To protest or not to protest? Migrant workers’ participation in protests in China

Abstract

We present a theoretical explanation for why migrant workers in China should be less likely to participate in protests than other categories of workers. While grievance-based theories of protest would suggest that migrant workers have more incentive to protest than other categories of workers, resource mobilization theory suggests that their capacity to mobilize for collective action is impeded by the conditions of their work situation and their residence. Using survey data from CGSS 2010, we test propositions derived from this framework. We find that a greater sense of relative deprivation is associated with a greater likelihood of participating in protest across all categories of workers. However, we also find that migrant worker status functions as a moderator between grievances and protest participation: compared to urban registered workers, migrant workers are significantly less likely to take part in protest activities when both of them have high levels of perceived unfairness. These findings are robust across all models.

1. Introduction

While there has not been another national protest movement in China since Tiananmen Square in 1989, the number of local protests increased dramatically in the two decades after Tiananmen Square, from approximately 9,000 in 1994 to over 74,000 in 2004 (Tanner, 2004: 138). By 2010, ‘mass incident’ reports totaled over 180,000 according to a report in the Economic Observer (25 February 2011).

Existing studies on post-Tiananmen Square protests point to two main differences between those events and the Tiananmen Square movement itself. First, post-Tiananmen Square protests are local, not national, in scope and motivated by grievances over local issues such as land expropriation, factory layoffs, labor rights abuses, wage delays, environmental damage, or bad working conditions that have emerged amid China's dramatic economic reforms and rapid growth of the last four decades (O'Brien, 1996; Fan, 2002; Cai, 2003; Chen, 2003; Jing, 2003; Hurst, 2004; Lee, 2007; Chan, 2010; Tong and Lei, 2010; Ong and Han, 2019). Second, post-Tiananmen protesters demand tangible and material benefits rather than the democratic transformation of the national political system, as was the case with Tiananmen Square and the 1978 ‘Democracy Wall’ movement (Chan et al., 2014). Third, post-Tiananmen Square protesters are mainly peasants and workers, whereas Tiananmen Square and previous national protests in 1976, 1978, and 1986 were largely student-led movements (Mason, 1994; Mason and Clements, 2002).

How do we account for this wave of protests during the height of China's rapid economic development? Why do workers and peasants protest while the standard of living of the average citizen has been improving steadily for several decades? The trajectory of China's post-Mao developmental success suggests that relative deprivation theory might be useful in explaining protest participation in post-Tiananmen Square China. These protests have occurred not under conditions of widespread severe deprivation but among populations who have benefited from China's four decades of rapid economic development but suddenly face a sharp decline in their living conditions and economic security, relative to their own past and/or to others in their own community. Peasants who received their own plot of land under the ‘household responsibility’ reforms of the 1980s have been displaced from their land to make room for industrial development. Workers in state-owned enterprises (SOEs) have been laid off as their plant was downsized, privatized, or otherwise restructured to make it profitable. Urban residents have been displaced from their residences so the land could be devoted to infrastructure projects or new industrial plants or commercial buildings (Tong and Lei, 2010; Ong and Han, 2019). To those affected by these events, the gains of the past 40 years appear to be in jeopardy for them.

However, even if segments of society experience a sense of grievance that might motivate participation in protest, collective action problems impede successful mobilization of protest for certain categories of workers. Resource mobilization (RM) theory suggests that, to explain protest participation, the focus should be less on the extent of deprivation and more on structural factors that enable or impede the ability of aggrieved groups to solve collective action problems and mobilize participation in protest events (McAdam, 1992; McAdam et al., 1996, 2003).

To date empirical research on protest in China, whether grounded in grievance or RM theories or some other framework, has involved mainly open-ended face-to-face interviews with subjects in a sample of locales (e.g., Hurst, 2004; Lee, 2007; Yan, 2008; Cho, 2009; Wang, 2011) or case studies of particular regions or particular protests (e.g., Chan and Pun, 2009; Pun et al., 2010). Going beyond the findings from the methods mentioned above, this paper uses data from a national survey to analyze the joint effects of grievances and mobilization processes that affect individual participation in those protests. We adopt a grievances-moderator-behaviors framework proposed by Kurer et al. (2018) to highlight the factors that moderate the link between individuals' grievances and protest participation. More specifically, we analyze how variations in both the working conditions faced by different categories of workers and the community ties in which their lives are embedded (especially whether they are registered urban residents or migrant workers) can function as moderating factors that explain variations in both individuals' grievances and their susceptibility to mobilization for protest participation.

RM theory predicts that different categories of workers should have different propensities to participate in protest as a function of their occupational/class status: workers have different sources of grievances depending on whether they are employed by SOEs, township and village enterprises (TVEs), or privately owned enterprises (including both domestic- and foreign-owned). The presence or absence (and strength) of preexisting mobilizing structures varies across these categories as well, as does workers' past experience with different forms of contentious action. As Hurst (2004) highlights, the frames that protest leaders employ are also likely to vary depending upon whether their audience consists of peasants, migrant workers, and employees of SOEs or of private- and/or foreign-owned firms. Finally, the opportunity structure facing workers – including their estimate of the costs/benefits of protest participation – should also vary across firm types and between migrant workers vs registered resident workers. Hence, under the same circumstances (i.e., similar shared grievances) we would expect workers in different types of enterprises to have different estimates of the political opportunity structure for protest that they face. For workers in similar occupational situations, we would also expect differences between registered resident workers and migrant workers in their perception of the political opportunity structure surrounding protest participation due to different cost–benefit calculus for protest participation facing migrant workers compared to resident workers.

