This study traces the evolution of systemic state-sponsored coercive labor in the cotton harvest in China’s northwestern Xinjiang Uyghur Autonomous Region (XUAR). The recent situation in the XUAR is compared to Uzbekistan, which implemented forced labor in cotton picking until 2021. Both regions create structurally coercive labor environments through a centralized authoritarian state apparatus that deploys human resource–intensive local grassroots mobilization efforts. The article finds that while both regional entities’ coercive labor dynamics are in many ways comparable, the resulting labor settings are not easily captured through static standard measures such as the ILO forced labor indicators. Instead, state-sponsored forced labor is characterized by pervasive state-induced and systemic dynamics of coercion that are deeply embedded within sociocultural contexts. Whereas Uzbekistan’s coercive labor practices were primarily driven by economic considerations, Xinjiang’s labor transfer program pursues some economic aims but is predominantly designed to achieve Beijing’s wider ethnopolitical goals in the region.
Introduction
In 1930, the International Labor Organization (ILO) adopted the Forced Labor Convention (29) that generally targeted practices of forced labor, especially in response to forced labor in western colonies in the 1920s (Maul, 2007, pp. 478–480). However, in a nod to colonial rule, the Convention created vague exceptions for “any work or service which forms part of…normal civic obligations,” including “minor communal services” (ILO, 1930). In 1957, the ILO adopted Convention 105 to close unaddressed loopholes, specifically targeting forced labor imposed by states for the purpose of furthering political goals, including the use of forced labor for economic development, as a means of racial or religious discrimination, or to punish dissent (ILO, n.d.). These clauses sought to counter states’ excuses for coercive labor practices associated with “development dictatorships” where coercive labor was justified with promises of socioeconomic development and technocratic modernization (Maul, 2007, p. 484).
LeBaron and Phillips (2019) argue that analyses of the role of states in coercive labor practices remain a weak link in the literature. States are often just “lurking in the background” as they are “considered predominantly as the vehicles for responses to forced labour, rather than as actors who play a causal role in shaping the conditions that give rise to it” (LeBaron & Phillips, 2019, p. 2). Scholarship’s preoccupation with the relationship between capitalism and unfree labor promoted a “certain invisibility for the state and state actors” as direct perpetrators of such unfreedom (p. 2). LeBaron and Phillips’s framework explores how states promote conditions that increase vulnerabilities to forms of unfree labor. This represents an important first step toward conceptualizing the role of states in forced labor. However, understanding the situation in Uzbekistan and Xinjiang requires more specific framing for conceptualizing systemic state-sponsored forced labor, given that these places outright initiated whole-of-government and whole-of-society approaches to herd large numbers of citizens into coercive work.
Conversely, studies of state-led prison labor schemes, such as the Soviet Gulag or the People’s Republic of China’s (PRC) laogai (“reform through labor”) system, are similarly ill-suited for analyzing state-sponsored coercive labor that does not occur in prison or internment settings (although in Xinjiang the threat of internment in re-education camps looms large). In both Xinjiang and Uzbekistan, coercive labor mobilization for cotton picking occurs outside of internment contexts, even though Xinjiang still operates some cotton-producing laogai prison farms (distinct from re-education camps). In both places, coercion is instead primarily predicated on overall coercive labor recruitment environments. State-sponsored forced labor of this kind, occurring within broadly coercive environments under authoritarian governments, is a comparatively neglected topic in the literature on modern slavery, and conceptual analyses of this form of forced labor are even sparser (an exception is, for example, McGuire and Laaser, 2021).
In Xinjiang, labor coercion accelerated after 2016 in the context of a campaign of mass internment and other highly oppressive policies designed to coercively assimilate and re-engineer ethnic (non-Han Chinese) communities (Roberts, 2020; Zenz, 2018, 2021b, 2022a; Zenz & Leibold, 2019). In the Uzbek context, McGuire and Laaser (2021) aptly describe the resulting coercive environment as “structurally forced consent,” defined as the “absence of freedom to actively decide for or against taking part in the cotton industry” (p. 560). “Structurally forced consent” is embedded and enacted from within the sociocultural context and key community institutions incentivized by the state (McGuire & Laaser, 2021, p. 560).
ILO conceptualizations of forced labor have been critiqued as dealing with the issue “in isolation from the wider political, economic and social context” (McGuire & Laaser, 2021, p. 555). Overly binary conceptions of forced labor overlook how coercion may be constituted through “a variety—perhaps a continuum—of forms and degrees of unfreedom” (Phillips, 2013, p. 177). A helpful framing here is Hickey and du Toit’s (2007) concept of “adverse incorporation,” where vulnerabilities and short-term needs of the poor are utilized to create exploitative labor relationships (pp. 4–5). Adverse incorporation may meet workers’ immediate material needs but perpetuates structural poverty; it may be framed as a “choice,” but one made in the context of unfreedoms (Phillips, 2013, p. 176).
In the literature, unfree labor has also been analyzed at the corporate level and been framed as an exploitative company management technique (Crane, 2013). While companies participating in state-sponsored forced labor frequently employ such techniques, I argue that the unique dynamics of state-sponsored forced labor in Xinjiang and Uzbekistan are best understood by analyzing state mobilization efforts that engineer “structurally forced consent.” This framing then also facilitates contextualized understandings of coercive dynamics in companies co-opted and incentivized by government forced labor schemes. Moreover, Xinjiang’s coercive labor policies are comparatively well documented, whereas there is essentially no ethnographic evidence on the conditions of Uyghur workers in cotton fields in the late 2010s, given the impossibility of conducting fieldwork there.
This study compares Xinjiang with Uzbekistan as two major centers of cotton production that rely (or relied) on large-scale coercive mobilization. (Comparable contexts such as Turkmenistan are smaller and less well documented.) It argues that Xinjiang’s ongoing and Uzbekistan’s prior coercive labor dynamics are largely comparable, both based on centralized authoritarian state apparatuses deploying coercive grassroots mobilization techniques through governance mechanisms that retain substantial features of their respective socialist eras. However, whereas Uzbekistan’s coercive labor practices were primarily driven by economic considerations, Xinjiang’s labor mobilization program pursues predominantly political goals.
In Xinjiang, both the overall oppressive context and threats of sanctions for refusal to comply with state work requests are significantly worse than in Uzbekistan. Uyghur resistance to state policies and ethnic discrimination is framed as a threat to national security resulting from alleged religious “extremism” (Roberts, 2020; Zenz 2021b, 2022a). Xinjiang’s capabilities to create structurally forced consent by identifying and punishing deviant behavior in real time are considerably greater than Uzbekistan’s. Prior to the mass internments, the government created a security state with one of the world’s highest per-capita police ratios (Zenz & Leibold, 2019). Designed not only to uncover but also to preempt acts of deviance, Xinjiang’s extensive systems of surveillance and predictive policing pursue complete coverage “without any chinks, blind spots, or blank spaces” (Zenz & Leibold, 2019, p. 14). Through its Integrated Joint Operations Platform (IJOP), Xinjiang’s surveillance mechanisms generate automated real-time “push notifications” that trigger police investigations, often resulting in internment (Roberts, 2020; Zenz, 2020a; Zenz, 2022a, pp. 29–36). Based on past acts, social interactions, and present behavior patterns, Uyghurs are liable to be labeled “untrustworthy” by an internal rating system that assigns guilt by association and ethnic profiling (Byler, 2021; Leibold, 2020; Zenz, 2020a).
