The conceptual evolution of poverty alleviation through labour transfer in the Xinjiang Uyghur Autonomous Region

ABSTRACT

This paper argues that in China’s Xinjiang Uyghur Autonomous Region, work placements of re-education detainees and Xinjiang’s implementation of the national Poverty Alleviation through Labor Transfer programme for the transfer of rural surplus labourers operate under fundamentally different policies. Drawing on new documentary and witness evidence, it is argued that within Xinjiang’s unique context of frontier settler colonialism, its recent coercive labour transfer programme evolved alongside decades-long efforts to facilitate surplus labour transfers throughout China. From 2014, when Beijing shifted the region’s work focus towards de-extremification, Uyghur underemployment was framed as a matter of social stability and national security. Between 2017 and 2019, labour transfer coercion dramatically increased alongside campaigns of mass internment and of enforcing poverty alleviation work goals. Xinjiang’s shift in 2021 from a campaign-style mobilizational to an institutionalized approach deepens coercive risks of this often poorly understood coercive labour strategy.

KEYWORDS:

• China
• Xinjiang
• Uyghurs
• poverty alleviation
• labour transfer
• forced labour

Introduction

In early 2017, the Xinjiang Uyghur Autonomous Region (XUAR) in north-west China embarked on a campaign of interning 1–2 million Uyghurs and other predominantly Turkic ethnic groups into re-education camps (Zenz 2018, 2022b). Initially called Transformation Through Education Centres, their terminology was later unified to Vocational Skills Education and Training Centres (职业技能教育培训中心 – VSETCs). At a US Congressional hearing in October 2019, evidence was presented for multiple XUAR coercive labour programmes: (1) work placements of VSETC detainees; (2) labour transfers of (non-detained) rural surplus labourers into non-agricultural work; and (3) as part of labour transfers, programmes to ‘send work to people’s doorstep’ through construction of village ‘satellite factories’ (Zenz 2019; cf. Lehr and Bechrakis 2019).¹ This led to the important and still necessary presumption that Xinjiang’s products are tainted with forced labour, legislated as the Uyghur Forced Labor Prevention Act (UFLPA), which came into effect in June 2022 (Bloomberg 2022). Subsequent research on coercive labour transfers into cotton-picking led to the US import ban on all Xinjiang cotton in January 2021 (Sudworth 2020; Zenz 2020).

Xinjiang’s forced labour extends the region’s decades-long campaign of frontier settler colonialism, an effort that accelerated dramatically under the region’s party secretary Chen Quanguo (2016–21) who oversaw mass internments and efforts to ‘optimize’ the ethnic population structure through birth prevention and population transfers (Smith Finley 2021; Zenz 2018, 2021b). Labour transfers are enforced through large-scale transfers of land usage rights (Zenz 2021a, 2023a). Chinese academics themselves argue that these transfers ‘crack open solidified [Uyghur] society’ and ‘reduce Uyghur population density’, linking them with an attack directed against the population (Zenz 2021a, 14–15, 2021b).

From a sociological perspective, Xinjiang’s coercive labour programmes can be viewed as complementary efforts to indoctrinate and transform ethnic (non-Han Chinese) farmers into docile industrial workers as part of the state’s assimilatory project of coercive social re-engineering. Both schemes create a workforce that is considerably cheaper than eastern Han Chinese workers, fuelling the region’s integration into global supply chains (Murphy and Elimä 2021).

However, the central argument of this paper is that the work placements of re-education detainees and Xinjiang’s coercive implementation of the national Poverty Alleviation through Labor Transfer (转移就业脱贫) policy for the transfer of rural surplus labourers (农村富余劳动力) are fundamentally different. While both systems can interface, potentially placing targeted groups into similar work environments, they evolved at different times from different precursors, operate under different policy and implementation schemes, and achieve different (albeit complementary and partially overlapping) aims. While VSETCs administer intensive de-extremification to ‘cure’ persons already ‘infected’ with the ‘virus’ of ‘religious extremism’, labour transfers are to preventively inoculate those who are not yet ‘infected’ (Zenz 2023b). While labour transfers occur in a generally coercive environment that is heightened by the threat of internment for resistance to state policies, the labour transfer programme generates its own grassroots-based coercive pressures and tightly administered processes to enforce mandated goals.