This paper focuses on migrant workers and how their Household Registration status (hukou) serves as a moderating factor in the relationship between the conditions of their employment situation and their likelihood of participating in protests. China's economic reforms induced large numbers of citizens in rural communities in rural areas and interior provinces to migrate to urban areas where most of the industrial expansion has occurred since 1980. However, China's hukou has confronted these migrant workers with a unique set of constraints on the terms of their employment. Those constraints generate a different set grievance (compared to registered workers) arising from those conditions of employment. Those constraints should also, then, affect their capacity to mobilize for protest to seek redress of their workplace grievances. Scholars such as Chan (2010), Chan and Pun (2009), Butollo and ten Brink (2012), Wong (2011), Wong et al. (2007), and Zhu (2016) have described the marginal living and working conditions of migrant workers in contemporary Chinese society, compared to other categories of urban workers. We argue that these differences give migrant workers and urban registered workers different incentives to protest and, perhaps more importantly, different capacities to mobilize their compatriots for collective action.

Considering people's residence status, we present a theoretical explanation for why migrant workers should have a lower capacity to protest, despite the fact that their marginal living and working conditions should have generated ample severe and widely shared grievances to motivate protest. We argue that, due to the conditions of their residence status (hukou), migrant workers face additional costs and risks of participation that a registered urban worker in the same occupation in the same factory would not face. For these reasons, migrants' capacity to mobilize for collective action should be weaker than that of registered urban workers. These factors should make migrants less likely than registered urban workers to resort to protest as a way to seek redress for their grievances, all else being equal.

In this paper we use survey data from the Chinese General Social Survey in 2010 (CGSS 2010) to conduct a series of empirical tests of propositions derived from this framework. Our findings suggest that grievance is an important factor in explaining protest participation in China. However, our findings also support the hypothesis that a worker's hukou status – migrant worker vs registered urban worker – functions as a moderating factor between grievances, mobilization, and protest participation: compared to urban registered workers, migrant workers are significantly less likely to take part in protest activities, all else being equal. These findings are robust across all models.

In the following sections we present a theoretical framework for analyzing how China's hukou system affects the protest participation of migrant workers by increasing the risks they face from participating in protests and impeding their ability to mobilize for protest (compared to urban registered workers), thus constraining their willingness to participate in protests, despite the grievances that their employment conditions and hokou status generate among them. After presenting the research design and results of empirical tests from the CGSS 2010 survey data, we conclude with a discussion of the implications of these findings for the prospects for and patterns of protest participation in China.

2. Grievances, mobilization, and protest participation

Relative deprivation theory argues that protests are more likely ‘when a prolonged period of objective social and economic development is followed by a short period of sharp reversal’ (Davies, 1962: 6). As expectations continue to rise while achievements decline, the resulting gap between expectations and achievement generates frustration that, if widely shared and attributable to the state, can motivate large numbers of citizens to engage in political protest (Davies, 1962; Gurr, 1970). China has certainly experienced a prolonged period of social and economic development, initiated by the reforms of the Deng Xiaoping era. As China's post-Mao economic and social system has matured, however, certain segments of the Chinese society have experienced what could be described as a period of sharp reversal in their own economic well-being that, according to relative deprivation theory, should make them more inclined to participate in protests. Won (2004) describes the grievances of laid-off SOE workers who have lost not only their job but the package of social welfare benefits (e.g., housing, health care, childcare) often referred to as the ‘iron rice bowl.’ Hurst (2004: 94) notes that between 1989 and 2004 some 30 million SOE workers were laid off. He documents regional variations in how laid-off workers perceive their deteriorating economic conditions resulting from state reforms of SOEs, to which actors they attribute these grievances, and what frames appeal to which type of workers in which regions as part of a mobilization strategy (Hurst, 2004).

Reforms in the agricultural sector had enabled peasants to obtain secure leaseholder rights to a plot of land under the Household Responsibility System (HRS). Between 1989 and 2005 per capita income of rural residents rose by 300%. Yet, during the same period, the ratio of urban to rural income per capita rose from 2.1:1 to 3.3:1 (Wright, 2010: 115–116). Even peasants who obtained farmland under the HRS are now faced with stagnant incomes and the prospect of having their land expropriated by local governments so that it can be devoted to more lucrative manufacturing enterprises or infrastructure projects (Ho, 2010; Ong, 2014; Wang et al., 2016; Heger, 2020). Between 1990 and 2002, 66.3 million farmers had their land expropriated; in the first half of 2004 alone, government sources reported nearly 47,000 cases of ‘illegal land activities’ (Wright, 2010: 112). ‘Rightful resistance’ movements in the countryside emerged over issues of land expropriations, local cadre corruption, excessive taxes, and illegitimate fees (O'Brien, 1996; Guo, 2001; Yeh et al., 2013; Chen, 2020). Alternatively, tens of millions of rural residents have migrated to urban areas in search of wage labor jobs.

Issues such as farmers' displacement from their land, collective layoffs at SOEs, or wage delays incite protest actions around demands for redress of the grievances that such state and corporate actions generate among the affected populations (Cai, 2003; Chan, 2010; Tong and Lei, 2010). These grievance-based theories of protest participation lead to our first hypothesis:

H1: Individuals who perceive higher levels of unfairness in society are more likely to participate in protest actions.