In the early 2000s, Xinjiang first introduced programs for the labor transfer of rural surplus laborers in the context of coercive poverty alleviation schemes (Zenz, 2021a, 2023b). From 2017, the region greatly increased the scope and coerciveness of transfers targeting especially Uyghurs for cotton picking (Zenz, 2020b). New evidence presented below indicates (a) a further intensification of this policy in 2019, forcing even elderly Uyghurs to pick cotton or else face “repeated…thought education” (Jinqi tuopin, 2019, p. 3); (b) a continuation of state mobilization tactics coercing Uyghurs to pick cotton in 2021/2022; and (c) mechanized harvesting does not necessarily reduce coercion, because mechanization often means Uyghur farmers are removed from their land and subjected to labor transfers into other industries. Xinjiang’s labor transfer program features its own mechanisms of coercion that operate in tandem with societywide systems of surveillance and predictive policing (Zenz, 2020b, 2021a, 2022c, 2023b). Transfers are part of wider schemes to reduce concentrations of non-Han populations (Zenz, 2021b). Since 2018, the region also coercively places camp detainees into heavily surveilled factory work (Byler, 2021; Zenz, 2019b, 2023b).
Research Methods
This article builds on the earliest study demonstrating forced labor in Xinjiang’s cotton harvest since the beginning of the recent repressive policies (Zenz, 2020b). Conceptually, it follows a distinction between two systems of coercive labor in Xinjiang first proposed by the author in 2019 (Zenz, 2019b), and subsequently adopted by the UN Special Rapporteur on contemporary forms of slavery (Obokata, 2022, p. 8) and researchers, for example, Lehr and Bechrakis (2019) and Murphy et al. (2021). Namely: (a) forced labor assignments for re-education camp detainees; and (b) poverty alleviation through labor transfer (Ch. zhuanyi jiuye tuopin), a program that exists in a broadly similar but typically much less-coercive form in many Chinese regions. This approach is used here to compare Xinjiang’s and Uzbekistan’s systems of non-prison state-sponsored coercive labor, focusing primarily on Xinjiang as the less-documented and more urgent contemporary context.
The transfer of Uyghurs into picking cotton operates through seasonal labor transfers and not camp-linked labor, whereas the subsequent processing of cotton involves both systems (besides an unknown but likely smaller amount of prison labor). Before 2017, most cotton pickers in Xinjiang were Han migrants from other provinces. Given that Xinjiang produces a comparatively small share of global textiles and garments but over 20% of global raw cotton, the primary driver of labor coercion in cotton production is labor transfer policies and not internment camps (labor transfer employment also likely predominates in textile and garment production). Consequently, reports that frame recent coercion in Xinjiang’s cotton production primarily in relation to camps or prisons, without referencing foundational policy frameworks behind labor transfer coercion, are misleading (for example, Daubenberger & Guckelsberger, 2022).
In 2017 and 2018, I developed an investigative policy analysis approach to elucidate surreptitious state policies (Zenz, 2018; compare 2019a). I used this approach to first document the campaign of mass internment by (a) identifying the genesis and usage of key terms and (b) ascertaining the development of related policy frameworks based on triangulated documentary evidence at different administrative levels (from autonomous region to counties and townships). Having established basic conceptual cornerstones, the research then proceeded to assess policy implementation through a triangulating analysis of diverse related sources (state reports, budgets, procurement bids, propaganda stories) in connection with witness accounts. I adapted this methodological approach in mid-2019 for the earliest study of Xinjiang’s forced labor programs (Zenz, 2019b),1 and later expanded it to include a six-phase framework for evaluating coercion in labor transfers (Zenz, 2021a; see Table 1 later in article).
Identifying Systemic Coercion through Six Phases
Phase . | Uzbekistan . | Xinjiang . | General Risk Indicators . |
---|---|---|---|
1. Identification of labor needs or of employment opportunities | State agencies calculate labor needs based on national production mandates and assign recruitment quotas to local entities. | Surplus labor and cotton-growing regions jointly identify labor needs and plan labor mobilization. Regions use seasonal labor to fulfill mandatory labor transfer quotas, achieve Beijing’s poverty alleviation goals, and transform ethnic communities. | State policies or discourses that seek to change a targeted population’s livelihoods in tandem with state employment campaigns. Local recruitment quotas for state-directed employment programs, assigned to local governments or community-level institutions. Policies that match workers to corporate or state labor needs. Policies that systematically identify state-mandated vocational training needs among targeted populations. Potentially applicable ILO indicators: N/A |
2. Recruitment | Community institutions conduct recruitment campaigns involving intense psychological pressure and threats. | Recruitment based on databases and intrusive door-to-door campaigns, involving intense psychological pressure (“thought education”) and latent or overt threats, including potential internment. Supported by a state-induced ideological environment requiring target groups to not be “lazy.” | Concerted grassroots mobilization efforts such as door-to-door campaigns, involving the state or community institutions acting on behalf of the state. Pressure campaigns using peer pressure on those resisting recruitment. State labor assignments have to be met unless locals “buy” themselves out of it (or find replacements). State efforts to define labor assignments as a matter of patriotic duty or social obligation (and refusal to accept them as sanctionable). Potentially applicable ILO indicators: (1) abuse of vulnerability, (2) deception, (3) restriction of movement, (6) intimidation and threats |
3. Training | N/A | Ideological indoctrination, “thought education,” and training in legal obligations and rights. | Mandatory vocational training programs. State-determined training quotas. Vocational training programs with state-led mandates to instill “work discipline.” Potentially applicable ILO indicators: (2) deception, (3) restriction of movement, (6) intimidation and threats |
4. Transfer to work destination | Accompanied by officials or work supervisors, potentially also law enforcement. | Centralized state-organized transfers, accompanied by officials and law enforcement. | Centralized and/or state-directed collective transfers of workers in groups. Transfers supervised by officials, security personnel, or other authority figures (such as employers). Workers cannot choose how to get to work destinations. Potentially applicable ILO indicators: (2) deception, (3) restriction of movement, (6) intimidation and threats |
5. Worker management | Supervision by officials or pickers’ employers, potentially law enforcement. Failure to achieve quotas results in verbal or physical abuse. Pressured until harvest is completed. | Supervision by accompanying officials, in some cases law enforcement. Additional surveillance and indoctrination visits by village-based work teams to evaluate their “state of mind” and “motivate” them to complete the harvest. | Work environments: With on-site supervision of workers by officials or security personnel. Designed to coerce workers to meet company or state targets. With state-mandated training at work, especially of an ideological, political, or assimilatory nature. Designed to collectively employ members of a group targeted by state-directed labor programs or policies. Where companies have been incentivized (e.g., financially) or commandeered by the state in the context of state-mandated labor policies or programs. Potentially applicable ILO indicators: (1) abuse of vulnerability, (2) deception, (3) restriction of movement, (5) physical and sexual violence, (6) intimidation and threats, (10) abusive working and living conditions, (11) excessive overtime |
6. Worker retention | Supervision in the fields, pressure on pickers’ employers. | Monitoring and surveillance in the fields. Intense police surveillance throughout Xinjiang, restricting freedom of movement. | Workers cannot freely leave workplaces or are pressured to remain. On-site supervision of workers by officials, security personnel, or third parties acting on behalf of the state. Work environments with exit/entry barriers, security and surveillance systems that restrict movement. Concentrated work settings such as factory parks with comprehensive on-site facilities (dormitories, childcare, etc.). Remote/isolated work settings that lack easy or affordable transportation. Societywide surveillance and security mechanisms that restrict movement. Potentially applicable ILO indicators: (3) restriction of movement, (4) isolation, (5) physical and sexual violence, (6) intimidation and threats, (7) retention of identity documents, (8) withholding of wages, (9) debt bondage |
Phase . | Uzbekistan . | Xinjiang . | General Risk Indicators . |
---|---|---|---|
