Working without Wages: Network Structure and Migrant Construction Workers’ Protests in China

Abstract Migrant construction workers are among the most vulnerable working populations in China as they are prone to facing the problem of wage arrears under the multi-tier subcontracting system. Based on ethnographic research of migrant construction workers in Tianjin, Shenzhen, Nanchang and Shaoguan, we examine workers’ divergent responses to wage arrears. While extant literature focuses on the positive role of informal networks in facilitating collective action, our findings indicate that the network structure between labour subcontractors and migrant workers plays a key role in enabling or constraining labour protests. We identify two network structures: the satellite network – characterized by arm's-length relationships between subcontractors and clusters of workers; and the spider-web network – characterized by strong relationships between subcontractors and their workers. We found that workers in satellite networks were prone to stage protests over wage arrears, but those in spider-web networks never held collective actions when facing the same problem. We argue that strong guanxi is a double-edged sword for the mobilization of labour protests and that workers’ responses to wage arrears are mediated through the network structure. Future studies may further scrutinize the role of a social network and its operating mechanisms in shaping workers’ working conditions and labour politics.

In the two cases that open this article, both Zhou and Hou were migrant construction workers and faced the same problem of wage arrears when their subcontractors failed to pay them on time. However, their construction teams responded very differently to the problem. While Zhou and his teammates staged collective actions to pressure the labour subcontractor and contractor company to pay them, Hou and his teammates stayed silent. Given that both Zhou and Hou worked under the same subcontracting system in China's construction industry, why did their teams respond to wage arrears differently? Why did some construction workers stage protests while others remained silent when their labour rights were violated under the same institutional and industrial context?
Drawing on ethnographic data, we examine migrant construction workers' divergent responses to wage arrears and explain why some resorted to labour protests while others remained silent. From the 69 migrant construction workers interviewed, 22 had encountered wage arrears, of whom 12 had experience joining collective actions against their subcontractors. The other workers waited patiently, sometimes for years, in the hope that their subcontractors would eventually pay them. The difference in responses, we argue, lies not in macro institutional contexts or individual personal profiles but in the informal network structure where the workers were embedded.
We found that while almost all migrant construction workers were recruited to construction projects through informal ties, how they were recruited resulted in different network structures. Some workers were recruited indirectly by labour subcontractors through other workers who served as intermediaries. In this case, the workers maintained only a distant relationship with the subcontractors. These intermediary workers normally were close to those they recruited, which led to the formation of informal clusters. The intermediaries played the informal role of being cluster leaders and maintained an arm's-length relationship with the subcontractors. The network structure that is made up of subcontractors, cluster leaders and workers was like a "satellite." Workers embedded in this kind of structure were more prone to take collective action in order to deal with wage arrears, as our data will show. On the other hand, if workers were recruited directly by a subcontractor and each worker shared a strong tie with the subcontractor and one another, the network structure was like a "spider-web." Our data indicated that workers embedded in this network never dealt with wage arrears through collective action. We found that strong ties between labour subcontractors and workers largely restrained workers from taking collective action.
Thus, departing from the extant literature that focuses on the positive role of interpersonal ties (especially horizontal ties among members of an interest group) in mobilizing collective actions, our findings highlight the constraining force of interpersonal ties, especially when the ties are structured vertically between superordinates and subordinates, just as between subcontractors and workers in our case. At the same time, we examine how social ties in different network structures enable or constrain the mobilization of labour protests. Our objective is to contribute to the emerging debates in China studies over the effects of informal networks on collective action mobilization and invite more studies along these lines.

Wage Arrears for Migrant Construction Workers in China
In 2020, there were 285.6 million migrant workers in China, among which 18.3 per cent (52.26 million) worked in the construction industry. 3 One of the greatest hardships that Chinese migrant construction workers face is wage arrears. 4 According to available data, the total amount of wage arrears for migrant construction workers accumulated to 33.7 billion yuan by 2005. 5 The wage arrears ratio in the construction industry (measured by the percentage of construction workers experiencing wage arrears) is persistently the highest among all industries. For instance, the wage arrears ratio among migrant construction workers in 2013 was 1.8 per cent compared to 0.9 per cent in the manufacturing sector; and in 2014, it was 1.4 per cent in construction compared to 0.6 per cent in manufacturing. 6 When we look at the absolute number, around one million migrant construction workers per year (1,065,002 in 2013 and 855,271 in 2014) suffered wage arrears.