As scholars such as McAdam et al. (1996, 2003) and Regan and Norton (2005) have argued, grievances alone are not sufficient to explain why people engage in dissident collective action. Tilly (1974: 302) argued that ‘fluctuations in grievances account for the outbreak of collective protest as poorly as fluctuations in the oxygen content of air explain the incidence of fires.’ Even large numbers of aggrieved individuals usually cannot overcome collective action problems: if the protest succeeds, the resulting benefits will be public goods that all can consume regardless of their participation/non-participation in the protest. This makes ‘free-riding’ a preferred alternative to participation because it enables the individual to avoid the costs and risks of participation while still enjoying the benefits of protest, should it succeed in producing the desired public goods.

To overcome this problem, RM theory points to the role of mobilizing structures, framing processes, and variations in the opportunity structure facing different categories of individuals as essential to explaining patterns of participation/non-participation across segments of the population. Mobilizing structures are preexisting networks of social interaction in which individuals' lives are embedded. They constitute long-standing mechanisms by which members of a community cooperate to provide themselves with collective benefits. These can include village organizations, labor unions, religious institutions, and civic associations. If political entrepreneurs can frame issues in such a way that members recognize that their grievances are shared and can be resolved by protest, these mobilizing structures can be activated to mobilize members to participate in protest just as they enabled cooperation for the production of shared benefits.

The willingness of individuals to participate will still vary according to their estimate of the costs and risks of participation measured against the prospects for the protest succeeding in producing the desired reforms. This calculus will vary across individuals according to their perception of changes in the political opportunity structure for protest. More specifically, we believe that the work environment of different categories of workers can enable some categories of workers to mobilize for collective action while other categories' capacity to protest is impeded by conditions of their employment and their residence status.

In this paper we use the Kurer et al. (2018) framework of grievances-moderator-behaviors to add workers' hukou status as moderating factors between individuals' grievances and their protest participation in contemporary China. Among workers in a given enterprise type, hukou status should be associated with greater grievances among migrant workers than among urban registered workers. Enterprise type should also be associated with variations in the availability of effective mobilizing structures and of political entrepreneurs capable of framing worker grievances in ways that persuade them of the legitimacy and potential efficacy of protest as a way to resolve their grievances. On the other hand, workers' perception of the political opportunity structure surrounding protest participation should vary with their hukou status. Existing studies have pointed out that migrant workers face objectively more severe deprivations compared to registered resident workers. According to grievance-based theories, then, they should be more likely than urban registered workers to protest. Using the grievances-moderator-behaviors framework, however, we expect migrant workers to manifest lower levels of protest participation because their work circumstances and residential status impede their ability to overcome free rider temptations inherent in the choice to participate in collective action or not. The unique collective action problems facing migrant workers arise from the effects of China's hukou on the political opportunity structure surrounding their choice to participate in protests or not. In the following sections we introduce China's hukou system and how that affects people's willingness to participate in protests.

3. The Household Registration System and grievances of migrant workers

With its first Five-Year Plan, the Chinese Communist Party adopted an economic development strategy that concentrated on expanding heavy industry in the cities while extracting an agricultural surplus from countryside to finance rapid industrialization (Chan, 2010). In order to make this strategy work, a strict urban–rural population segregation system was implemented to control rural-to-urban migration. The Household Registration System (hukou) was introduced in the mid-1950s for that purpose. Under the hukou system, people who lived in urban areas and worked as laborers were assigned an urban hukou, while those who lived in rural agricultural communes were assigned a rural hukou. One's hukou was assigned at birth and could not be changed except under rare circumstances.

The hukou system defines Chinese citizens' rights to education, housing, social welfare benefits, and employment (Zhao, 1999; Lee, 2007; Wong et al., 2007; Chan, 2010; Xu et al., 2011; Wang and Fan, 2012). Each citizen has a claim to public services and benefits only in the locale where he or she is registered. Given an economic strategy that emphasized heavy industry, urban workers enjoyed full access to social welfare benefits from the state. By contrast, peasants in rural areas received far fewer state benefits (Zhao, 1999; Chan, 2010; Qiu et al., 2011). For instance, Cheng and Selden (1994) report that in 1954 the fixed allocations of flour or rice were 184–212 kg per urban resident but only 143–186 kg per rural resident.

Economic reforms in both rural and urban areas in the early 1980s began to generate strains in the hukou system. The HRS, launched in the early 1980s, enables residents of rural communes to lease (but not own) a plot of farmland. In return, the household is required to remit a certain amount of their output to the state. Beyond that, peasants are free to sell the remaining output in open markets. They are also free to engage in other economic activities for pay as well. Because the amount of land available for distribution under HRS was not enough to provide each commune household with their own plot sufficient to support their family, the dismantling of the commune system and the shift to the HRS created a surplus of rural labor. While some of this labor was absorbed by local TVEs and rural service cooperatives, many peasants were attracted to cities by the boom in new jobs in the service sector, in construction, and in labor-intensive export-oriented industries that marked the beginnings of China's industrial expansion (Oi, 1999; Zhao, 1999; Chan, 2010).