1. Identification of labor needs or of employment opportunities | State agencies calculate labor needs based on national production mandates and assign recruitment quotas to local entities. | Surplus labor and cotton-growing regions jointly identify labor needs and plan labor mobilization. Regions use seasonal labor to fulfill mandatory labor transfer quotas, achieve Beijing’s poverty alleviation goals, and transform ethnic communities. | State policies or discourses that seek to change a targeted population’s livelihoods in tandem with state employment campaigns. Local recruitment quotas for state-directed employment programs, assigned to local governments or community-level institutions. Policies that match workers to corporate or state labor needs. Policies that systematically identify state-mandated vocational training needs among targeted populations. Potentially applicable ILO indicators: N/A |
2. Recruitment | Community institutions conduct recruitment campaigns involving intense psychological pressure and threats. | Recruitment based on databases and intrusive door-to-door campaigns, involving intense psychological pressure (“thought education”) and latent or overt threats, including potential internment. Supported by a state-induced ideological environment requiring target groups to not be “lazy.” | Concerted grassroots mobilization efforts such as door-to-door campaigns, involving the state or community institutions acting on behalf of the state. Pressure campaigns using peer pressure on those resisting recruitment. State labor assignments have to be met unless locals “buy” themselves out of it (or find replacements). State efforts to define labor assignments as a matter of patriotic duty or social obligation (and refusal to accept them as sanctionable). Potentially applicable ILO indicators: (1) abuse of vulnerability, (2) deception, (3) restriction of movement, (6) intimidation and threats |
3. Training | N/A | Ideological indoctrination, “thought education,” and training in legal obligations and rights. | Mandatory vocational training programs. State-determined training quotas. Vocational training programs with state-led mandates to instill “work discipline.” Potentially applicable ILO indicators: (2) deception, (3) restriction of movement, (6) intimidation and threats |
4. Transfer to work destination | Accompanied by officials or work supervisors, potentially also law enforcement. | Centralized state-organized transfers, accompanied by officials and law enforcement. | Centralized and/or state-directed collective transfers of workers in groups. Transfers supervised by officials, security personnel, or other authority figures (such as employers). Workers cannot choose how to get to work destinations. Potentially applicable ILO indicators: (2) deception, (3) restriction of movement, (6) intimidation and threats |
5. Worker management | Supervision by officials or pickers’ employers, potentially law enforcement. Failure to achieve quotas results in verbal or physical abuse. Pressured until harvest is completed. | Supervision by accompanying officials, in some cases law enforcement. Additional surveillance and indoctrination visits by village-based work teams to evaluate their “state of mind” and “motivate” them to complete the harvest. | Work environments: With on-site supervision of workers by officials or security personnel. Designed to coerce workers to meet company or state targets. With state-mandated training at work, especially of an ideological, political, or assimilatory nature. Designed to collectively employ members of a group targeted by state-directed labor programs or policies. Where companies have been incentivized (e.g., financially) or commandeered by the state in the context of state-mandated labor policies or programs. Potentially applicable ILO indicators: (1) abuse of vulnerability, (2) deception, (3) restriction of movement, (5) physical and sexual violence, (6) intimidation and threats, (10) abusive working and living conditions, (11) excessive overtime |
6. Worker retention | Supervision in the fields, pressure on pickers’ employers. | Monitoring and surveillance in the fields. Intense police surveillance throughout Xinjiang, restricting freedom of movement. | Workers cannot freely leave workplaces or are pressured to remain. On-site supervision of workers by officials, security personnel, or third parties acting on behalf of the state. Work environments with exit/entry barriers, security and surveillance systems that restrict movement. Concentrated work settings such as factory parks with comprehensive on-site facilities (dormitories, childcare, etc.). Remote/isolated work settings that lack easy or affordable transportation. Societywide surveillance and security mechanisms that restrict movement. Potentially applicable ILO indicators: (3) restriction of movement, (4) isolation, (5) physical and sexual violence, (6) intimidation and threats, (7) retention of identity documents, (8) withholding of wages, (9) debt bondage |
Note: ILO indicators are listed with indicator numbering in parentheses.
Building on this methodological, conceptual, and empirical foundation, this article further adopts my more recent method of examining the evolution of the region’s forced labor schemes through the lens of authoritative macro-level planning documents, including Five-Year Plans (Zenz, 2022c, 2023b). The material analyzed here also includes key political demands of central government leaders contained in the so-called “Xinjiang Papers,” internal state documents that were leaked to the Uyghur Tribunal (Zenz, 2021c). This article additionally discusses important new evidence from internal government documents contained in the “Xinjiang QQ Files,” which I obtained from unsecured local government document caches stored within the Tencent QQ network (Zenz, 2022a); and the “Xinjiang Police Files,” which were obtained by an anonymous third party by hacking into XUAR police computers (Zenz, 2022a, 2022b).
Overall, the analysis triangulates different types of documentary sources across administrative levels to analyze the coercive dynamics underpinning Xinjiang’s cotton production, including state planning and policy documents, implementation reports, local media and propaganda accounts, and Chinese academic studies. This form of multilevel policy analysis is facilitated by “politically binding standard phrases” (tifa) that connect policies across administrative levels and along stages of the policy cycle, from deliberation to implementation (Heilmann, 2017, p. 313). While this article examines the coherence between higher-level policy directives and their local implementation by evaluating both dimensions, in the specific Xinjiang context such discrepancies tend to be narrower than in other Chinese regions, a phenomenon described as a “crisis mode” of governance that is common in politically sensitive contexts (Heilmann, 2017, p. 161).
The article first traces the evolution of coercive labor in cotton picking in Uzbekistan and Xinjiang. Second, it demonstrates how labor transfers into cotton picking in Xinjiang became more coercive from 2017. Third, it shows how labor transfers for manual picking have continued despite recent increases in mechanized harvesting. Fourth, it concludes with an overview of the key elements that underpin state-sponsored forced labor and presents an alternative evaluation framework for such labor besides the ILO indicators (which had been mainly designed to measure company-based, not state-imposed, forced labor).
Cotton Production in Uzbekistan and Xinjiang in the 1990s
In the 1930 and 1940s, cotton production in the Soviet-ruled Uzbek SSR relied to a significant extent on forced labor from more than 100,000 ethnic minorities that the Soviets relocated from their homelands to prison labor camps (Pohl, 2007). From the 1950s, women and children in particular were coerced into harvesting cotton as their labor was inexpensive (Keller, 2015). Uzbekistan produced more than 60% of all Soviet cotton (Schmitz, 2020, p. 9). A rapidly growing rural population supplied ample labor, rendering mechanization unattractive in a corrupt system that effectively disincentivized modernization. In the 1990s and 2000s, systemic stress created by the post-Soviet transition further exacerbated economic pressures, leading to declining mechanization rates and increased use of inexpensive coercive labor (Bhat, 2013; Cannell, 2007). In 2000, an estimated 1.4 million or 22.6% of Uzbek children aged 5 to 14 were subjected to cotton picking (Cannell, 2007, p. 217). An ILO analysis of adult pickers aged 18 to 50 during the 2014 and 2015 harvests found that 21%–24% of that age bracket (3 million persons) were involved in this work, nearly three-quarters of them rural women (ILO, 2017, p. x). Since 2012, the focus of coercion has shifted from forced child to forced adult labor, but the overall motivation behind the system has remained similar, as profits from cotton production have enriched corrupt political elites (Foley, 2016).2
As for Xinjiang, in the early 1990s, in the context of decades-long efforts to solidify state control through settler colonialism in this strategic frontier region, the Chinese government massively expanded its infrastructure and other investments into the region, turning it into China’s main cotton-producing region. In 1996, Xinjiang’s party secretary Wang Lequan announced that the Uyghur heartland of southern Xinjiang was to become the “largest cotton-producing area in the country” (Seymour & Anderson, 1999, p. 103). Xinjiang’s economic development and cotton production was in part boosted by Beijing’s investment in the Xinjiang Production and Construction Corps (XPCC), an entity established in 1954 as a military-agricultural colony that facilitates large-scale Han in-migration (Seymour & Anderson, 1999; Cliff, 2009). The XPCC operates a large network of prisons and laogai labor camps that in the 1980s and 1990s absorbed tens of thousands of prisoners from eastern China. At its peak in the mid-1990s, an estimated 2.4% of Xinjiang’s cotton production derived from laogai labor (Seymour & Anderson, 1999, pp. 104–105).