The precarity and vulnerability of construction workers is in part attributed to the multi-tier subcontracting system of labour deployment. 7 In China, a multitier labour subcontracting system emerged in the 1980s as a result of the construction industry's privatization. 8 The basic structure of the labour subcontracting system can be divided into five levels (see Figure 1). Level one is composed of 3 NBS 2021. "2020 nian nongmingong jiance diaocha baogao" (The national monitoring survey report on migrant workers in 2020 property developers (investors) who are in charge of land reclamation and project design. Once a developer has finished designing a project, it outsources the whole project to a general contractor (level two) through a bidding system. The general contractor then divides the project into a variety of sectors and contracts them to various specialized contractor companies (level three). The specialized contractors do not directly employ workers; instead, they depend on labour subcontractors (level four) to recruit workers into different construction teams (level five). A subcontractor normally recruits, manages and supervises one team of construction workers per project but he may have more than one project at a time and end up managing more than one team of workers. If a subcontractor has many projects at a time, he often recruits a few labour supervisors and assigns each to manage the daily operation of the workers on a team. Through the subcontracting system, the higher-tier players (investors and general contractors) can externalize their responsibilities and rely on subcontractors to invest money in advance, thus reducing the higher-tier players' upfront pay-outs. Generally, an investor pays the general contractor who then pays the specialized contractors in instalments according to the project completion schedule. The specialized contractors then pay the labour subcontractors in instalments upon the completion of certain tasks. After receiving payments from specialized contractors, the subcontractors then disburse wages to the workers.
There are two reasons why migrant construction workers are paid in arrears. One is that labour subcontractors may fail to receive payments from the higherlevel specialized contractors, making them economically incapable of paying their workers. In this case, labour subcontractors may mobilize migrant workers to go on strike to pressure the specialized contractors to make the payments. Another reason is that the labour subcontractors have received payments from specialized contractors, but they are either unwilling or unable to pay workers due to other reasons. For instance, some labour subcontractors may have taken up a number of projects and may transfer the payments of one project to another or invest the payments for other purposes. In this article, we focus on the latter case where wage arrears were caused by labour subcontractors, as it is in this situation that we found variation in worker responses to wage arrears.

Informal Networks, Guanxi and Protest
An intriguing question concerning the study of labour unrest in China is how workers initiate and organize collective action in an authoritarian context. Studies on Chinese contentious politics point to the inherent contradictions in China's authoritarianism and institutionalization of legalism that provide political opportunities for the mobilization of collective action. 9 However, the question remains how workers mobilize when formal, independent organizations conducive to collective resistance are absent and personal risks for participation are high. 10 Lacking these components, collective action in China largely depends on informal interpersonal networks to serve as an alternative mobilization structure. 11 Informal interpersonal ties are found to foster group solidarity and group norms that are conductive to collective action, even in democratic countries. 12 In authoritarian contexts like China, the role of informal networks in mobilizing collective action is imperative as they serve as an informal meso-level organization that marshals and channels individual grievances.
The role of informal networks in recruiting participants and sustaining collective actions in contemporary China has been well researched. In rural contexts, guanxi ties such as family, clan and kinship play a key role in mobilizing villagers into collective protests pertaining to land rights. 13 In urban contexts, homeowners' interpersonal ties with local government officials and the media were found to sustain their collective actions on property right issues. 14 In labour protests, informal interpersonal networks circulated information, amassed materials, and fostered group solidarity and identity toward the emergence of protests. The living arrangements of migrant workers based on locality, kinship and ethnic ties often fostered solidarity among these workers. 15 Laid-off workers were often neighbours and lived in the same buildings due to the former socialist housing programme, which naturally nurtured strong bonds among these workers and facilitated their collective resistance to mass lay-offs. 16 In terms of migrant construction workers, their status is characterized by informal work and precarious employment. Being half-peasants, half-workers and 9 Lee 2007;Cai 2008;Chen, Xi 2012;Gallagher 2017;Elfstorm 2021. 10 Chen, Xi 2012Tao 2017. 11 Hurst 2009;Becker 2012. 12 McAdam and Paulsen 1993. 13 Kuang and Göbel 2013Wang, Chi-mao, and Woods 2013. 14 Shi and Cai 2006. 15 Chan, Chris King-chi, and Pun 2009. 16 Lee 2007 second-class citizens, they are often excluded from the basic rights of urban citizens. 17 These workers largely rely on kinship and locality ties to organize their life in cities. 18 Interpersonal ties were found to function as information channels that were critical to initiating collective petitions that demand payments against wage arrears. 19 Swider provides a comprehensive study of the role of social networks in the employment of Chinese migrant construction workers. 20 She identifies different forms of employment in relation to informal networks, which she terms "employment configurations." One such configuration is "mediated employment," where workers find jobs through informal intermediaries who connect them to larger labour contractors. These large contractors normally hire hundreds of workers from different provinces and therefore do not know the workers personally. Another form of employment is "embedded," where workers are directly recruited by smaller labour contractors through multiple social networks. Workers in embedded employment usually come from one province and develop their social networks in the city enclave.