The advantage of leaving their native villages to work in urban areas is that rural laborers can earn significantly more income as wage laborers in urban areas than they can from farming or from what wage labor positions are available in TVEs and rural service cooperatives (Wang and Zuo, 1999; Wong et al., 2007; Chan, 2010; Pun et al., 2010).¹ These earnings are, for many, an essential supplement to what they can earn from farm production. For this reason, many peasants leave their family in the countryside (where they are registered and perhaps have a plot of farmland) and migrate to cities to find work as temporary wage laborers (Qiu et al., 2011; Becker, 2012; Gui et al., 2012; Saich, 2015; Wang and Chen, 2019). Those peasants who have migrated to urban areas in search of wage labor positions are referred to as migrant workers (nong-min-gong). The movement of people with a rural hukou to urban areas in search of job opportunities has turned China's rural–urban divide into a three-tiered social structure: registered urban workers, rural residents, and migrant workers (Chan, 2010: 664). Migrant workers have become a critical component of the urban labor force, with their numbers expanding when demand for labor is high but contracting when demand for labor declines (Chan, 2010: 662; Chan and Selden, 2017).

Migrant workers in urban areas, however, are in a status of being economically accepted but socially excluded. Numerous studies point out that the hukou system negatively affects the life conditions of migrant workers in several ways. First, the hukou system affects the job security and pay of migrant workers. In order to protect the employment rights of local registered residents, local governments in urban areas often impose regulations to prevent migrant workers from taking certain occupations with higher pay and better benefits (Wang and Zuo, 1999; Fan, 2002; Gu et al., 2007; Wong et al., 2007; Chan, 2010; Wang and Fan, 2012). As a consequence, migrant workers are heavily concentrated in jobs that are dangerous, dirty, low-pay, and lacking in any sort of job security (Solinger, 1991, 1999; Wong et al., 2007; Chan, 2010; Qiu et al., 2011; Wang, 2011). Many do get work in foreign-owned firms in coastal provinces, but those firms are not subject to the same labor regulations as SOEs or private firms that are Chinese-owned (Wang, 2011). Lee (2007: 162–163) finds that many migrant workers employed in foreign-invested firms are forced to work 12–14 h a day with few days off while China's Labor Law stipulates a 40-h work week with no more than 36 h of overtime per month.

Second, numerous studies have documented the income gap between urban registered workers and migrant workers. Meng and Zhang (2001) find that in Shanghai, migrant laborers work an average of 14 h more per week than urban residents, but their monthly income is only 61% of urban residents' average earnings. Gu et al. (2007: 3) report that in Hunan, Sichuan, and Henan, ‘the actual average monthly working hours of rural migrant workers is over 1.5 times that of those urban workers, whereas the average monthly income of rural migrant workers is 60% lower than that of urban workers.’ The 2007 national household survey of the China Household Project reports ‘the ratio of the average monthly wage of urban residents to that of rural–urban migrants to be 1.49’ (Knight et al., 2011: 587). Using the CGSS 2008 data, Tian (2010: 96) found a 31.6% income disparity between urban registered workers (16,624 RMB) and migrant workers (12,635 RMB). Tian argues that the hukou system is the barrier impeding any rise in migrant workers' income level.

Third, migrant workers in private- or foreign-owned factories are subject to a variety of other abuses as well (Lee, 2007). It is not uncommon that their wages are not paid for months at a time (Chan, 2010). Migrant workers have little if any legal recourse to seek relief from such treatment precisely because they are not registered residents of the locale where they work. Migrant workers also have few opportunities to move up the career ladder in these firms (Wang and Fan, 2012: 739) (Table 1).

Table 1. Working condition of migrant workers

Source: National Bureau of Statistics of China, various years (http://www.stats.gov.cn/tjsj/zxfb/201704/t20170428_1489334.html, accessed on 06/16/2022).

^a Statistics after 2016 are not available.

Besides job security and pay, a second area of impact of the hukou system concerns migrant workers' lack of access to state-provided benefits. Wang and Fan (2012: 741) find that private enterprises usually do not offer migrant workers social security entitlements nor the employment contracts that they normally provide to urban residents. They found that in Wuhan less than 17% of migrant workers are offered a paper contract, and only 16% are covered by medical insurance. Xu et al. (2011: 12) report that in China only 10–15% of migrant workers participate in employment-based pension and healthcare programs. By contrast, registered residents are entitled to a range of public benefits and services provided by their local governments. These include pensions, medical care, and state-funded education for their children (Wang and Zuo, 1999; Zhao, 1999; Butollo and ten Brink, 2012). Local governments receive funding from provincial and national levels of government to pay for the public benefits of registered residents, but they receive no extra funding to provide the same benefits to migrant workers (Wong et al., 2007; Pun and Chan, 2013). Offering migrant workers benefits therefore would simply increases the financial burden on local governments (Gu et al., 2007; Qiu et al., 2011).

The public education system offers tuition-free primary and secondary education for children. Public schools, however, are only for the children of registered residents. If migrant workers choose to bring their children to the cities, they can enroll their ‘migrant children’ in urban public school only if there is available space and they are willing to pay high tuition fees. As an alternative, migrant workers can send their children to privately run schools (called ‘migrant schools’) where the educational quality is substandard compared to that of public schools (Zhao, 1999; Fleisher and Yang, 2003; Wong et al., 2007; Gui et al., 2012). Lai et al. (2014: 76) find that migrant children in urban public schools significantly outperform their counterparts in migrant schools. They also find that the ‘school resources and teacher qualifications in migrant schools are inferior to those in even the poorest rural schools.’ Lai et al. conclude that migrant schools have been unsuccessful in providing quality education. Otherwise, migrant workers have to leave their children with family for schooling in their native village; they are called ‘left-behind children’ (Lai et al., 2014). However, Yue et al. (2019) find that most of the caregivers in the rural areas (mainly grandparents) do not engage with children (such as playing with toys and reading stories to children), so they find high rates of developmental delay in their samples from rural Shaanxi province. Li et al. (2015) indicate that while China has been engaged in the expansion of college education since 1990s, rural youth from poor counties are about 7 times less likely than urban youth to access any college and 11 times less likely to access China's elite 211 colleges. As a result, the education of migrant children is one of the greatest challenges faced by migrant families.

