensions of relationships and conduct quantitative research on network relationships.

To identify relationships, a literature survey was first conducted. This research obtained common network relationships based on existing research [44,45,46], optimized them according to the functional characteristics of the engineering project, and an initial list of relationship measurement indexes was produced. Considering the uniqueness of the tower crane safety collaborative governance, this research investigated stakeholders involved in tower crane safety governance, i.e., all interviewees were experienced in project management. A total of 13 experts participated in the interviews, including a project manager, an on-site chief engineer and three contract managers, i.e., a construction investor, general contractor, and tower crane property owner, as well as two engineers, i.e., a supervisor and subcontractor, two government staff from the Emergency Management Agency and the Housing and Urban-Rural Development Authority, respectively, and four professors in the construction engineering management field. The length of each interview was approximately 30 min.

After the expert interviews, the trust index is removed and the network relationships among the stakeholders for collaborative governance of tower crane safety were divided into four measurement dimensions: transactions, supervision, dependencies, and communication. By integrating the qualitative expert scoring method and quantitative AHP, the measurement index of the network relationships among stakeholders and their weights was determined, as shown in

Table 3. Measurement index of the network relationships among stakeholders.

	Measurement Dimension	Measurement Index	Index Weight	Index Description
Network relationship	Transaction	JY	0.2939	A contractual relationship with the partner
	Supervision	JG	0.2517	A supervisory relationship with the partner
	Dependency	YL1	0.1837	A task-dependent relationship with the partner
		YL2	0.1008	A resource-dependent relationship with the partner
	Communication	JL1	0.1009	$\mathrm{In}\mathrm{formal}\mathrm{communication}\mathrm{with}\mathrm{the}\mathrm{partner}$
		JL2	0.0690	$\mathrm{In}\mathrm{informal}\mathrm{communication}\mathrm{with}\mathrm{the}\mathrm{partner}$

.

2.3. Quantification of Relationships among Stakeholders

Considering that some variables (i.e., geographical variation, management level, and project characteristics) are not directly measurable, multiple projects were investigated to diminish the potential effects thereof. Both qualitative and quantitative data collection methods were employed in this study, such as questionnaire surveys, field interviews, and direct literature. To compensate for the potential weakness of the limited sample size, face-to-face interviews with experts were conducted in the pre-research stage in order to elicit their insights on stakeholder collaboration relationships. Meanwhile, project documents were reviewed to facilitate direct observations and obtain in-depth understanding.

The original data were obtained by a questionnaire from four relationship measurement dimensions based on Table 3. A Likert scale was used to demonstrate the consistency between the description of each problem and the actual situation. This research distributed questionnaires to the tower crane operation stakeholders listed in Table 2. The research period lasted from June to August 2021. The research group enjoys a good cooperation relationship with several large building construction companies in China, which guaranteed the reliability of the questionnaire. The questionnaire was administered through an online platform (wjx.cn). The average time for respondents to complete it was around 15 min. In this research, 158 questionnaires were distributed and 144 were returned, with a recovery rate of 91.13%. Invalid questionnaires were eliminated according to two criteria: too short response time or too similar option answers. We obtained 127 valid questionnaires, i.e., efficiency rate of 88.19%. The specific characteristics of the sample are shown in

Table 4. Summary of respondents’ profiles.

Information	Options	Respondents
Work Unit	Government Authority	14
	Construction Investor	11
	Main Contractor	14
	Subcontractor	9
	Supervisor	15
	Tower Crane Manufacturer	7
	Tower Crane Property Owner	13
	Tower Crane Installer	10
	Tower Crane Inspector	12
	Tower Crane Maintenance Company	13
	Insurance Company	9
Job Position	Project Leader	32
	Safety director	41
	Clerk	54
Education Level	High School Graduate/Technical Secondary School Graduates	42
	Undergraduate	68
	Master/Doctor	17
Experience	≤5 years	39
	5 years ~10 years	73
	≥10 years	15
Worksite	Provincial City/Municipality	57
	Prefecture-level City	70
Number of projects involved	<3	42
	3~5	56
	>5	29

.

2.4. Deciphering the Collaborative Governance Structure

SNA originates from social metrology and graph theory. It is an analytical method that integrates sociology, psychology, mathematics, and other disciplines [47,48,49]. SNA emphasizes the integrated application of social science variables and complex project management, and can intuitively reflect and objectively analyze the overall characteristics of social networks and the relationships among individuals in the network, rather than simply analyzing each individual [50].