Our study follows Swider's finding that labour contractors may indirectly or directly recruit workers, resulting in satellite and spider-web networks respectively. Our findings also echo hers in that informal workers are not as powerless and disorganized as expected. 21 However, we diverge from her findings on the issue of wage arrears. This problem is absent in her studies on embedded employment, but our research shows that wage arrears could still occur in this form of employment (i.e. in spider-web networks). We found that precisely because there is an absence of protest in wage arrears cases in embedded employment, it appears as if there were no wage arrears. As we will show, wage arrears also occurred for workers in embedded employment but this form of employment actually impeded mobilization of collective action against subcontractors. Furthermore, while Swider found that protests staged by workers encountering wage arrears in mediated employment (i.e. in satellite networks) were often in the form of retributive violence, 22 our study shows that these workers' protests were more organized and strategic under the current political environment.
As a whole, the extant literature focuses primarily on the enabling power of informal networks in mobilizing protests, and less attention has been paid to their constraining force. An exception is Deng and O'Brien's finding that interpersonal ties were mobilized by authorities to demobilize workers' collective action, which they called "relational suppression." 23 In our study, we go a step further to examine how network structures may facilitate or constrain workers' collective action. Thus, our study examines both the positive and negative 17 Solinger 1999. 18 Lei 2005. 19 Pun and Lu 2010. 20 Swider 2015a, 2015b. 21 Swider 2015b Ibid. 23 Deng and O'Brien 2013. functions of interpersonal connections in mobilizing labour protests and specifies under what conditions which function is more likely to prevail.
Our analytical framework is derived from the literature on Chinese guanxi. The importance of guanxi in Chinese society has been widely documented and analysed in various contexts. For our purposes, we found Fei Xiaotong's insights into the general pattern of guanxi 24 and co-author Cheris Chan's concept of relational properties particularly useful in analysing network structure and labour protests. 25 According to Fei, the general pattern of Chinese social relations is characterized as a "differential mode of association" (chaxu geju 差序格局). It is analogized as concentric circles that appear on the surface of a lake when a rock is thrown into it. Any individual actor (as the Ego) is in the centre of these layers of concentric circles extending from the strongest ties to medium and weak ties and finally to those at arm's length. This ego-centric network structure entails multiple cultural meanings and implications for Chinese behavioural patterns; among which the cultural logic of particularism is found in many guanxi practices and sustained by relational mechanisms like renqing 人情 (interpersonal feelings) and mianzi 面子 (face). 26 Building on Fei's concentric circles and Granovetter's distinction of strong and weak ties, 27 Chan postulated that different tie strengths embodied different relational properties and cultural content. Accordingly, strong ties that are closest to the Ego are defined by asymmetric obligation, affection and trust. In other words, members with strong-tie relations are obligated to provide favours to the other party without expecting immediate reciprocal returns, to be sympathetic with each other and highly trust one another. On the other hand, arm's-length relationships at the outer layer of the circle are defined by symmetric obligations and instrumental calculations. Immediate reciprocity and market exchanges are expected in arm's-length relationships. In between these two poles is a spectrum of variation in the types of relational properties of trust, affection, calculation and obligation. 28 We found that the relational properties associated with different tie strengths that make up network structures help to explain the variation in migrant construction workers' responses to wage arrears.

Method and Data
This study was based on ten months of ethnographic research conducted from February 2016 to July 2017 in four Chinese cities. Field research was first carried out in Nanchang, central China, for three months (February to April 2016) and Tianjin, northern China, for five months (April to August 2016). After coding 24 Fei 1992. 25 Chan, Cheris Shun-ching 2009. 26 Hwang 1987Yang 1994;Yan 1996;Lin 2001. 27 Granovetter 1973. 28 Chan, Cheris Shun-ching 2009 and analysis, another three months (May to July 2017) of field research was conducted in Shenzhen and Shaoguan, in southern China. The first author was stationed in one construction site in each city, except in Tianjin where he engaged in regular observations at two construction sites. We chose five construction sites in four different cities for both theoretical and practical considerations.
Theoretically, we adopted a grounded-theory approach by choosing different cases to find similarities. We selected cases in different regions to increase the possibility of diversity in employment patterns and strengthen the generalizability of our findings and analyses. Workers in the north are often seasonal and turnover rates of these workers in the middle of a project are high; whereas workers in the south are relatively stable as workers usually stay until a project's completion. For practicality, we relied on our personal networks to gain access to the construction sites. Thus, while we selected different regions on purpose, the particular construction sites where observations were conducted were largely determined by their accessibility.
The data presented in this article were mainly derived from interviews. Altogether 87 face-to-face interviews were conducted, of which 69 were with migrant construction workers, four with labour subcontractors, seven with labour supervisors, four with managerial staff from specialized contractor companies, and three with managerial staff from general contractor companies. The construction workers interviewed came from different sectors of the construction industry, such as bar benders and fixers, bricklayers, carpenters, concrete workers and curtain wall installers. They ranged in age from 20 to 69 but were mostly males in their 40s. There were only four female workers in our interview pool. All the migrant workers came from rural areas. Those in Nanchang came from local rural areas, while those in Tianjin, Shenzhen and Shaoguan came from rural areas in neighbouring provinces.