This set of grievances faced by migrant workers suggests the following hypothesis:

H2a: All else being equal, migrant workers are more likely than urban registered workers to take part in protests.

4. Mobilizing migrant workers

While grievances may provide motivation to participate in protest, one also needs to consider differences between migrant workers and registered urban workers with regard to the political opportunity structure they face – including the costs and benefits of participating – as well differences between migrants and residents with regard to the presence and strength of mobilizing structures (Kurer et al., 2018). Kurer et al.'s grievance-mediation-behavior framework suggests that, besides the collective action problems highlighted by RM theory, migrant workers' participation is mediated by the special constraints that the hukou system imposes on their behavior.

First, with regard to the opportunity structure facing migrant workers, the economic risks of protest participation are higher for migrant workers than for registered urban workers. In the 1990s, the reforms of the state-owned sector with the resulting mass layoffs of SOE workers, the replacement of the commune system with the HRS, and the growth of the rural population all contributed to a surplus of labor in the urban economy. China's continued record of economic growth and development has tightened urban labor markets considerably in subsequent years, but this has not necessarily led to job security for migrant workers. As Figure 1 indicates, there continues to be a steady increase in the number of migrant workers entering the urban labor market.

Figure 1. Total number of migrant workers by years (in million) (statistics before 2008 are not available).

Source: National Bureau of Statistics of China, various years (http://www.stats.gov.cn/tjsj/zxfb/201704/t20170428_1489334.html), accessed on 06/16/2022.

Job security is a primary concern for migrant workers, especially those who relocate their families to urban areas. To the extent that job security is more tenuous for migrant workers than for urban registered workers, the risk of being laid off from their job for participating in a protest event should be a stronger deterrent for migrant workers than for registered urban workers: if migrant workers participate in protests to demand more pay, shorter hours, and/or better working conditions, the management of the enterprise can simply fire them and hire someone else to take their job (Knight et al., 1999; Zhao, 1999).

Migrant workers' job security and job satisfaction are not the responsibility of local governments in the same sense that they are for registered urban workers. Chan (2010: 667) points out that in 2009 migrant workers unemployment rate was about 16.4% (estimated at 23 million) compared to 4.3% for the urban hukou workers. Local governments prioritize the unemployment rate of urban hukou workers over migrant workers. In response to rising unemployment, local governments usually encourage jobless migrant workers to go back to their home villages (Lee, 2007). However, the majority of migrant workers, especially younger ones, do not have much experience in farming, nor do they have a farm plot to which they can return. Many returnees have found that they have become ‘displaced peasants’ as their HRS leaseholder plots have been seized by local authorities who designated these as ‘abandoned’ lands available for new construction (Cho, 2009; Ong, 2014). Third, the income from farm work, especially after China entered the World Trade Organization, has not been sufficient to meet the family's basic income needs, which is why peasants leave their villages in the first place (Pun et al., 2010). As a result, returning to their home village is not an attractive option for migrant workers (Chan, 2010).

The second factor that impedes migrant worker mobilization is weakness or near absence of pre-existing mobilizing structures, compared to urban registered workers. This too is attributable in large part to the hukou system. The presence or absence of such mobilizing infrastructures affects potential participants' cost/benefit calculus for protest (McAdam, 1992; McAdam et al., 1996, 2003). Table 2 shows that the average job tenure for migrant workers is less than 5 years, compared to almost 20 years for registered urban workers. Thus, migrant workers in urban areas face the problem of unstable and unsecure employment with high rates of turnover (Wang, 2011). The high job turnover rate makes it difficult for migrant workers to build stable and strong networks of cooperation among themselves that would enhance their capacity to overcome the collective action problems inherent in mobilizing protests (Fan, 2002; Gu et al., 2007; Becker, 2012; Butollo and ten Brink, 2012).

Table 2. Average and median tenures of urban residents and migrant workers in 1999

Source: Knight and Yueh (2004).

Based on this analysis, we argue that the lack of job security and high turnover rate among migrant workers, and their concentration in categories of jobs where organizing is difficult create organizational obstacles to migrant workers' capacity to participate, individually and as a group, in protests. This is despite the fact that those same conditions should give them ample incentive to participate in protests. This leads to the following competing hypothesis:

H2b: All else being equal, migrant workers are less likely than urban registered workers to take part in protests.

5. Research design

We use survey data from CGSS 2010 to test the propositions discussed above.² Given this, the findings are relevant to protest participation in the Hu Jintao era and earlier, which, as noted earlier, is the era that experienced the largest number and the most remarkable annual rate of increase in local protests in the post-Mao era.

The dependent variable, Protest, for each of our models is a dichotomous measure of whether or not the respondent had the experience of participating in protests. We used two questions, questions D12 and D20, to construct our dependent variable.³

The first hypothesis relates to individuals' perception of unfairness. We created two variables to measure that: Unfair Treatment to measure whether an individual considers him/herself to have suffered any unfair treatment from government officials, while Unfairness measures to the degree to which a respondent considers society in general to be unfair. The wording of the questions can be found in the Appendix.