The obtained relationship strength questionnaire data were input into the database (an Excel spreadsheet). Based on all the data results to calculate the average, a directed matrix of the relationships among stakeholders was constructed, as shown in Figure 2. Then, the relationship matrix was imported to visualize the SNA network, as shown in Figure 3. The node represents the stakeholders and an arrow line indicates that a relationship points from one stakeholder to another. Meanwhile, the obtained results included different indicators. According to these indicators, the positions of stakeholders in the composition and four-dimensional relationship of tower crane safety collaborative governance were analyzed.

The network indicators used in SNA usually include density, centrality, and structural hole index [51]. Network density and centralization can measure the status of the entire safety collaborative governance network, while indicators such as centrality and structural hole evaluate the status of each stakeholder in the network [52]. The positions of stakeholders in the network can influence their behavioral decisions and the effect of safety governance. The individual behavioral decision of each stakeholder could influence the entire network, thereby achieving enhanced safety in tower crane operations [53].

3. Results and Discussion

3.1. Overall Network Analysis

An overall network analysis, with indicators including network density, network centralization, and average path length, helped reveal various structural characteristics of the entire network. Network density is the ratio of the number of relationships that actually exist in the network to the number of relationships that may exist. It shows the degree of closeness among stakeholders; the greater the network density, the closer the relationship between the stakeholders in the network and the more stable the overall structure of the network. Centralization explores the tendency of a network graph to concentrate on a certain point, and indicates the overall cohesion and integration of the graph, including its degree centralization, betweenness centralization, and closeness centralization. The greater the centralization, the more centralized the network, the more uneven the distribution of power in the network, and the more unstable the network. The average path length refers to the least number of edges that can be connected between any two points in a network. The larger the average path length, the closer the relationship between the stakeholders in the network and the stronger the cohesion. The calculated data are shown in

Table 5. Overall network analysis.

Density	Standard Deviation	Degree Centralization/%		Network Centralization Index/%	Average Distance	Compactness
		Out	In
0.700	0.458	33.200	26.600	16.45	1.309	0.848

.

Table 5 shows that in the network of tower crane safety collaborative governance, the overall density is 0.7000, which indicates that the connections among stakeholders are close, the network structure places severe restrictions on the behavior of stakeholders, and that there is a close relationship among stakeholders. The in-degree centralization is smaller than the out-degree centralization, which indicates that the power distribution among stakeholders who output resources and those who utilize such resources is slightly unbalanced. However, both indicators are low, indicating that the overall stability of the network is acceptable. A betweenness centralization of 18.29% indicated that there were stakeholders in the network who could control resources even though they had little control over others and could not effectively restrict them. Because there was no strong connection in the constructed social network graph, the closeness centralization was not analyzed. The average path was short, and the average clustering coefficient was high, which indicated that the network of tower crane safety collaborative governance has the characteristics of a ‘small world’, and further demonstrates that the relationships among stakeholders are close and that the overall cohesion within the network is strong.

3.2. Centrality Analysis

To analyze the core stakeholders in the network of tower crane safety collaborative governance and reveal the characteristics of the roles played by various stakeholders and the differences between the assumed and actual roles of various stakeholders in tower crane safety governance, this study conducted a centrality analysis, mainly including degree centrality, betweenness centrality, and closeness centrality. Degree centrality evaluates the connectivity of a node by measuring the number of other nodes directly connected to it. The degree of a node is divided into in-degree and out-degree, and the corresponding degree centrality is expressed as in-degree centrality and out-degree centrality, respectively. Betweenness centrality measures the betweenness of a node in other nodes, thereby showing the controllability of the node over interactions among other nodes in the network. Closeness centrality measures the shortest distance from one node to another. If a node can connect to other nodes in a short distance, that node has high closeness centrality, as shown in Table 5.

Table 6. Centrality analysis.