Seymour and Anderson (1999) argue that after 1994, declining prisoner numbers would have reduced the share of laogai labor in Xinjiang’s agricultural production (p. 105). From 1994, the region began to annually mobilize over a million schoolchildren starting from grade three for the cotton harvest to address the “problem of insufficient local labor” (China News, 2017). Meriyem Sultan (2021), a Uyghur informant from Aqsu prefecture who picked cotton between 1998 and 2004 (during middle and high school), reported that they were assigned daily quotas. At times, they were sent to fields operated by the Tarim laogai, working adjacent to prisoners who were guarded by police. She and other students were beaten when they didn’t fulfill quotas. She personally witnessed the death of a schoolmate who collapsed due to a known heart condition (which did not exempt him from picking cotton) during a poorly ventilated and crowded bus ride to the cotton fields. From 2006, primary (but not secondary) school students were exempted from cotton picking. Between 2001 and 2021, Xinjiang’s share of China cotton production grew from 27.4% to 89.5% and, in 2021, made up around 20% of the global cotton supply (PRC government, 2021b; Murphy et al., 2021; Wang, 2015, p. 29; Zenz, 2020b).
Coercive Labor in Uzbekistan’s Cotton Harvest Until 2020
Uzbekistan’s state-mandated cotton production quota system constituted a main driver of its coercive labor system. Soon after its abolishment during reforms initiated by President Mirziyoyev in 2020, the country’s forced labor problem subsided (Keller, 2015; RFERL, 2020). Before then, the state controlled every aspect of cotton production, reinforced by its repressive apparatus (McGuire & Laaser, 2021, p. 556). The system was instituted during Soviet rule and continued after independence under President Karimov. Karimov’s gradualist post-socialist reforms were designed to retain state control over cotton production, intentionally continuing key elements of the Soviet-era planned economy while additionally leveraging newly-created local institutions (farmers associations) to enforce government crop production plans among smallholders (Kandiyoti, 2007, pp. 33–35; Schmitz, 2020, p. 9).
National production quotas were handed down to regional officials, who in turn pressured subordinate entities to recruit commensurate numbers of pickers (ILO, 2017, p. xi; Kandiyoti, 1998, p. 569; UGF, 2018, p. 21). Bilal Bhat (2013) speaks of “a wider hierarchy of mobilization in which a national target for the harvest is broken down into quotas which are then enforced at each level of the state administration right down to the individual quotas for a [picker] in the field” (p. 80). This took place through so-called hokims, heads of local executive authorities that are broadly equivalent to local governors (cf. ILO, 2017, p. xi; McGuire & Laaser, 2021, p. 560). Central government documents mandated the creation of “cotton command units” under regional and district hokimiats (administrative bodies; UGF, 2018, p. 27). Informants recalled nightly meetings where hokims issued threats against their district’s institutions when quotas were not met (UFHR, 2019, p. 29). Hokims in turn used the leaders of mahallas (residential community associations) to recruit assigned worker numbers (UFHR, 2019, p. 15). Mahallas possess considerable power since they control the disbursement of social benefits and exert social control, including through auxiliary law enforcement functions (Kandiyoti, 2007, pp. 34–36). Hokims also coerced public organizations to send pickers, and the system further relied on leaders of picking brigades and private employment agencies paid by the state. Those responsible for mobilization often had direct control over the employment or financial well-being of the people they recruited, and fear of losing one’s job was a common reason for compliance amid high employment (McGuire & Laaser 2021, p. 562; UGF, 2018, p. 1). The same officials who supervised people during their regular work often recruited and monitored them in the fields (UGF, 2018, p. 35).
Not surprisingly, surveys identified “a strong correlation between method of recruitment…and various forms of labor exploitation” (UFHR, 2019). Mahallas were tasked with mobilizing able-bodied citizens and formed teams of officials and public sector employees to “compile lists of unemployed people and go door to door to recruit people to pick cotton” (UGF, 2018, p. 28). Teams conducted daily visits to those who failed to participate. Local leaders in charge of disbursing social benefits were pressured to recruit pickers and created blacklists of those who refused, barring them from receiving state benefits or services (UGF, 2018, p. 37; UFHR, 2019, p. 7). Pickers were given large daily quotas, worked 12- to 13-hour days without weekends off for low pay, and faced threats of violence, verbal abuse, and fines when missing their quotas (McGuire & Laaser, 2021, pp. 563–566). Because the primary objective was not full employment but completing the harvest, citizens could pay for replacement pickers (ILO, 2017, p. xi; UGF, 2018, p. 1).
The Uzbek recruitment system created an atmosphere of coercion where the latent threat of consequences often sufficed to enforce compliance. A local business owner remarked that even in the absence of overt threats, businesses would comply with state requests to send workers to the fields: “If we do not fulfill their requests, they [government entities] will surely cause problems for us. The tax inspection would come and find some wrongdoings or violations” (McGuire & Laaser, 2021, p. 561). In 2014, only about 15% of those called to pick cotton refused, and even in 2019, when efforts to abolish coercion had made significant progress, 48.3% of respondents replied “no” to the question “Could you refuse to pick cotton?” (ILO, 2017, p. xi; UFHR, 2019, p. 21). Most rejectors in the 2014 ILO survey were already unemployed and could therefore not be penalized with loss of employment (ILO, 2017, p. xi). An interviewed mahalla recruiter admitted that she tells pickers that “the work is voluntary,” however, “refusal is not an option” (ILO, 2017, p. 19). Mahalla recruiters were at times accompanied by police officers, who also often supervised the transport of pickers to the fields. As noted by outside observers, “to a population with a deep and well-founded fear of law enforcement, law enforcement presence reinforces the message that cotton picking is mandatory” (UGF cited in McGuire & Laaser, 2021, p. 562).
After the start of a 13-year international boycott of Uzbek cotton in 2009, state harassment of independent monitors and systemic coercion remained endemic until their gradual elimination by 2021 (Imamova, 2022).
Xinjiang’s Labor Provision Strategies in the 2000s
Between 1990 and 2007, driven by China’s strong economic growth in the east and a massive state-led campaign to “Open Up the West” under Jiang Zemin, Xinjiang’s planted cotton area quadrupled, from 0.42 to 1.66 million hectares (Guan, 2008). By 2008, the region needed more than 1 million pickers annually (Bao & Li, 2008, p. 15). Finding labor for this grueling and comparatively poorly paid work remained a persistent problem despite rural underemployment. In the late 1990s, some of the crop could not be picked in time despite attracting 200,000 pickers from other provinces (Su et al., 1998, p. 65). Therefore, the state embarked on a two-pronged strategy.
From 2003, the XPCC in particular pursued a proactive, large-scale “organized” recruitment and transfer of Chinese migrant pickers on special trains, organized through close “labor service cooperation relationships” with respective local state agencies in other provinces (Bao & Li, 2008, p. 17; China News, 2014).These were on top of pickers who arrived in Xinjiang through other individual or private channels. In 2008, the XPCC had 686,000 pickers, 228,000 from Xinjiang and 458,000 (approximately two-thirds) from other provinces (Bao & Li, 2008, p. 15).