While the grievance variables measure the incentive to protest, we argue that the capacity to overcome collective action problems varies across categories of occupation. We create three main occupation variables: Urban Worker, Migrant Worker, and Rural Worker. The variable Urban Worker refers to regular urban registered workers with a non-agricultural hukou. The variable Migrant Worker refers to those workers who have an agricultural hukou but now work as wage laborers but not in their registered locale. The variable Rural Worker refers to those who have agricultural hukou and work as wage laborers in their registered locale. How these variables are constructed from the survey questions is explained in the Appendix.

In addition, we created several interaction variables to measure the interaction between individuals' perception of perceived grievances and their occupational category to test the hypotheses that types of occupation function as mediator between individuals' perception of grievance and their protest participation.

We included several standard control variables in the models. Age refers to the respondent's age in 2010. The protest literature points out that younger people have fewer obligations to families and careers and are therefore more likely to participate in protests than older citizens (Schussman and Soule, 2005). The variable Male is included as a control since existing literature indicates that men are more likely than women to participate in protest (McAdam, 1992; Wu, 2012). The variable Education measures the respondent's highest level of education to determine if better-educated citizens are more or less likely to participate in protests (Machado et al., 2011). Ethnic identities make it easier to generate the sentiment of ‘us’ vs ‘them’ between ethnic groups (Tong and Lei, 2010), so ethnic minorities might be more likely to feel deprived by the ethnic majority. Poverty is also related to social unrest (Gurr, 1970), so we include the variable Income that measures the respondent's annual income in 1,000 yuan of RMB (US$141). Data description and descriptive statistics are presented in Tables 3 and 4, and the text of all questions used is presented in the Appendix.

Table 3. Data description

Table 4. Data summary

6. Findings and analyses

Because the dependent variable in each model is a binary outcome, we estimated a series of logistic regression models to test our hypotheses. The occupational categories are measured with a series of dummy variables. Logistic regression requires that one of those categories be omitted from the model to serve as the reference category. We use Migrant Worker as the reference category so that the coefficient for each of the other occupational categories is that category's probability of protest participation compared to migrant workers. Positive coefficients indicate that group is more likely to protest than migrant workers, and negative coefficients indicate that group is less likely to participate in protest than migrant workers.

Table 5 reports the results of five logistic regression models, with the binary protest/not protest variable as the dependent variable. Model A is the baseline model that includes only the two perceived grievance variables and the controls. The result of model A indicates that the two measures of individuals' perception of grievance are indeed associated with protest participation. First, as expected, those who see themselves as having suffered unfair treatment from government officials are significantly more likely to participate in protests: predicted probabilities from model A show that holding other variables at their means, those who have experienced unfair treatment from government officials are five times more likely to take part in protests than those who have not (9.4 vs 1.8%). Second, the more an individual perceives society to be unfair in general, the more likely that person is to participate in protests: the probability of participating in protests is 3.3% for those who perceive society as unfair vs 1.4% for those who perceive it as fair. These results are visually presented in Figure 2. These findings not only support our first hypothesis that the perception of unfairness increases the likelihood that an individual will participate in protests. They also provide statistical support for existing qualitative studies on why people protest in contemporary Chinese society.

Figure 2. Predict probabilities of protest participation by unfairness (from model A).

Table 5. Predictors of individual protest participation

Robust standard errors are given in the parentheses.

***P < 0.001, **P < 0.01, *P < 0.05.

H2a and H2b are based on the logic of RM in that our theory argues that the capacity to mobilize for collective action varies across categories of workers. Model B adds the five occupational categories to test the two competing hypotheses: for given levels of perceived unfairness, protest participation varies across categories of occupation. The finding indicates that migrant workers are significantly less likely than urban registered workers to participate in protests: holding other variables at their means, the predicted probabilities of protest participation are 2.65% for urban workers compared to 0.78% for migrant workers, which means urban workers are 3.4 times more likely than migrant workers to take part in protests.

Since we adopt a grievances-moderator-behaviors framework (Kurer et al., 2018) to argue that being a migrant worker moderates the link between individuals' grievances and their protest participation, models C and D added terms for the interaction between different occupational categories and the two grievance variables.⁴ The results in these two models indicate mixed and interesting findings. First, the result in model C indicates that among those who experienced unfair treatment by government officials (Unfair Treatment = 1), the probability of urban registered workers participating in protests is 13.8% while that of migrant workers is less than 0.01%. Therefore, there was a significant difference between the migrant workers and urban workers regarding their protest participations, even among those in both categories who experienced unfair treatment by government officials.

As shown in Figure 3, model D indicates that among those who perceived higher degree unfairness in society in general, urban workers are substantially more likely than migrant workers to take part in protests as well: the predicted probabilities show that for those who strongly consider society to be unfair (Unfairness = 5), the probability of participating in protests is 4.8% for urban workers compared to 0.14% for migrant workers. These findings suggest that while holding a higher degree of perceived unfairness constant, migrant workers are substantially less likely than urban workers to engage in protest activities. Therefore, hypothesis H2b receives support: protest participation is less likely among migrant workers compared to urban registered workers, controlling for their high degree of perceived unfairness.

Figure 3. Predict probabilities of protest participation by unfairness (from model D).

However, when we hold the degree of perceived unfairness at lower levels, migrant workers are instead more likely than urban workers to engage in protest activities (1.12% for urban workers vs 4.36% for migrant workers when Unfairness = 1, and 1.68% for urban workers vs 1.86% for migrant workers when Unfairness = 2). This finding shows that when holding the degree of perceived unfairness at higher levels, then migrant workers are less likely than urban workers to protest, but migrant workers are more likely than urban workers to protest when the degree of perceived unfairness is low. In other words, when they both consider the society unfair, migrant workers are indeed less likely than urban workers to protest as expected. But when they both consider the society to be fair, then migrant workers are more likely than urban workers to protest.