Our interviews with workers covered a wide range of topics, including how they entered the construction industry, their relationships with subcontractors and teammates, perceptions of working conditions, experiences of wage arrears, thoughts and feelings about wage arrears, whether they participated in protests, and the reasons why they did or did not. Our interviews with subcontractors focused on their relationship with workers, and how they organized and managed workers. We also inquired about their roles and strategies in wage arrears and protests. To protect our informants, we use pseudonyms throughout this article.

Divergent Responses from Workers in Different Network Structures
All the workers we interviewed became construction workers because they had some close social ties working in the construction industry. They were referred by their social circles to specific projects and construction teams; however, whether the workers were brought to the industry by fellow workers or labour subcontractors also reliably predicted their differing responses to wage arrears. Those brought in by fellow workers and embedded in "satellite networks," where they shared strong horizontal ties with workers but maintained weak vertical ties with subcontractors, were prone to protesting when faced with wage arrears. On the other hand, those workers recruited directly by subcontractors and thus embedded in "spider-web networks," where they shared strong vertical ties with subcontractors as well as strong horizontal ties with fellow workers, would never stage protests even when subcontractors failed to pay them. Therefore, it is the network structure where workers were embedded and the strength of the vertical ties that determined the presence or absence of collective actions against subcontractors in the case of wage arrears.

Protests in satellite networks
Of the 22 interviewed workers who encountered wage arrears, 12 had previously participated in protests. The forms of protests included strikes, besieging up-line contractors' offices, blocking the highway or construction-site gates, and collective petitioning to local governments. These workers came from various regions but what they had in common was that they were all recruited through fellow workers and hence had weak vertical ties with their subcontractors.
By examining the network structures of the construction teams that protested, we found that they functioned like "satellites." In this structure, a labour subcontractor is connected to several cluster leaders through weak ties. In a satellite network, workers are recruited to construction teams through intermediaries who are themselves workers (e.g. persons Wang, A, B, C and D, respectively in Figure 2), and their relationships with the subcontractors are indirect and weak. Low levels of trust, economic calculations and symmetric obligations are the relational properties of these weak ties. The intermediaries usually also have weak ties among themselves and with the subcontractors because they often come from different places. These intermediaries do, however, share strong ties with the workers they recruited directly. Thus, the intermediaries act as informal leaders and give rise to a number of clusters within a team of construction workers. Workers in a cluster often share strong ties with each other with high levels of trust, affection and asymmetric obligations. The combination of weak vertical ties with subcontractors and strong horizontal ties with workers within clusters foster labour protests in the event of wage arrears.
To appreciate the relationship between the network structure and collective action, we will present a typical case in detail. The case materials were acquired from interviews with three workers (Wang, Liu and Yang) from a construction team specializing in curtain wall installation for a project in Shenzhen. 29 There were slightly more than 100 migrant workers in this construction team. The labour subcontractor, Fu, originally came from Henan but had lived in Shenzhen for many years. He recruited the workers through five migrant construction workers from five different rural localities. Three came from three different towns (Chengguan 城关镇, Xiangbei 向北镇 and Tangnan 塘南镇) in Henan province, and the other two came from Bianshan 边山镇 in Guizhou and Shanquan 山泉镇 in Sichuan. 30 These five workers recruited friends, relatives and neighbours from their villages to join the team, forming five informal clusters of workers and serving as informal cluster leaders. Each leader shared strong ties with workers in a cluster and the workers themselves knew each other well. Each cluster, therefore, was a close-knit group. As three of the clusters came from Henan, they accounted for about 75 per cent of workers in the team. Most of them had never worked for Fu before. On the other hand, most of the workers from Guizhou and Sichuan had worked for Fu before through the same cluster leaders. However, according to the interviewees, the five cluster leaders and all the workers maintained only weak ties with Fu and the ties between clusters were weak. Figure 2 illustrates workers' relationships with one another and with Fu.
Accordingly, a series of collective actions were taken by the workers between 2012 and 2013 in the face of wage arrears caused by Fu. It was a few weeks before Chinese New Year, when some workers from Henan were making plans to go home for the Spring Festival. They asked Fu to pay their wages but Fu found excuses to delay payment, even though he had already received payment from the specialized contractor company. The workers spread the news to other workers, including those from Guizhou and Sichuan, that Fu might not pay them on time. Other workers then urged Fu to pay before Chinese New Year. The total amount of wages Fu owed to the workers was about three million yuan, ranging from approximately 20,000 to 50,000 yuan for each worker. Even though the workers kept pressuring Fu to pay, he kept making excuses. The fact is that he had taken up too many projects and that had led to much financial difficulty. Realizing that Fu was not going to pay, the three cluster leaders from Henan coordinated to mobilize their workers to stop working. They also told workers from Guizhou and Sichuan about the problem and these workers then joined the strike.