Stakeholder	Degree Centrality		Betweenness Centrality	Closeness Centrality
	Out	In		Out	In
SC1	19	12	4.563	1.000	0.769
SC2	19	18	1.417	0.909	0.714
SC3	26	22	4.100	0.909	0.909
SC4	10	13	0.000	0.625	0.667
SC5	22	23	1.900	0.833	0.833
SC6	3	3	0.000	0.526	0.526
SC7	32	29	16.550	1.000	0.909
SC8	21	20	2.900	0.833	0.909
SC9	14	18	0.000	0.714	0.769
SC10	16	19	2.167	0.769	0.909
SC11	4	9	0.000	0.625	0.714

shows that tower crane property owners and main contractors have a high degree centrality, indicating that they are highly influential stakeholders. In particular, tower crane property owners have the highest in-degree centrality and out-degree centrality, indicating that they are in the core position of the network, have absolute controll over other stakeholders, and are more collaborative and active in risk governance. At the same time, tower crane property owners have the highest betweenness centrality (16.550), indicating that they are an important bridge for the interconnection of stakeholders, have large amounts of resources and information, and are more likely to control other stakeholders. The main contractor and government authority also have high betweenness centrality, playing the role of a ‘bridge’ in the network of tower crane safety collaborative governance. Additionally, these parties are of great significance to the development of collaborative relationships.

For the closeness centrality indicator, the main contractor, tower crane property owner, tower crane installer, and tower crane maintenance company have the largest inward closeness centrality, indicating that the four are highly independent of other related entities in terms of resource input. At the same time, the government authority and tower crane property owner have the largest outward closeness centrality, indicating that they are less dependent on other related entities in terms of resource output.

Our comprehensive analysis results show that tower crane property owner is at the center of tower crane safety governance, and entities with a high degree centrality and betweenness centrality (i.e., the tower crane property owner, main contractor) play the role of intermediaries in the network, creating cohesion for other stakeholders. If these important roles are lost, some stakeholders could lose important ways or even the only way to obtain information and resources. As a consequence, the collaborative relationship in the risk governance structure could disappear and tower crane safety issues could arise. In contrast, the government authority has a larger radiation scope in the collaborative governance structure and more management and control over other stakeholders, even though the control effect is low.

3.3. Structural Hole Analysis

Structural holes refer to the gaps between nodes with complementary resources or information but no direct connection in a social network. In a complex relationship network, by connecting groups of scattered and nonrepetitive nodes, the actors occupying positions of structural holes have more network resources, obtain more nonrepetitive information, benefit from more rights, and control the flow of resources and information to other nodes. Thus, these actors have the highest potential to achieve their goals. Structural hole analysis generally measures the following four indicators: effective size, efficiency, constraint, and hierarchy.

To analyze the constraint, restrictions, and control among stakeholders in collaborative governance, a structural hole analysis was used to measure the network structural hole indicators, as shown in

Table 7. Structural hole analysis.

Stakeholder	Effective Size	Efficiency	Constraint	Hierarchy
SC1	6.249	0.625	0.431	0.129
SC2	4.946	0.550	0.468	0.189
SC3	5.285	0.587	0.406	0.109
SC4	3.388	0.484	0.516	0.148
SC5	4.525	0.566	0.421	0.062
SC6	1.689	0.844	0.750	0.713
SC7	6.586	0.659	0.350	0.115
SC8	4.970	0.552	0.445	0.161
SC9	3.636	0.519	0.467	0.091
SC10	4.650	0.517	0.474	0.198
SC11	3.190	0.532	0.482	0.118

.

The structural hole indicators show that the overall power distribution of tower crane safety collaborative governance network is scattered, and that tower crane property owners and main contractors have the fewest constraints and most extensive relationships in the network. Tower crane property owners and government authorities have the largest effective size; as such, these stakeholders have a good ability to occupy structural holes. Having information advantages and opportunities to control benefits could enable them to collaborate in risk governance and improve the overall resilience and robustness of the collaborative structure of tower crane safety governance. In addition, these stakeholders are more flexible and dominant in collaborative governance, which can further promote their leadership and following capabilities regarding overall collaborative governance. Meanwhile, the supervisor has few constraints and a low effective size; therefore, he/she is weak in terms of collaborative governance.

3.4. Cohesion Subgroup Analysis

Cohesion subgroup analysis is an important method to study the internal structural characteristics of a network. It examines the tightness, closeness, and accessibility of nodes within a network to identify subgroups and relationships therein, i.e., subgroup differences among internal and external nodes of each subgroup. Additionally, it analyzes the heterogeneous roles of different groups in the entire network and the interaction mechanisms among them.

As shown in Figure 4 and

Table 8. Cohesion subgroup density.