From the 2000s, Xinjiang began to promote the transfer of rural surplus laborers (Ch. nongcun fuyu laodongli zhuanyi jiuye; Zenz, 2021a, 2023b). In the classic dual-sector or Lewis-Ranis-Fei model of economic development, surplus labor is defined as “labor [that] can be transferred out of the traditional [agricultural] sector without reducing the volume of farm output” (Cook, 1999, p. 18). After decollectivization in the 1980s, China’s impoverished rural populations that had been trapped in low-productivity agriculture through China’s rural-urban dual system transferred themselves into secondary and tertiary sector employment (Chan & Wei, 2019). However, in western regions including Xinjiang, ethnic discrimination, language barriers, and strong linkages between communities, land and ethnic identity rendered non-Han populations reticent to leave their land, especially to enter relationships of economic dependence with the Han (Cliff, 2009; Li, Ma & Sun, 2017; Roberts, 2020). In 2003, amid ongoing high rural ethnic population growth, Xinjiang mobilized 130,000 local peasants into cotton picking (People’s Net, 2003). However, Han migrants continued to predominate. By 2016 locals (mostly from southern Xinjiang) still constituted only 100,000 of 300,000 XPCC cotton pickers (XJASS, 2019). In the wake of increased mechanization, and the social instability and aggressive securitization following the 2009 Ürümchi riots that significantly impacted Uyghur labor transfers to non-Uyghur regions, mobilizing Uyghurs into cotton picking may not have been a primary concern (Li, Ma & Sun, 2017, p. 36; Zenz & Leibold, 2019).
Political Mandates and the Promotion of Poverty Alleviation Through Labor Transfer in Xinjiang
In 2014, the central leadership in Beijing reoriented its policies for the XUAR from a primary focus on economic development to de-extremification and stability maintenance (Zenz, 2021c, 2022b). Between 2014 and 2019, and especially concurrent with the campaign of mass internment in 2017, employment creation became a major political priority and its enforcement was increasingly coercive.
In confidential speeches, General Secretary Xi Jinping stated that large numbers of unemployed persons are liable to “provoke trouble,” and that company employment is “conducive to ethnic interaction, exchanges and blending,” making ethnic groups “imperceptibly study Chinese culture,” and therefore helping them to “resist religious extremist thinking” (Central Office Bulletin, 2014, p. 20). At the same event, Yu Zhengsheng, then head of the Central Committee Xinjiang Work Coordination Small Group, argued that
a series of supporting policies and requirements have been put forward for the development of the textile and garment industry, which is to drive at least one million people into employment. This matter is of vital importance to Xinjiang’s social stability and long-term stability. (Central Office Bulletin, 2014, p. 72)
By 2017, these arguments had deeply impacted policy priorities. Following Xi Jinping, Kashgar prefecture’s party secretary Li Ningping exhorted officials that poverty alleviation is closely intertwined with counterterrorism and social stability:
Over 90 percent of illegal religious activities take place in rural areas, over 90 percent of all violent terrorists live in rural areas.…The reasons why rural areas account for such large proportions are inseparable from poverty.…Some people have nothing to do for a long time, no jobs to work, they idle about, and make trouble out of nothing. They are easily instigated and seduced into wrongdoing by the “three evil forces.”…If we do a good job in the work of alleviating deep poverty and guide people of all ethnic groups to live a modern and civilized life, it will be a powerful hedge against religious extremism.…Therefore…doing a good job of alleviating deep poverty is not only a livelihood issue, but also a [social] stability issue; [it is] an economic issue as well as a political issue. (Kashgar Party Office Bulletin, 2017, p. 6)
In December 2017, Huafu Corporation, which operates the world’s largest textile mill in Xinjiang, showed photos of 600 transferred Uyghur laborers in military fatigues who were being subjected to military drill and thought education, noting that
due to a lack of information, a lack of courage, and a fear of going out, large numbers of rural surplus laborers are idle at home, which increases the burden on their families and brings hidden dangers to public security. (Huafu Corporation, 2017)
Xinjiang’s February 2016 Opinion on implementing the central government’s mandate to “Win the Battle Against Poverty” specified the use of village-based work groups (officials going from door to door) to implement Targeted Poverty Alleviation (jingzhun fupin) work and warned that teams would not be withdrawn from villages until poverty alleviation targets had been achieved (China Network, 2016). The battle against poverty would involve “stimulating the inner motivation of the masses,” while “curing poverty means to first cure ignorance and backwardness” (China Network, 2016). Some 1.74 million persons would be freed from poverty through increased industrial production and labor transfers. Chinese academic research found that substantial numbers of Uyghurs resisted such transfers even when offered adequate financial remuneration and free housing (Deng, Mamati & Wang, 2016, p. 84). Academics and officials associated such unwillingness with “closed-mindedness” and religious “extremism,” arguing that labor transfers can “crack open the solidified society in southern Xinjiang” (Lu & Guo, 2017, p. 194).
The region’s June 2016 Special Action Plan for implementing the February 2016 Opinion specified a comprehensive approach to labor transfers, including state-led transfers during agricultural “slack periods” into seasonal work such as cotton picking (Laodong fagui, 2016). In September 2016, Xinjiang issued a notice about the state management of seasonal workers, including cotton pickers, as part of implementing the June 2016 Special Action Plan (Shangye xinzhi, 2016). It prescribed a process of close supervision and intensified indoctrination of these workers in collaboration with public security agencies, including “thought education.” All counties were charged with proactively inquiring about labor needs from XPCC cotton planting regions and then mobilizing the required workers. In December 2016, Xinjiang announced plans to intensify the employment cooperation mechanism between the region’s north and south, expanding the scale of government-organized labor transfers, including for seasonal work (PRC government, 2016). This significantly accelerated transfers of southern Uyghur laborers to northern XPCC cotton fields.
Xinjiang’s 13th Five-Year Poverty Alleviation Plan from May 2017 stated that poor people’s “labor and employment willingness and abilities are insufficient” and repeated five times that the “inner motivation” (Ch. neisheng dongli) of locals is insufficient and must be “stimulated” (Yecheng county government, 2017). It specifically highlighted the role of seasonal labor transfers in achieving state poverty alleviation targets. In early 2017, the XUAR even officially abolished the so-called hashar, a longstanding system of coerced unpaid labor in agricultural and public works programs, with the stated aim of promoting intensified Uyghur participation in labor transfers (Emin county government, 2017).
In early 2017, the state also began to systematically replace Han migrant pickers with transferred non-Han laborers. The September 2016 Notice mandated that “under the same conditions,” XPCC recruitment of seasonal laborers should “prefer local farmers and herdsmen from Xinjiang” (Shangye xinzhi, 2016). A XUAR government report about Kashgar’s Yopurga county states that local Uyghurs’ “deep-rooted ideology of laziness” was overcome through labor transfers. It cites a Han cotton farmer as saying that due to work teams implementing transfers starting in 2017, he “no longer used outside [migrant] cotton pickers” (XUAR Transport Department, 2019). In 2016, Xinjiang-based pickers made up only one-third (about 100,000) of all pickers in XPCC cotton fields (XJASS, 2019). Han migrants also constituted most pickers in southern Xinjiang as Han cotton farmers recruited pickers from their eastern home regions (Li & Hu, 2018). In contrast, by 2018, only 2.5% of the pickers in Kashgar prefecture were migrants, while 210,900, or 84.4%, were from the same prefecture, and other Uyghur regions reported similar trends (Li & Hu, 2018; Shandong Legal Network, 2020). That year, Aqsu and Khotan (Hotan) prefectures alone sent 210,000 pickers to the XPCC (PRC Ministry of Agriculture, 2018).