The workers went directly to Four Stars Company, which had subcontracted the project to Fu, and urged the project manager to pressure Fu to pay them. The manager at first did not entertain the workers' demands and instead ordered them to return to work; however, the workers refused to leave the company's compound. The manager had no choice but to summon Fu to show up immediately to pacify the workers. Fu first persuaded the workers from Guizhou and Sichuan who had worked for him before to return to work, promising that he would definitely pay them. These workers then left but the workers from Henan, who were working for Fu for the first time, refused to leave. After almost two hours of negotiation, Fu promised to pay the workers in three days. With the manager bearing witness to the promise made by Fu, the workers from Henan also went back to work.
However, Fu did not show up three days later and the workers could not get a hold of him. Enraged by Fu's empty promises, the workers decided to escalate their action. The cluster leaders played a significant role in mobilizing and organizing the escalated action. When asked why he joined the collective action, Liu, a worker from the Chengguan cluster said, "[The cluster leader] coordinated our actions. We knew him well and we trusted him. We simply followed his instructions. We did not know workers from other groups, and I think they also just followed their leaders." 31 Wang, a cluster leader from Henan, told us that he and other cluster leaders discussed the problem and made decisions together. Frustrated by Fu's evasive stance, they decided to lead the workers to Four Stars again. The workers threatened to file a complaint to the local government and sue the company if they failed to resolve the wage arrears problem. The project manager, nonetheless, did nothing to meet their demands. As a result, the workers petitioned the Shenzhen government for help. The government officials stepped in to pressure the contractor company to pay the workers on behalf of the labour subcontractor. The payments were made within four days.
The case above demonstrates the process through which migrant construction workers staged protests until they successfully received their wages. It shows that, on one hand, weak vertical ties between the subcontractor and workers made workers fear that the subcontractor might be untrustworthy, which compelled 31 Interview, Tianjin, April 2016. them to take action against him early. On the other hand, the strong horizontal ties among workers within a cluster and their trust in their informal leaders fostered a sense of solidarity and willingness to participate in risky collective action. The cluster leaders helped to formulate strategies and mobilize fellow workers, and their leadership facilitated coordination and cooperation among workers.
This network pattern in relation to workers' readiness to engage in protests was found in all other protest cases in our data. The workers in satellite networks all pointed to their market relationship with the subcontractors whereby fair exchange was expected. For instance, when asked why he went to petition against the subcontractor, Li, a carpenter from Jiangxi province, said, "[B]ecause we were trying to get back our hard-earned money. It is protected by the Labour Law. [The subcontractor] should have paid us when we finished our job." 32 When workers maintained only weak ties with subcontractors, there were no reasons for them to tolerate wage arrears. "We generate profit for our boss and the boss gives us wages. That is the rule!" This statement was commonly expressed by the workers who participated in protests. Furthermore, labour management in satellite networks is more likely to be harsh and oppressive because of the low level of mutual trust in arm's-length relationships. Master Fang, a bar bender from Hubei, shared of his subcontractor, "The subcontractor's supervisor was constantly overseeing us, counting how many tasks we had completed, and pushing us to do more and do it faster. It was like having a guard pointing a gun at us." 33 In short, when workers had an arm's-length relationship with subcontractors and the labour process was characterized by coercion, these workers would become enraged in the face of wage arrears and were more ready to stage protests.

The absence of protests in spider-web networks
For the ten workers in our interviews who suffered wage arrears but never went to protest, we noted they were all recruited directly by labour subcontractors. They were usually relatives, close friends or hometown neighbours of these subcontractors, so they shared overlapping kinship and social circles. The interpersonal relationships between subcontractors and workers, and among workers themselves, were strong. For these workers, they preferred to work for subcontractors they knew well because they believed strong ties assured timely payment. Zhang, a bar bender from Jiangxi province, explained, "If you don't know [the subcontractor], you don't know what will happen and you don't know if you can be guaranteed your wage." 34 However, to the workers' disappointment, strong ties could also work against them when facing wage arrears. To understand what deterred these workers from 32 Interview, Nanchang, March 2016. 33 Interview, Tianjin, May 2016. 34 Interview, Nanchang, March 2016 taking collective action against subcontractors who owed them wages, we will present another typical case. The case materials were provided by three workers (Lao Yao, Bi and Du) who worked for the subcontractor, Xiao Yao, but did not get paid on time. 35 Xiao Yao just started his business as a labour subcontractor in 2013. He set up a construction team of about 30 workers from his hometown in Henan province. For instance, Lao Yao and Bi were relatives of Xiao Yao's, and Du was a neighbour. The workers therefore shared strong ties due to their overlapping social circles. Of course, as with any relationship some types of guanxi were stronger than others, even within these close social circles. We represented the different modes of association in two concentric circles, one being very close to Xiao Yao, and the other being close but not as close. We then derived the network structure in which the workers were embedded, and it came to look like a spider web. Here, a labour subcontractor is like a spider at the centre sharing a strong tie with each of the recruited workers. While the strong horizontal ties among the workers facilitated group solidarity and coordination, strong vertical ties between the workers and subcontractor constrained collective action.