	Cliques
		1	2	3	4
Cliques	1	0.667	1.000	0.111	0.111
	2	1.000	1.000	0.333	1.000
	3	0.111	0.000	0.000	0.444
	4	0.222	0.667	0.778	1.000

, the overall scale of the entities interested in tower crane safety governance in this study is small. In the iterative, correlation-based CONCER program, the maximum segmentation depth is selected as 3, and the convergence standard is set as 0.200 for the cohesion subgroup analysis of the tower crane safety collaborative governance network. Four subgroups are generated: (1) Government authority, construction investor, and subcontractors; (2) Main contractor, supervisor; (3) Tower crane manufacturer, tower crane inspector, insurance company; (4) Tower crane property owner, tower crane installer, tower crane maintenance company. An analysis of the density coefficients of the cohesion subgroups showed that the density coefficient of the internal risk governance connection between subgroups 2 and 4 was the largest, i.e., higher than the overall density of the network, indicating that the main contractor, supervisor, tower crane property right owner, installer and maintenance company occupy the core area of collaborative tower crane risk governance and are dominant in collaborative governance thereof. An analysis of the density coefficient among the subgroups showed that the collaborative governance connection between subgroup 3 and other subgroups is low, and that the stakeholders (tower crane manufacturer, tower crane inspector, insurance company) are in the most marginal areas of the governance structure, with the least decision-making independence. In addition, the collaborative connection between subgroups 1 and 4 was also low, indicating that government authorities and construction investors are less collaborative with the stakeholders involved in three important stages, i.e., tower crane leasing, installing, and dismantling, as well as maintenance. Additionally, any unsafe decision made by them may easily lead to the emergence of unsafe conditions.

3.5. Four-Dimensional Network Analysis

As shown in

Table 9. Overall network analysis of the relationship network.

Indicator	Composition Relationship Network	Transaction Relationship Network	Supervision Relationship Network	Dependency Relationship Network	Communication Relationship Network
Density	0.700	0.546	0.564	0.691	0.800
Standard deviation	0.458	0.498	0.496	0.462	0.400
Out-degree centralization/%	33.2%%	39%	48%	34%	22%
In-degree centralization/%	26.6%	28%	26%	23%	11%
Network Centralization Index	16.45%	20.16%	23.56%	16.51%	11.58%
Network out-Centralization	48.61%	51.01%	67.78%	51.32%	32.34%
Network in-Centralization	28.93%	33.15%	33.59%	31.05%	13.84%
Average distance	1.309	1.491	1.509	1.327	1.200
Compactness	0.848	0.568	0.643	0.506	0.400

, the density of the communication relationship is the highest, the density of the dependency relationship is slightly lower than that of the composition relationship, and the densities of both transaction and supervision relationship dimensions are much lower than that of the composite relationship network, indicating that there is no direct connection between half of the stakeholders in terms of transactions and supervision. Therefore, the influence of transaction and supervision relationships on stakeholders’ attitudes and behavior is much less than dependency and communication relationships.

From the degree centralization indicator, we found that the centralization-out was significantly larger than the centralization-in in the transaction and supervision relationship, which indicated that the power distribution was uneven in the whole collaborative governance structure, and that the stakeholders occupying the position of benefit exporting and control are freer than those of benefit receiving with an absence of regulations. Additionally, their behavioral decisions are less constrained by other stakeholders. Meanwhile, combined with the intermediate network central potential indicators shown in Figure 5, the supervision relationship has the largest value, the transaction relationship has the second largest, and both are larger than the composite relationship network, indicating that there are control cores and interest oligarchs in the collaborative governance structure who can efficiently master the resource transfer and security control measures to maximize their interests.

All centralization indicators of the communication relationship network were the lowest, indicating that the communication intensity among stakeholders in the collaborative governance structure is low and relatively balanced. Even for stakeholders with more complex task crossover or closer resource dependence, communication intensity is not focused. Meanwhile the average distance of the communication relationship network is the shortest and the network connectivity is the best, indicating that the information accessibility between stakeholders is high. However, the clustering coefficient is too low, so in the whole collaborative governance process, safety information collaboration between stakeholders is poor, which leads to reduced tower crane safety.