Between 2016 and 2020, Xinjiang transferred an annual average of 2.87 million (Xinjiang UFWD, 2021). Seasonal transfers played a predominant role. Between 2017 and mid-2020, approximately half of all transfers in southern Xinjiang’s Khotan prefecture focused on seasonal labor such as cotton picking (PRC Ministry of Human Resources, 2020). Between 2017 and 2018, Qaraqash (Karakax) county in Khotan prefecture increased numbers of labor-transferred pickers from 40,600 to 54,000, mobilizing 15.7% of its population aged 18 to 59 years to pick cotton in other regions (Sina News, 2017, 2018). In 2018, Xinjiang is estimated to have mobilized at least 570,000 surplus laborers from predominantly Uyghur regions into cotton picking (Zenz, 2020b).
Coercive seasonal labor transfers for cotton picking have been linked to Xinjiang’s coercive Poverty Alleviation through Labor Transfer program, but not to the coercive work placement of former detainees of the so-called Vocational Skills Education and Training Centers (VSETCs), a state euphemism for re-education camps (Zenz, 2022a, 2023a). While the two systems are distinct and operate under different policy and administrative systems, Uyghurs who refuse to participate in poverty alleviation or employment policies could face internment. In 2014, the region published a list of 75 “expressions of religious extremism,” one of them being “refusal to accept government management,” a broad expression that can include poverty alleviation and employment measures (Phoenix Information, 2014).
The threat of internment does not alter the conceptual distinctions between Xinjiang’s coercive labor systems (Zenz, 2023b). Poverty Alleviation through Labor Transfers generate their own coercive pressures, independent of the camps. However, this latent threat significantly enhances such pressures.
The Yarkand Poverty Alleviation Rectification Campaign and Lists of “Lazy Persons”
Previously unpublished internal government documents from the Poverty Alleviation Work Group in Arslanbagh Township, Yarkand (Shache) county, show that state efforts to compel Uyghurs into poverty alleviation measures, including labor transfers and seasonal labor, intensified after 2018. Yarkand’s Rectification Work Notice from January 2019 mandated an increase in the “political status” of poverty alleviation work, sternly warning cadres of “severe” repercussions for not achieving mandated outcomes (Zhonggong shache, 2019, p. 3). Within its secret five-year plan designed to suppress any opposition to its policies, a plan that began in 2017 and ended in 2021, 2019 was the year when Xinjiang was supposed to achieve a state of “basic normalization” (Zenz, 2022b).
Yarkand’s July 2019 document on “Recent Key Work in Poverty Alleviation” warned officials to “prevent the problem of insufficient inner motivation among cadres and masses” (Jinqi tuopin, 2019, p. 1). It mandated that “lazy persons, drunkards, and other persons with insufficient inner motivation” would need to be subjected to “repeated…thought education” (p. 3). If this failed to produce “obvious results,” they were to be dealt with “according to the document” (the exact meaning of this is unclear, but the context suggests coercive measures). Persons over 60 years old and students were to pick crops including cotton, vegetables, tomatoes, and peppers. The names of those who “have not realized stable employment [and therefore] must [engage] in seasonal labor” were to be listed by the Poverty Alleviation Employment Office, and they were to be subjected to centralized, organized transfers into seasonal labor such as cotton picking (p. 5). The state was to organize centralized childcare for toddlers to “prevent the problem from happening that these types of persons neither have stable employment, nor are able to earn a seasonal labor income” (p. 5).
By late 2019, Yarkand county was compiling lists of “lazy persons,” “drunkards,” and persons “without sufficient inner motivation” (Qi cun, n.d.). One list labeled individuals as old as 77 years as “lazy,” and of five individuals with listed solutions for their “laziness,” two were sent to another county to pick cotton.
Xinjiang’s Labor Transfers for Cotton Picking: Local Implementation Cases
Like Uzbekistan, the coercive pressures in Xinjiang’s cotton harvest are primarily found at the recruitment stage. Village-based work teams (Ch. zhucun gongzuodui) go door-to-door collecting information on each household’s employment status and coercing villagers into labor transfers:
The factors that once prevented the local workforce from entering the cotton-picking market no longer exist. This is thanks to the continuous advancement of the work of transferring surplus labor performed by all levels of the Xinjiang government. Especially since 2014, Xinjiang has sent a total of 350,000 cadres to villages for five consecutive years to help the masses out of poverty and misery. (Li & Hu, 2018)
In a village in Payzawat county (Kashgar prefecture) where the state found that Uyghurs were “unwilling to go out to work,” officials entered every home for a second time and undertook “thought education work” until 60 persons had been mobilized into picking cotton (Jiashi county government, 2017). In September 2018, a village-based work team mobilized 345 villagers in a single village in Qaraqash county (with a total workforce of 983), which previously had very low participation in labor transfers, to pick cotton in neighboring Kashgar prefecture (Zhou, 2020). Three village cadres supervised them.
A September 2018 account from Aqsu’s Bai (Baicheng) county describes how a village-based work team “liberated” Uyghur villager Er’eli Hekim from his “serious” thought problems (Aqsu Daily, 2020). Thanks to “thought transformation work,” he then followed the government’s call to pick cotton:
In the past, my lazy thoughts of “waiting, relying, and asking” were serious. I only knew how to ask for things from the party and the government. Now, I finally understand that the happiest [thing] is to use the money I earn with my hard-working hands and my sweat. In the future, with the support of the party and the government’s policy…, I will earn more money and make my family’s life better every day.
Recruitment drives are characterized by tightly organized transfers and close supervision. A 2017 account from Kashgar prefecture notes that pickers are accompanied by officials, and “work-team cadres and police station guards regularly visit them” (Chen, 2017). A 2020 report from Aqsu prefecture states that cotton pickers are transferred to their work destinations in a “point-to-point transfer” fashion and cadres must be with them at all times:
Give full play to the front-line [cadres acting as] “instructors,” “security staff,” and “service staff.” Except under special circumstances, these must eat, live, study, and work together, actively carry out ideological education during cotton picking…and assist in solving issues related to wages or accidental injuries. (Shandong Legal Network, 2020)
These accounts also indicate that Xinjiang’s new employment policies can come with material benefits. Centralized state assistance and related policies are designed to protect laborers from exploitation through employers, improve work conditions and safety standards, increase awareness of their rights, and propagate social benefit coverage (Shangye xinzhi, 2016; Zenz, 2023b). Given that cotton picking as an economic sector has historically been associated with exploitation, these benefits should not be lightly dismissed, and they indicate how Xi Jinping’s signature goal of alleviating absolute income-based poverty constitutes an important policy goal.
Cotton pickers recruited through these state schemes are monitored by accompanying cadres and work teams. For example, in late 2019 a Kashgar law enforcement work team visited 150 Uyghurs who were picking cotton in an XPCC region (XUAR Natural Resources Department, 2019). They assessed pickers’ “state of mind” and exhorted them: “We hope you will develop a spirit of hard work and of being willing to suffer, work diligently, and finish the task of picking cotton according to the plan.”
Such accounts indicate significant similarities between the Xinjiang and Uzbekistan contexts. In both regions, pickers are (or were) coercively mobilized through extensive grassroots efforts. The result is a societal normalization of coercive labor. When Meriyem Sultan (2021) later attended university in Beijing, she sincerely expected to have to perform mandatory labor tasks:
When I asked my tutor if we have to carry out any forced labor, everyone burst out laughing. I was so relieved that finally I was at an institution where I can just focus on my studies without worrying about forced labor. (p. 4)
Beyond Picking: Two Forms of Coercive Labor in Xinjiang’s Cotton-Based Value Chain
A key difference between Xinjiang and Uzbekistan is that the latter primarily produces raw cotton, whereas in Xinjiang, state policies after 2014 drove a massive expansion of the cotton-based value chain into the production of yarn, textiles, and garments. In July 2014, Beijing announced the goal of placing at least 1 million workers into jobs in textile and garment industries by 2023, with 650,000 of them coming from the southern Uyghur majority regions (PRC government, 2014).