Toward the end of 2014, when workers completed part of the project and planned to go home for Chinese New Year, Xiao Yao was unable to pay them. He owed each worker about 30,000 yuan and altogether a total amount of about 900,000 yuan. This was a significant amount, considering the workers desperately needed it for Chinese New Year. Xiao Yao told the workers that because he had just started his business, most of the money he had received from the specialized contractor had been spent on production materials and tools. He asked the workers to forgive him for delaying their payment and promised he would pay them as soon as he could. Xiao Yao gave each worker 1,000 yuan for them to return home. The workers went home quietly.
The amount Xiao Yao owed each worker was comparable to Fu in the above case, so then why did these workers simply return home instead of protesting? Lao Yao (Xiao Yao's uncle) explained, "It was not easy for [Xiao Yao] because he had just started his business. We are one big family and if we don't help him, who will? It is alright for him to delay my payment for some time. I trust that once he has money, he will pay us." 36 His uncle's explanation illustrated that it was his sense of trust in the moral ethics of the subcontractor and his sympathies for him that prevented him from staging a protest. Furthermore, members of strong ties are bound by asymmetric obligation. Bi, as another relative of Xiao Yao's, added, "As relatives, it is unacceptable to pressure each other. On the contrary, we should help each other in difficult times." 37 While Xiao Yao asked a favour from the workers by delaying their payment, the "asymmetric obligation" from the strong ties obligated these workers to help. This sense of obligation constrained the workers from staging protests against the subcontractor for fear of ruining the existing relationship. The relational properties of trust, affection and asymmetric obligation made workers in the most inner concentric circle in Figure 3 tolerant of wage arrears.
As for the workers who were Xiao Yao's neighbours, their affection and sense of asymmetric obligation were not as strong; however, they still felt they could trust Xiao Yao because they had known him for a long time and shared overlapping social circles with him. They had reliable information about Xiao Yao's reputation and Xiao Yao also relied on them to maintain a good reputation in his hometown. This gave workers assurance that Xiao Yao would keep his word. When asked why he and his fellow workers tolerated the wage arrears, Du said, "We are from the same hometown, which means I know where he lives and I can always see his house. It's like the saying, 'The monk can flee but the temple will never run away.' I don't worry about the delayed payment. We are from the same hometown and we can't push him too hard. I am sure that he will give me the money sooner or later." 38 Du and other hometown neighbours trusted Xiao Yao's promise because it was grounded in the information they had about him and their overlapping social circles. It was rational for workers to believe that the subcontractor would fulfil his obligation to maintain his reputation in his hometown, otherwise he would have difficulties recruiting workers in the future. In this case, guanxi served the function of "assurance" because subcontractors who violated that trust would be sanctioned socially and their business would eventually suffer as a result. 39 Xiao Yao's case illustrates how the relational properties of strong ties discourage workers from staging protests over wage arrears. We further observed that the stronger the ties were between subcontractors and workers, the longer the delay to payments could be. For instance, Xiao Yao paid most of his hometown neighbours in less than a year, but it took him one and a half years to pay his relatives. This differential treatment was based on the cultural expectation that the closer the relationships are, the more tolerance and forgiveness the favourgivers should have.
Other workers who did not engage in protests over wage arrears shared similar experiences in our interviews. They were all embedded in spider-web networks. The constraining force came from a high degree of trust, affection and sympathies, and a sense of asymmetric obligation: the relational properties of strong ties. The source of trust could come from a belief in the moral character of the subcontractor, as with Lao Yao and Bi, or from densely overlapping social circles, as with Du.
Our findings indicate that migrant construction workers in spider-web networks never staged protests to handle wage arrears caused by labour subcontractors. What happens then if a subcontractor exploits these relationships and keeps delaying payments? Instead of staging collective action, we found that the workers would choose to leave the subcontractor who kept delaying their payments. Given their negative experience of working for relatives or hometown neighbours, they left these spider-web networks and joined construction teams organized in satellite networks. Huang, a technician from Jiangxi province, shared, "My uncle had been a labour subcontractor specializing in curtain wall installation for many years. Many relatives and friends from our hometown had worked for him and I joined his team in 2005. Two years later, he had financial difficulties and delayed paying us for a long time. As a result, many workers, including myself, left and worked for other subcontractorsthose whom we didn't have any pre-existing relationships with." 40 Even though workers in spider-web networks would tolerate wage arrears from their subcontractor, it does not mean that their relationship would not be damaged. For instance, Liu, a concrete worker from Sichuan province, once worked for his brother-in-law who owed him 20 per cent of his yearly wage. After failing to be paid for two years, a grudge grew within their relationship and the two became almost strangers. 41 Nevertheless, as construction teams would eventually collapse when subcontractors persisted in breaking the trust constituted by strong ties, most subcontractors did their best to pay off their debts when they had the financial capacity, sometimes years later. 42

Network Structures as a Key Factor
Our research highlights informal network structures as a key factor in explaining the mobilization of migrant construction workers' collective actions. The impact of this factor is clear when a number of contextual variables are kept constant. These variables include the macro political-economical background, absence of formal organization, labour subcontracting system, causes of wage arrears, amounts of wage owed to each worker, and the duration of wage arrears.