In contrast, the centralization indicators and the average distance of the supervision relationship network are higher compared to the dimensions of transaction, dependency, and communication, while the network density and the average clustering coefficient indicators are lower. This result indicates that at present, the collaborative risk governance of tower crane safety is still dominated by regulatory means, and the core group has a strong degree of freedom regarding decision making, while the marginal group can only obey the regulatory control and has a low motivation to participate in collaborative governance, leading to ineffective collaborative risk management.

Taken together, the average distance of the composite relationship network was smaller than that of the transaction, supervision, and dependency relationship network, and its average clustering coefficient was much larger than that of the other four sub-dimensional relationship networks, indicating that the best effect of tower crane safety management can only be achieved with the synergy of transaction, regulation, dependency, and communication multidimensional relationships, which also proves the importance of collaborative tower crane safety management.

Figure 6 and

Table 10. Centrality analysis of the four-dimensional relationship network.

Stakeholders	Degree Centrality
	Transaction		Supervision		Dependency		Communication
	Out	In	Out	In	Out	In	Out	In
SC1	11	8	37	8	12	13	20	26
SC2	19	21	17	17	18	19	19	21
SC3	22	20	22	18	30	24	35	26
SC4	7	8	2	15	17	16	15	21
SC5	10	11	25	18	28	32	30	34
SC6	5	4	2	4	3	3	3	5
SC7	29	27	24	29	35	29	39	32
SC8	15	12	15	22	27	25	32	23
SC9	10	8	18	20	17	23	18	24
SC10	12	11	14	23	19	22	24	22
SC11	6	16	1	3	7	7	8	9

indicate that the tower crane property owner, the main contractor, and the construction investor have a higher status in the transaction relationship. At the same time, the betweenness centrality of the tower crane property owner is the highest. As an important contributor and beneficiary of the whole life cycle of the tower crane, this individual can easily influence other stakeholders through transactions. The generally low centrality of the remaining stakeholders illustrates the uniqueness of the transaction relationship among the various stakeholders.

From the perspective of the supervision relationship, the out-degree centrality of government authority is much higher than that of other stakeholders, and its betweenness centrality is the highest, indicating that the government authority plays a leading role in the tower crane safety regulatory system, and that it can significantly influence the behavior of other stakeholders. Meanwhile, its in-degree centrality is only higher than that of tower crane manufacturers and insurance companies, indicating that its supervised relationship is missing, which means that other stakeholders have little supervision and control over government authorities. In contrast, supervisors and tower crane inspectors in the supervision relationship did not show high levels of centrality; the out-degree is particularly low, indicating that the two do not effectively fulfill their regulatory roles in the collaborative management of tower crane safety, resulting in safety weaknesses (hidden trouble detection, inspection and testing, operation and maintenance protection and others).

In the dimension of dependency relationships, except for the tower crane manufacturer and the insurance company, the degrees of centrality of other stakeholders are high. In the collaborative governance process, constrained by their own and environmental factors, stakeholders need to rely on capital from elsewhere, policy systems, technology, information, and other resources to perform tower crane safety governance.

The high degree of centrality of supervision indicates that other stakeholders have a great dependence on the systems of supervision, decision-making, and information. The out-degree of the tower crane property owner is greater than the in-degree, which shows that the tower crane property owner has a high level of dependence on the technology of other stakeholders.

In the communication dimension, the degree centrality of the main contractor, supervisor, tower crane property owner, and tower crane installer are very high, i.e., much greater than those of other stakeholders, indicating that these four occupy a core position in the communication relationship. Meanwhile, the out-degree centralities of the main contractor, tower crane property owner, and tower crane installer are greater than their in-degree centralities, which shows that they have considerable control over outside communication. The supervisor has the highest in-degree centrality, so it receives the most information about the safety of the tower crane.

Other stakeholders are not prominent in the communication relationship. Such a structure makes the communication between various stakeholders one-way and closed. As such, stakeholders at the edge do not have enough understanding of the processes and measures of collaborative governance. This lack of dialogue and exchange of opinions reduces their enthusiasm for participation, resulting in effective governance schemes in some cases which are not understood and accepted.

In addition, combined with the two dimensions of dependency and communication, the degree centralities of the main contractor, the supervisor, the tower crane property owner, and the tower crane installer are all higher, indicating that the tasks and resource dependence among them are strong, and that communication is more frequent, making it easies to achieve collaboration.