This paper argues that labour transfers evolved from an in itself legitimate socio-economic policy that has now come to be part of a multi-faceted atrocity. State-sponsored labour transfers from agriculture to other sectors arose from needs faced by most developing countries transitioning toward industrialization, in China’s case exacerbated by Mao Zedong’s Stalinist industrialization strategy. In Xinjiang, from its inception in the early 2000s, this policy interfaced with colonialist marginalization and Han dominance over resources. It then became much more coercive and pervasive due to (1) Beijing’s shifting political priorities for Xinjiang after 2014 prioritizing full (off-farm) employment, (2) mass internments from early 2017, and (3) Xi Jinping’s national campaign to eradicate absolute poverty by 2020, prompting an intense poverty alleviation work rectification campaign in 2019.

The Chinese state defines rural surplus labourers, also referred to as rural migrant workers (农民工), as persons engaged in forms of subsistence agriculture who are considered to be superfluous labour within modernizing agricultural production (Chen, Zhang, and Shi 2019; NBS 2005; PRC Ministry of Agriculture 2003; Xinhua 2020). State policies and statistics on these labourers' transfer from primary to secondary or tertiary economic sectors include various forms of self-initiated labour migration (“self-transfers” 自发), transfers facilitated by private intermediaries (often incentivized by the state), and transfers directly supervised by state agencies, all of them occurring within an increasingly draconian employment policy context.

Labour transfers target non-detained rural populations and involve a wide range of sectors, including cotton-picking, tomato-picking and polysilicon production for solar panels (Swanson and Buckley 2021; Murphy and Elimä 2021; Zenz 2020, 2023a). By contrast, VSETC-linked labour placements began in 2018, having evolved from China’s Re-Education Through Labor system, to effect a controlled release of detainees into a narrower range of low-skilled manufacturing or other workplaces, with no established links to cotton harvesting or polysilicon production (Zenz 2023b). In contrast to labour transfers, many detainees were not agricultural workers before internment, and some can eventually return to their original (non-agricultural) workplaces. VSETC-linked labour victims can receive almost no pay and work in highly securitized environments. By contrast, transferred labourers face multiple unfreedoms but can potentially increase incomes compared with farming despite receiving much lower remuneration than average Chinese workers. Labour transfer implementation is governed by a formal poverty alleviation policy pursuing both political and economic goals. The camp-to-labour policy likely ended late 2019 when many lower-security camps were closed, with many released detainees likely remaining in forced labour. By contrast, labour transfers further intensified from 2021, and now constitute Xinjiang’s primary coercive labour system (Zenz 2022c).

This conceptual framing of Xinjiang’s two labour systems, first outlined in my study on forced labour (Zenz 2019), was adopted in the United Nations High Commissioner for Human Rights’ report on Xinjiang (Office of the United Nations High Commissioner for Human Rights (OHCHR) 2022, 37). While some non-scholars may consider such conceptual distinctions irrelevant, inaccurate conceptual framings can cause policy responses that the Chinese state could rightly challenge. They could also lead to highly misleading conclusions about the continued existence of forced labour if China credibly demonstrated camp-linked labour policies had ended, a scenario that is most likely already reality.

This paper will (1) contextualize the issue in the existing literature, (2) elucidate the historical and socio-economic context of rural surplus labour transfers in China and Xinjiang, (3) examine the evolution of such transfers in Xinjiang to date, (4) contrast labour transfers with VSETC-linked work placements, and (5) present important new evidence for the threat of internment for refusing to comply with state employment policies. It focuses mainly on labour transfers due to their larger scope and scale. Conceptual disagreements are not intended to diminish other authors’ very important contributions.

Literature discussion

Given the complexities of Xinjiang’s labour systems and that labour transfers became most coercive at the start of the mass internments, scholars’ knowledge of this subject has been evolving. While clearly distinguishing between different coercive labour placement mechanisms, in the earliest research on Xinjiang’s forced labour I framed camp-linked forced labour in the context of poverty alleviation policies and expressed doubt that this involved genuine attempts to alleviate poverty, especially concerning camp detainees (Zenz 2019). This interpretation was influenced by the presence of poverty alleviation terminology in lower-level government sources. However, this work elucidated related distinct policy systems, and I subsequently provided detailed distinctions between both labour systems (Zenz 2020, 2021a).