Nonetheless, it is important for us to consider alternative explanations for our observations. First, we ask if it is possible that the scale of the construction teams affects the likelihood of collective action. Satellite network teams are generally larger while spider-web networks are smaller; yet, if scale was a factor we would almost expect an opposite result. Smaller spider-web networks have strong ties that foster solidarity and unity, coupled with easier coordination and speedier organization of action due to their size, which would predict more protests. Meanwhile, large satellite networks composed of clusters with weak ties would undermine workers' collective mobilization. Thus, network size is unlikely to be a viable factor in producing the patterned outcome in our study.
Second, the role of leadership in mobilizing and organizing collective action cannot be ignored. 43 We consider whether leadership is a key factor in determining mobilization toward labour protests in our case study. We find that satellite networks had more experienced informal cluster leaders who organized and coordinated workers. However, other studies show that leadership is not a necessary condition for labour protests in China as workers' grievances can lead to spontaneous collective actions. 44 Our interview data also showed that workers have spontaneously protested when wage arrears occurred. For instance, Zhou, the welder mentioned in the introduction, went on strike with other workers without coordinated leadership. Furthermore, leadership itself could not be a sufficient explanation for mobilization. The effect of leadership on protest mobilization against subcontractors can be realized only when migrant construction workers keep an arm's-length relationship with their subcontractors. Thus, leadership takes effect only under certain network context.
While eliminating alternative explanations, we are not arguing that a satellite network is a sufficient condition for the emergence of labour protests in the Chinese construction industry. Rather, we are asserting that weak ties between workers and subcontractors facilitate worker mobilization, whereas strong ties in spider-web networks constrain protests against subcontractors. In other words, not every wage arrears incident leads to collective action even in satellite 42 Interviews, Tianjin, May 2016, and Shenzhen, June 2017. 43 Cai 2006Chen 2008. 44 Lee 2007Chen and Xu 2012. networks because other factors may hold that action at bay. However, a satellite network is necessary for protests against labour subcontractors to take place. The arm's-length relationship between workers and subcontractors facilitates labour protests when other facilitating elements (e.g. leadership) are also present.
The worker-led protests in satellite networks are not the only type of collective action that migrant construction workers engage in. When an upper-level company is at fault for wage arrears, the subcontractor might mobilize their workers to protest in order to pressure the company to pay, regardless of his tie strength with the workers. In subcontractor-led protests, the weak vertical ties between the subcontractor and workers are compensated by their shared interest. According to our analytical framework, if a subcontractor has a strong tie with the contractor company, the subcontractor may hesitate to initiate collective action against the company. However, that subcontractor will fail to pay his workers who may then initiate protests, resulting in worker-led protests.
To what extent does our analytical framework apply to manufacturing workers in contemporary China? Like the construction industry, the manufacturing sector has a large proportion of migrant workers who rely on informal social ties to secure employment. Familial kinship and hometown network ties among workers are often mobilized during job searches, facilitating the formation and solidarity of clusters on the factory shop floor. 45 However, different from those workers in spider-web networks in the construction sector, migrant factory workers hardly have opportunities to establish close ties with their employers. It is expected that satellite networks characterized by different provincial clusters are more common in the manufacturing sector, 46 and this network structure is likely to contribute to protests. Future studies may probe into the network structures in the manufacturing industry and their impact on labour protests. Furthermore, it would be theoretically and empirically interesting to explore the extent to which our findings apply to other Asian contexts. Studies on migrant construction workers in other Asian countries also show the influence of informal networks on labour markets and labour protests. 47 For example, Khurana recorded how contractors used ties of caste, kinship and locality to recruit women workers in India and examined how workers' ties with contractors constrained their resistance against contractors. 48 This finding is similar to the absence of protests in spider-web networks in the Chinese case. While our analytical framework is built upon Fei Xiaotong's concept of "differential mode of association," we believe that the idea that vertical tie strength could suppress labour protests is applicable to other Asian contexts that share a similar mode of interpersonal relationships. However, the form of those network structures and which structures are prominent may vary and require study. 45 Chan, Chris King-chi, and Pun 2009. 46 Lee 1995. 47 Bal 2016Deshingkar et al. 2019. 48 Khurana 2017.