However, the tower crane maintenance company is also a key stakeholder in the safety governance of the tower crane, and its transaction, supervision, dependency, and communication indicators are low, reflecting the low degree of attention paid to the maintenance of tower cranes at present.

From the overall data, the degree centrality index of tower crane property owners is high in all four-dimensional relationship networks, which confirms the results of the overall network analysis regarding the composite relationship described in the previous section, i.e., in tower crane safety collaborative governance structure, the tower crane property owner occupies a central position.

Second, participation by government authorities in the dimension of dependency and communication is low, which is in stark contrast to its supervision leading role. In the whole collaborative governance structure, redundant regulations and controls are imposed by the government authority, and the efficiency of control is low, resulting in the supervision relationship of all stakeholders being carried out under the requirements of government regulations.

Therefore, at present, China’s tower crane safety collaborative governance still has excessive regulatory dependence on the government, which restricts the enthusiasm of other stakeholders to participate in collaborative governance. Additionally, the lack of supervision coordination practice among these parties leads to insufficient collaboration in the entire tower crane safety governance structure.

Third, the supervisor’s regulatory role is seriously problematic. This individual can obtain a large amount of information about tower crane safety, and has close contact with the core stakeholder (i.e., the tower crane property owner), but cannot capitalize upon these advantages in the communication and dependency relationship in the supervision relationship, resulting in weak supervision.

Finally, the insurance company is only very marginally involved in the collaborative governance of tower crane safety. At present, the business development of insurance companies in the field of engineering safety is relatively backward, and the promotion of safety liability insurance is not adequate. At the same time, awareness related to tower crane safety is poor, and participation among insurance representatives is low,

4. Conclusions

This study systematically categorizes the stakeholders involved in the collaborative governance of tower crane safety, discusses the structure of collaborative governance, and clarifies the relationships among stakeholders in different dimensions. This research reveals the position and influence of each stakeholder on the collaborative governance of tower crane safety from a social network perspective. Some practical findings are as follows:

(1)

The effective governance of tower crane safety is based on effective collaboration among stakeholders. This study combines stakeholder analyses with a social network analysis to study such social aspects, embodied by stakeholder collaboration, in contrast to previous studies, which have not focused on the participants of tower cane safety management. Clarifying the collaborative relationship among stakeholders provides new ideas for collaborative governance of safety from four dimensions, i.e., transactions, supervision, dependencies, and communication.

(2)

The tower crane safety governance process has a strong collaborative effect. Under existing single subject supervision modeled by the government, by only relying on administrative measures to implement safety supervision, it is difficult to realize optimal benefits for tower crane safety collaboration governance. Therefore, by the means of safety information collaboration, a new phase of integrated management of tower crane needs to be gradually realized to promote a greater dependence of stakeholders on funds, technology, resources, and policies, and to improve the mutual supervision and self-supervision capabilities of stakeholders. The establishment of diversified collaborative governance is of great significance to improving the level of tower crane safety governance.

(3)

Tower crane property owners are the core of the whole collaborative governance structure, which has a great impact on connecting and strengthening other weaker relationships. Meanwhile, the safety decisions of the general contractor and the tower crane property owner largely affect the overall safety governance level. As such, it is necessary to focus on checking and accepting their safety planning scheme. Additionally, the administrative power of government authorities is too great, which seriously affects overall collaboration. Government authorities need to reasonably decentralize their powers to other stakeholders to promote collaborative will. Significantly, supervisors can get more information from other stakeholders, but it is considered that they have failed to perform their duties well, and their role in collaborative governance has not been fully brought into play, which increases the possibility for the dissemination or superposition of unsafe behavior among other stakeholders.

This research may seem to have relatively limited sample data from a specific country; however, the selection of this sample was scrupulous and can be considered a reliable representation of tower crane safety stakeholders in the construction industry in China. Although this research was conducted in China, its perspectives are potentially enlightening for other countries. The second limitation lies in the measurement of the relationship, based on the overall judgment of the entire life-cycle of tower cranes. Each stage of the tower crane life-cycle involves the entry and withdrawal of stakeholders, and the relationship strength among the relevant entities could change with risk evolution, which is a shortcoming in this study. Further research needs to be extended to include the different stages of the tower crane life-cycle for comparison purposes. Additionally, a broader study should be conducted in the future to analyze the impact of the stakeholder dynamic influences on the collaborative governance of tower crane safety.