Several puzzles follow from this finding. First, this finding is in conflict with our first hypothesis that individuals who perceive higher levels of unfairness in society are more likely to participate in protests. That hypothesis receives support from the results in models A and B. The results in model D, however, indicate that migrant workers do not think and behave in such a way: they protest when they feel higher degrees of fairness, not higher degrees of unfairness. Why is this the case? Here we use the concept of hierarchical political trust and the findings of Chen et al. (2021) to give an interpretation of these mixed results.

Existing studies on China's political power structure points out that different levels of government have different concerns and priorities (Blanchard and Shleifer, 2001; Cai, 2008; Tong and Lei, 2010; Chen 2017). On the one hand, the central government prioritizes the legitimacy of the regime; it makes general plans for the nation to maintain popular respect for the Chinese Communist Party. On the other, the main concern of local governments is to develop the local economy while preserving social stability; local political leaders are in charge of policy implementation, and their promotions depend on their performance in their locale. Nevertheless, the central government holds the power to investigate local issues. This system, also called Chinese Federalism (Jin et al., 2005), leads Chinese people to have more trust in the central government than in their local governments (Li, 2016; Huang, 2018). Based on these findings, Chen et al. (2021) further find that in China, people with higher degrees of political trust toward the central government combined with lower degrees of political trust toward local governments are more likely than others to protest. This is because they believe protest actions can draw the attention of the central government authorities to the grievances they have that are largely the responsibility of local government and/or local firms. In other words, people can appeal to the national authorities (whom they trust) to compel local authorities (whom they do not trust) to redress their grievances, which are caused by actions of local governments and/or private enterprises (whom they also do not trust).

The theoretical expectation of this paper is that economic unfairness, such as low pay with long working hours, pushes people to go to the streets. We hypothesized that people who consider society to be fair should have less reason to protest than those who see it as unfair. Recall that migrant workers are peasants who have left their native villages to work in urban areas in search of better economic opportunities. Without a strong network in urban areas, migrant workers might not be able to resolve any grievances through guan xi, the informal personal societal networks, which are more prevalent in rural than urban communities (Noland and Rowley, 2020). Therefore, they need some other mechanism to resolve the unfavorable situation they face in the urban environment. Based on what Chen et al. (2021) find, we suggest that if an individual considers society to be fair politically, then this perception might make the individual believe that his/her unfavorable situation will be resolved by national authorities if he/she participates in protests that draw the attention of national authorities to abuses by local authorities. Thus, on the one hand, migrant workers who believe that the society is fair economically might have less incentives to protest. On the other, if they consider society to be fair politically, migrant workers might believe that participating in protests will be an effective way to resolve their grievances. Unfortunately, the CGSS 2010 survey data do not allow us to disaggregate the concept of ‘fairness’ into economic fairness and political fairness to test this hypothesis for the unexpected findings on high levels of perceived fairness.

With respect to the control variables, the results from all models consistently suggest that males are significantly more likely than females to participate in protests. The odds ratio indicates that, compared to females, males were 1.57 times more likely to participate in protests (model D). This finding is consistent with the existing findings on gender differences in protest participation (McAdam, 1992). Other controls such as variables Age, Education, Minority, and Income, however, are not significantly related to individuals' protest participation.

Overall, our empirical tests offer interesting findings: in contemporary China, the perception of grievance explains why individuals participating in protest actions, but among different types of workers, migrant workers are less likely than urban registered workers to participate in protests. Thus, our findings provide some support for grievance-based theories, they also confirm propositions from RM theory on the critical importance of mobilization capacity and political opportunity structure in explaining who among the aggrieved is most likely to participate in protests in China.

7. Discussion and implication

This research investigates the interaction between grievances and capacity to mobilize in explaining which segments of China's labor force are more or less likely to participate in local protests. The results of our analysis of survey data (CGSS 2010) from a nationwide sample provide support for two theoretical expectations. First, over the course of China's unprecedented economic development, many who benefited initially from that development are now coming to see themselves as deprived, relative to others and/or relative to their own economic well-being and security at the peak of this developmental wave. Over the last 30 years, factory layoffs, land expropriation, and labor rights abuses have produced grievances across certain segments of China's population, and our findings show that those who share that sense of grievance are more likely to protest.

The second and, arguably, more significant finding of this study is that, while it appears that migrant workers have ample reasons to protest, they are substantially less likely to participate in protest than registered urban workers when the degree of grievance is high. We argue this is because, whatever a worker's incentive to participate (as a function of their grievances), their capacity to participate is affected by the conditions of their employment. For migrant workers especially, their residential status affects their perception of the political opportunity structure for protest because of the greater risks and expected costs migrant workers face.

Our study reveals some important features of the status of migrant workers in China. The dramatic increase in the number of rural residents migrating to cities indicates that the stagnation of the rural economy combined with the industrial boom in urban areas has motivated tens of millions of rural citizens to migrate to the city in search of wage labor positions. Chan (2010: 358) concludes that, for those who migrate, the hukou system ‘effectively circumscribed the peasantry's economic, social, and political opportunities and rights, creating a massive pool of super-low-cost rural labor tied to land of very little market value.’ Compared to registered urban workers, larger shares of migrant workers are concentrated in low-wage jobs with no job security and less access to legal protection from abuse by their employer. On average, they work longer hours than registered workers, but they earn less income. Migrant workers are not eligible for social welfare benefits that the government provides to registered resident workers in cities. By the logic of grievance-based theories, these conditions should give migrant workers greater incentive to resort to protests to seek reforms that would provide them with wages and working conditions comparable to registered urban workers.