Conclusion
Many studies on Chinese labour highlight the empowerment of migrant workers by focusing on workers' awareness of their interests and the political goals that make them conducive to collective action. Nonetheless, the external conditions affecting workers are often seen as critical when it comes to the workers' ability to fight for their interests. 49 In this article, we have shown that external conditions that facilitate and constrain workers' collective action are not confined to macro-political contexts and formal organizations, 50 but also include informal network structures where workers are embedded. For migrant construction workers, whether workers are recruited indirectly or directly by labour subcontractors will constitute different network structures that will facilitate or constrain workers' collective action.
Informal networks are usually seen as a positive agent of collective action mobilization in social movement literature and China studies. This enabling power is largely derived from the focus on horizontal ties among potential adherents. The vertical ties among different levels of actors in a hierarchical system are rarely taken into consideration. 51 Here, we move beyond the extant literature to account for both horizontal ties among workers and vertical ties between workers and subcontractors. Although migrant workers in both satellite and spider-web networks shared strong horizontal ties, the strength of the vertical ties between labour subcontractors and workers made a difference in the way workers mobilized over wage arrears. When vertical ties were weak, there were protests; but when vertical ties were strong, there were no protests even when ties between workers were strong. This finding suggests a hypothesis that strong vertical ties could suppress labour protests.
Our study contributes to the emerging literature on the constraining function of Chinese guanxi. Research on business practices and military corruption has begun to focus on the negative impact of guanxi. 52 In labour studies, however, the focus is still on the positive side of guanxi, such as helping migrants settle in and facilitating solidarity among workers for collective action. How guanxi may deter workers from fighting for their labour rights has not been looked at. By examining both positive and negative cases, this article advances our understanding of the relationship between guanxi and labour politics. While Deng and O'Brien found that guanxi ties were used by local authorities to demobilize people who had already engaged in protests, our study examines how guanxi ties could suppress the emergence of protests altogether. Adding to Swider's observation that wage arrears did not occur in spider-web networks, we explained how wage arrears did occur but protests did not, thus making it seem as if wage arrears were not an issue. 49 Lee 2016. 50 Blecher 2008Friedman 2013. 51 Lu and Tao 2017 is a welcoming exception. 52 Gold, Guthrie and Wank 2002;Wang, Peng 2016. Our study also shows the role of Chinese guanxi in migrant workers' labour markets or employment relations. In our case studies, traditional guanxi becomes an indispensable resource that these migrant construction workers rely on. On the one hand, the significance of guanxi is increasing in the reform era. 53 As an instrumental resource and cultural schemata, guanxi plays a significant role in migrant workers' urban livesinside and outside workespecially for those in the informal economy. In the case of migrant construction workers, employment relations were differentiated and mediated through workers' guanxi with other workers and subcontractors. On the other hand, different from Walder's neo-traditionalism that is premised upon political allegiance, 54 market clientelism constituted by guanxi in the reform era is an exchange system that is mixed with morality and interest. 55 Migrant construction workers in spider-web networks were informed by a sense of moral economy. When employers unrestrainedly took advantage of the moral economy to the extent that the balance between morality and self-interest was upset, they destroyed their close relationship with the workers. This caused those workers to move to satellite networks that are more oriented towards self-interest. Thus, worker perception toward their employment changes as the interplay between guanxi, morality and self-interest varies.
To conclude, our ethnographic details provide insight into the kind of network structures that facilitate and constrain collective action as well as workers' negotiating power. How workers' grievances influence political orientation is mediated by the way workers are organized and connected with various actors in the field. Strong guanxi is a two-edged sword that may foster solidarity among workers but can also be used to take advantage of those workers, inhibiting them from defending their rights. Future research on social relations and labour politics in China could focus more on network structures and their impact on the power dynamics among actors at different levels in different organizations and industries.

Biographical notes
Haitao WEI is assistant professor in the department of sociology of Central China Normal University. His research interests include collective action, labour politics, labour process theory and rural society in China. His current research focuses on the working conditions and labour agency of platform workers, and on the state movement of rural revitalization in contemporary China.
Cheris Shun-ching CHAN is associate professor of sociology at the University of Hong Kong. Her research interests include culture, economic practices, qualitative social network analysis, healthcare, new social movements and Chinese 53 Bian 2018. 54 Walder 1986. 55 Wank 1996 societies. She is author of the award-winning book, Marketing Death: Culture and the Making of a Life Insurance Market in China (Oxford University Press, 2012). Her writings have also appeared in the American Journal of Sociology, British Journal of Sociology, Theory and Society, Social Psychology Quarterly and Modern China, among others. She is currently working on two projects. One is about the moral economy of informal payment and social network in gaining access to hospital care in China and Taiwan, and the other is about the sociocultural implications of the social credit system in China.