While current evidence of the scale of laogai labor within Xinjiang’s cotton production is limited, multiple prison farms still engage in it. Qutubi (Hutubi) county’s Fangcaohu Prison Farm’s cotton-ginning factory was still operating in June 2016 (China News, 2016; Laogai Research Foundation, 2008, p. 472). Shayar county’s Tarim Runcheng Prison Ginning Plant, linked to Shayar’s Prison Farm, was still operating in September 2017 (XJDRC, 2017, p. 5; Laogai Research Foundation, 2008, p. 470). Abdulhamit Barat, a Uyghur from Yarkand county, stated that during 18 years of imprisonment in different prisons (1995 to 2013) he was continuously forced to produce various types of garments, including gloves, sweaters, and uniforms, working 15–16 hours daily.3
On a large scale, coercive cotton picking is only linked to Poverty Alleviation through Labor Transfer. However, the subsequent production of yarn, textiles, and garments has also involved a large-scale employment of VSETC detainees that are “released” into state-designated work positions (Byler, 2021; Murphy et al., 2021; Zenz, 2019b, 2023a). In early 2018, multiple Uyghur majority regions received a preferential state policy for moving detainees into labor-intensive textile and garment production, attracting eastern Chinese companies through subsidies of 1,800 Chinese yuan per trained detainee; that year, Kashgar prefecture alone planned to employ 100,000 camp detainees, 70,000 of them in textile and garment factories (Zenz, 2019b, 2023a).
Xinjiang’s textile industry also employs coercive management measures for ethnic workers from the labor transfer system. For example, Yarkand’s Eagle Textile Company employs impoverished and surplus laborers using “semi-militarized management” (Wang, 2018; XJDRC, 2018). The Han factory owner stated that these workers initially “lacked enthusiasm for labor” and had a mindset of “waiting, relying, and asking.” After one year of “imperceptible nurturing,” however, they “understood that they should use their hard work.” The Chinese term for “imperceptible” here is identical to the one in the classified speech where Xi Jinping argued that enterprise employment makes ethnic groups “imperceptibly study Chinese culture.” Such companies play a significant role in implementing the state’s campaign of indoctrination and forced assimilation.
Xinjiang’s Mechanized Harvesting and Ongoing Labor Transfers for Manual Picking
During the past decade, the state promoted mechanization, especially in northern Xinjiang and the XPCC regions, which are better suited to mechanized harvesting than the south due to larger-scale farming and more contiguous fields. In 2019, the share of mechanized cotton-planted land in XPCC regions reached 82.9%, whereas the same share for the entire XUAR stood at only 30.2% (0.77 of 2.54 million hectares) (PRC government, 2020; China Cotton, 2020).
After the first publication of evidence of systematic forced labor in Xinjiang’s cotton harvest in December 2020 (Zenz, 2020b), the Chinese Ministry of Foreign Affairs spokesperson claimed that “the mechanization rate for cotton picking in Xinjiang reached 70 percent in 2020” (PRC Embassy Canada, 2021). However, the figures underlying this claim pertain to only to 64.5% of Xinjiang’s total cotton-planted area (China Daily, 2021; China Statistics Network, 2021). On 6 December 2021, the government then claimed a XUAR mechanization rate of 80% for 2021, but without providing the sizes of mechanized versus manually-harvested cotton areas, and without disclosing southern Xinjiang’s much lower mechanization rate (PRC government, 2021a). Less than six weeks later, on 14 January 2022, Beijing suddenly claimed a mechanization share of “over 85%” without providing any substantiation (PRC Ministry of Foreign Affairs, 2022).
In contrast, an October 2020 government report published before the December 2020 revelations quoted a mechanization share for southern Xinjiang of only 34.7%, noting that “operating conditions of large-scale machinery and equipment are poor” and “land fragmentation is high,” all factors that “seriously affect the application of agricultural mechanization” (XUAR Finance Department, 2020). In 2019, 45% of Xinjiang’s cotton fields were in these southern five prefectures.4
Satellite imagery analysis by German media from spring 2022 used a proxy indicator of harvesting speed to estimate that at least 36% of the region’s cotton was still being picked by hand (this method potentially underestimates manual harvest share because scattered small plots are more easily overlooked than large contiguous fields) (Daubenberger & Guckelsberger, 2022). In southern Xinjiang’s Kashgar region, however, the estimated manual harvest share was 96%. Southern Xinjiang’s cotton is of especially high quality, not only because hand-picked cotton is considered superior to the machine-picked variety, but also because the region grows nearly three-quarters of the especially valuable long-staple cotton.5 Chinese reports confirm that by March 2022, long-staple cotton still could not be mechanically harvested as its long, soft fibers are better preserved through manual picking (Sina News, 2022).
Despite rising mechanization rates, the speed of the expansion of Xinjiang’s cotton production drives an ongoing need for pickers. In the past two decades, the region’s sown cotton area grew fastest in the least-mechanized areas of southern Xinjiang, nearly tripling from 574,200 to 1,605,900 hectares (2000 to 2020; Liu et al., 2021). In XPCC-controlled farmland, the total area of cotton fields also grew from 498,000 hectares in 2010 to 870,000 hectares in 2021. The fact that in 2018, Aqsu and Khotan alone sent 210,000 seasonal rural surplus laborers to XPCC regions—despite the latter’s 80.4% mechanization share that year—is likely linked to such rapid increases (PRC Ministry of Agriculture, 2018; XPCC, 2019).
Xinjiang’s transfers of ethnic surplus laborers for cotton picking continued in 2021–22 despite Beijing’s claims. In October 2021, Kashgar prefecture boasted that “the entire region’s counties, townships, and villages have done a good job in mobilizing cotton-pickers, and the relevant departments have also organized the masses of the townships and villages to work for large cotton farmers…solving large-scale cotton planters’ ‘urgent worries’” (Sohu, 2021). Using dedicated mass mobilization methods, Kashgar’s Poksam county mobilized 6,000 pickers. These are not isolated cases. Xinjiang’s 14th Five-Year Social and Economic Development Plan (2021 to 2025) mandates closer cooperation between XPCC and other regions for an “enlarged” promotion of seasonal (harvest-related) agricultural labor transfers (XUAR government, 2021). An April 2022 article on Xinjiang’s employment programs and labor transfers in Tianshan, a prominent media outlet co-sponsored by the XUAR propaganda department, confirms that the “vast scale” of cotton, tomato, and other plantations in southern Xinjiang continues to “provide an abundance of short-term employment avenues” (Tianshan, 2022).
Increased mechanization does not necessarily reduce the risk of forced labor. To promote mechanization, the state aggressively promotes land transfer schemes by which local farmers transfer their land usage rights to large-scale operators in exchange for lease payments. Displaced Uyghur farmers are then often coercively transferred into labor-intensive manufacturing work.