However, we argue the hukou system also gives migrant workers low capacity to mobilize for protest. First, the costs and risks of participating in protests are greater for them than for registered urban workers. Each migrant worker can be replaced at any time for any (or no) reason. If they engage in protests, they are subject to being fired from their jobs and even expelled from the city. The high job turnover rate among migrant workers also makes it difficult for them to build stable and strong networks of cooperation for protest mobilization. These factors together leave migrant workers with less capacity to protest, despite the fact that they may have more reasons to do so. We found that, although migrant workers face more unfavorable conditions than registered urban workers, migrant workers are still about three times less likely than urban workers to participate in protests. Our findings suggest that when it comes to protest participation, the capacity to organize and mobilize is the key factor that determines which segments of China's urban labor force would be most likely to engage in protests.

Our paper tells a sad story about what happens in the transition underway in Chinese society. Peasants leave their native villages for work as temporary wage laborers in cities. Although they live and work in cities, migrant workers are not registered urban resident because of the hukou system. In other words, they are not considered a part of the city. Many scholars have pointed out that China's hukou system is functioning as an internal passport system, and that China's rural–urban dual system creates ‘Two Systems in One Country’ (Chan and Wei, 2019). The hukou system has created impediments to migrant workers relocating permanently to the cities in which they work, including the inferior education opportunities for migrant children, and their ineligibility for most social welfare benefits (Lai et al., 2014; Li, 2016).

While the findings of this paper suggest that migrant workers are less likely than urban workers to protest when the degree of perceived unfairness is high, there is evidence that migrant workers do participate in protest activities (Friedman and Lee, 2010; Chen and Tang, 2013; Elfstrom and Kuruvilla, 2014). However, existing literature on this topic finds that even if they do want to protest, there are fewer conventional mobilizing structures for migrant workers to use to overcome collective action problems (Chan and Pun, 2009). For example, Palmer et al. (2011) study the work-unit-based communities that have been established in urban areas since pre-reform period (now called ‘Urban Residents Committees’). They find that, due to the social marginalization that they face in urban areas, few migrant workers have participated in such urban community organizations. Chan (2012) notes that while the number of labor NGOs is increasing in China, these organizations have to handle labor issues very carefully since Chinese authorities are very sensitive to the prospects for strikes and protests among workers. Local authorities can easily make things difficult for NGOs by doing such things as terminating a group's lease on office space or even hiring local gangsters to attack activist workers. Therefore, many of these NGOs offer legal advice only; they avoid organizing workers for any form of collective action. Trade unions do not have any substantial operational power other than to meet the requirements set out by local authorities (see Chan, 2012: 13 for details). In sum, there are no powerful organizations to safeguard migrant workers' rights (Gui et al., 2012). This factor could further impede their ability, as a group, to overcome free rider problems. Existing research also points out that migrant worker protests are mainly factory-based and ‘each strike was fought and settled on its own’ (Pun et al., 2010; Butollo and ten Brink, 2012: 434). How do we effectively measure the ‘dormitory labor regime,’ which is considered the main mechanism for migrant workers protest mobilization (Pun and Chan, 2013), and apply it to quantitative studies? The information about whether or not the respondent migrant worker is living in a dormitory environment and how long, the size and type of the dormitory, and the number of their residents in the dormitory is not available from existing nation-wide survey data. Further, among migrant workers, protest participation may vary between those who are employed in foreign-owned or Chinese-owned manufacturing firms and those who are employed in the service sector and construction. The latter are more difficult to mobilize and more easily replaced, should they participate in protests.

One question that deserves further examination is what impact a continued increase in the number of migrant workers will have on the frequency of protests and the likelihood of registered urban workers participating in them. To date, demand for labor in the urban economy has been growing, but if the pace of job expansion has declined as a result of COVID shutdowns, for instance, and greater competition in export markets for some Chinese-made goods. Registered urban workers could face even more deterioration in wage and working conditions that have motivated their participation in protests in the recent past. Shock events such as shutdowns due to the COVID pandemic have caused temporary declines in production in China's industrial economy. Will the resulting layoffs be confined exclusively to migrant workers, or will resident workers feel the pain as well? And as industrial production resumes with the anticipated waning of the pandemic, will employers prefer resident workers, or will they see this as an opportunity to hire more migrant workers at lower wages than the resident workers they laid off?

Another focal point is if China were to dismantle the hukou system, would the number and scale of migrant worker protests increase, given that cost of protest participation for migrant workers would decrease accordingly? Dismantling the hukou system would eliminate one major factor that makes migrant workers less likely than registered urban workers to participate in protests. By increasing the size of the pool of potential participants, dismantling the hukou system could make it easier for political entrepreneurs to overcome collective action problems and mobilize enough workers to stage a protest event. Increasing repression of protests under the Xi administration would represent a check on this dynamic. Were the hukou system to be eliminated or relaxed, our theory suggests that we would expect more protest participation by migrant workers, and this increase would be most likely among those employed in foreign- and privately owned manufacturing enterprises. With appropriate survey data, these questions can be explored in future research.