Xinjiang’s main state media has described land transfer as a crucial measure for reducing land fragmentation and promoting mechanized cotton farming (Xinhua, 2018). By 2019, Awat county in Aqsu planned to transfer 66,667 hectares from smallholders to state cooperatives, XPCC state farms, and large corporations in order to “fully implement mechanized cotton-picking” and simultaneously “liberate the peasants from [their] land” (through labor transfers) (China Cotton Network, 2018). By 2021, Lopnur (Yuli) county had transferred 26,667 hectares from farmers to Lihua Cotton Corporation to promote “modern, mechanized” cotton production, drastically increasing mechanized harvesting (Zhu, 2021). While usage rights transfers can theoretically be reversed, Xinjiang’s policy context makes it clear that transfers form an integral part of a long-term strategy to create a large-scale modern and mechanized agriculture and permanently convert smallholder farmers into industrial workers (Zenz, 2021a). Transferred laborers who were “liberated” from their land often become cheap labor for Xinjiang’s textile and garment industries.
Comparing Coercive Labor in Xinjiang and Uzbekistan
In 2012, the ILO published a set of 11 indicators for measuring forced labor, including abuse of vulnerability, restriction of movement, isolation, intimidation, abusive work conditions, violence, debt bondage, and withholding of wages (ILO, 2012). These indicators were designed to measure forced labor in individual companies or economic sectors. They are largely unsuited to evaluating the key mechanisms that underpin state-sponsored forced labor, especially in Xinjiang, where state goals for coercive mobilization are primarily political. Some of the indicators linked to economic exploitation are applicable to Uzbekistan’s profit-motivated cotton harvest. However, widespread worker exploitation would run counter to Beijing’s political aims of long-term social stability and of achieving Xi’s signature goal of eradicating absolute poverty. Consequently, Xinjiang at least nominally seeks to improve work conditions, and for labor transfers it prioritizes political objectives over economic exploitation (the work conditions of released VSETC detainees diverge negatively from this). This does not mean that the state does not exploit cheap Uyghur labor, given that pickers receive low pay for their extremely hard work. However, labor exploitation is not the primary goal of Xinjiang’s labor transfers, and outright abysmal work conditions run counter to the long-term objectives of the state’s employment policies targeting ethnic groups.
Instead of relying on the static and decontextualized ILO indicators, state-sponsored coercive labor in both Uzbekistan and Xinjiang is best understood and measured as a systemic and contextual process. Based on a review of Xinjiang’s labor transfers, I previously developed a six-phase framework for evaluating coercion in labor transfers (Zenz, 2021a). Below, besides comparing Uzbekistan and Xinjiang, sets of new indicators are shown for each phase that are specifically designed for evaluating the risk of state-sponsored forced labor (Table 1). The comparison with potentially applicable ILO indicators shows that the ILO indicator framework largely fails to evaluate coercion during the initial labor identification, recruitment, training, and transfer phases—which are the most pertinent stages for evaluating risks associated with state-sponsored forced labor.
Overall, mechanisms of coercion in both regions are comparable, although Xinjiang’s environment is far more repressive. In both cases, state-sponsored forced labor is similar in five key respects:
a centralized authoritarian state with a strong bureaucratic apparatus that
steers economic policy and incentivizes or commandeers relevant economic actors (including state-owned and private companies);
creates a coercive social environment among targeted populations;
leverages this environment in tandem with substantial grassroots-based human resources and institutions to
develop mobilizational pressures at the grassroots level.
In both regions, vestiges of socialist planned economies play a significant role, although not in equal measure. Uzbekistan relied on a continued centrally-planned quota system. Xinjiang’s top-down state planning also shapes the economy in significant yet complex ways, influencing crop cultivation priorities, commandeering state-owned enterprises and XPCC entities according to political goals, and steering seasonal worker flows (replacing Han with Uyghur pickers). However, Xinjiang’s current form of “state capitalism” or “socialist market economy” also systematically incentivizes the private sector to further state policies, especially since the 2010s (Cliff, 2016). State-sponsored forced labor can be implemented just as easily by strong developmentalist states, using the five methods listed above, as by socialist planned economies.
One key difference between the two systems lies in their ultimate goals. In Uzbekistan, coercive labor directly financially benefited elites. Beijing’s policies in Xinjiang focus on ethnic profiling and are framed as an urgent matter of national security for the state, which has portrayed Uyghurs as a type of “biological threat” (Roberts, 2020, p. 40; Zenz, 2021b; Zenz, 2022a, pp. 38–41). Compared to traditional rural livelihoods, surveilled factory environments in China provide the state with greater control over workers, weaken traditional community structures, and curtail the intergenerational transmission of indigenous culture (Byler, 2021; Zenz, 2019b, 2021b).
Conclusions
Both Xinjiang and Uzbekistan deploy(ed) systems of state-sponsored forced labor through a highly centralized coercive state apparatus leveraging significant grassroots human resources to coercively recruit and retain cotton pickers. Initially, both heavily relied on child labor before shifting to coercive adult labor, doing so in contexts where during past eras of collectivization and centrally-planned economies employment was largely coercively assigned by the state.
In both cases, coercive labor mobilization operates (or operated) through state dominance over economic policy and the state’s ability to commandeer or incentivize relevant economic actors, together with an extensive mobilizational grassroots apparatus created during socialist eras and maintained or evolved by subsequent leaders. While Xinjiang is far more repressive, both regions heavily rely (or relied) on centralized authoritarian systems to create coercive environments penalizing noncompliance. In Xinjiang, this includes the threat of internment and the detection of deviance through automated systems of preventative policing. Without legal recourse, citizens “choose” to comply rather than face potential ramifications within a system that leaves them powerless.
The resulting environments of “structurally forced consent” (McGuire & Laaser, 2021) are not necessarily immediately observable to outsiders, and may be challenging to assess through conventional means such as the ILO’s forced labor indicator framework, which was not designed to evaluate state-sponsored forced labor. State-sponsored forced labor is not readily detected by examining individual workers, especially in highly repressive environments such as Xinjiang where they cannot speak freely, or even at individual workplaces. These may only require levels of security and surveillance comparable to noncoercive workplaces, because coercive pressures are strongest during initial recruitment and because the social context precludes free movement. Structurally coercive environments are most aptly conceptualized as a “continuum” of “forms and degrees of unfreedom” where clearcut distinctions between voluntariness and involuntariness are inapplicable (Phillips, 2013).
In Uzbekistan, the forced labor system leveraged structural marginalization for economic exploitation. Xinjiang’s system at least nominally aims to increase incomes so that the state can declare ethnic regions to be free from poverty. Given China’s low absolute poverty threshold, even meager remuneration for picking cotton suffices to push ethnic pickers above that line. However, the state’s long-term goal is to shift people from traditional livelihoods and communities to “modern,” state-controlled settings of work, indoctrination, and social control (Zenz, 2019b, 2021a, 2023b).
Global sanctions targeted Uzbekistan’s economic rationale for coercion and eventually led to systemic change, also due to the country’s strong need to attract more foreign investment. In contrast, Beijing’s economic might, long-term political aims in Xinjiang, and their perceived implications for China’s national security mean that coercive labor transfers into cotton picking and related industries could persist for a long time to come (Roberts, 2020; Zenz, 2021b, 2022c). Chinese companies using cotton from Xinjiang evade Western sanctions by obfuscating supply chains (Murphy et al., 2021). While sanctions are necessary to avoid consumer complicity in the atrocity, Xi Jinping's prioritiziation of domestic security over economic growth and the enactment of a countersanction law penalizing companies that comply with sanctions render their impact more uncertain. (Drinhausen & Legarda, 2021).
Notes
A preprint of the cited December 2019 journal article had been published in July 2019.
Compare McGuire and Laaser (2021, p. 566).
Interview conducted in 2021 in Turkey by interlocutors collaborating with the author.
Calculated from the Xinjiang Bureau of Statistics (2020, table 12-20). Southern Xinjiang includes Khotan, Aqsu, Kizilsu, Kashgar, and Bayingholin.
Calculated from the Xinjiang Bureau of Statistics (2020, table 12-20).