Other authors have identified important empirical evidence which has yet to be rigorously conceptualized. They have tended to frame internment-linked labour in terms similar to the poverty alleviation policies that undergird labour transfers, often using evidence from one system to demonstrate the coercive nature of the other. A March 2020 report by the Australian Strategic Policy Institute (ASPI) grounded Uyghur labour transfer coercion substantially in the re-education campaign, while also invoking aspects of the labour transfer framework (Xu et al. 2020). It suggested that some Uyghurs are transferred ‘directly from detention camps’ and that labour transfer ‘work assignments’ are ‘enmeshed with the apparatus of detention’ (4). A July 2021 report by Sheffield-Hallam University argued outright that Xinjiang’s labour transfers are ‘tantamount to […] enslavement’ (Murphy and Elimä 2021, 7, 9). The authors loosely distinguished labour transfers from VSETC-linked labour, but described labour transfer coercion using an assortment of closely enmeshed sources pertaining to both systems. The report argued that labour transfers are ‘not […] designed to lift people out of poverty’ because many who work in the camps are ‘trained professionals’, and cited witness accounts from detainees not involved in labour transfers to suggest that such transfers constitute a form of ‘human trafficking’ (12–13, nn 29–33). Based on some interviews and his interpretation of one Chinese expert’s policy suggestion, Darren Byler suggests that surplus labourers and camp detainees are co-placed in the same industrial parks through a ‘development mechanism linkage’ (Byler 2022b), an interpretation analysed below that is in my view inaccurate and also does not constitute an integrative link between both programmes. Inasmuch as all these accounts present current surplus labour transfers in a largely decontextualized fashion – not situating them in the wider national socio-economic and policy context – they can overstate the internment campaign’s role in labour transfer coercion and blur our understanding of policy by framing poverty alleviation aspects as propaganda. I suggest that poverty alleviation and economic goals coexist alongside political goals.

Less-than-accurate conceptual framings of Xinjiang’s forced labour have ramifications. In February 2022, an International Labour Organization (ILO) report cited observations by the International Trade Union Confederation (ITUC) on forced labour in Xinjiang. It incorrectly suggested that Xinjiang’s entire system of coercive labour is based on ‘internment or ‘re-education’ camps’ (ILO 2022, 515). Presumably referring to camp-linked forced labour, it asserted that paid wages amount to ‘almost nothing’ (516). Similarly, a November 2022 report by the Australian Clean Energy Council claimed that ‘[a]pproximately 2.6 million Uyghur and Kazakh citizens are alleged to be subjected to ‘surplus labour’ programmes and face significant coercion, re-education programs and internment’ (Clean Energy Council 2022, 6). This statement cites the Sheffield-Hallam report, which misunderstood a partially erroneous official English translation of the 2020 Xinjiang white paper on Employment Rights to state that ‘2.6 million minoritised citizens’ were subjected to labour transfers (Murphy and Elimä 2021, 7, 9) (discussed below). The council also introduced the common error of closely associating labour transfers with re-education internment, perhaps not unexpected given that the cited reports (ASPI and Sheffield-Hallam) ended up framing both labour systems as largely enmeshed.

Xinjiang forced labour research has an outsized influence on policymaking. In early 2022, I reviewed high-level policy research commissioned by a supra-national policy deliberation body intended to inform the development of major legislation countering Xinjiang’s forced labour. Among several significant conceptual inaccuracies, the report wrongly implied that cotton-harvesting and polysilicon production are linked to camp labour. The August 2022 UN Xinjiang report stated that Xinjiang operates two ‘surplus labour’ and ‘labour transfer’ ‘schemes’ (OHCHR 2022, 37), an incorrect phrasing for this singular policy adopted from Murphy and Elimä (2021, 7). All this indicates how poorly understood Xinjiang’s employment policies often are.

Research method

The ‘Investigative Policy Analysis’ method of documenting Xinjiang’s atrocities that I developed identifies policy frameworks and their implementation by illuminating policies from different angles, drawing on government and other textual sources (Zenz 2018). In mid-2019, this approach was adapted to examine forced labour (Zenz 2019, 2021a). Subsequent research on Xinjiang’s forced labour by think tanks and scholars largely adopted this approach. However, its adoption necessitates rigorous contextualization of applicable material in immediately relevant policy contexts. Some studies employed the approach by serially concatenating assortments of texts containing select keywords.

Researchers had to contend with the fact that Xinjiang’s most salient policy implementation evidence often came from lower, county-level, policy. Higher-level policies are often classified, and research has noted that ‘[a]ll central policies eventually have to be carried out by county and township governments, and the cadre bureaucracies at these administrative levels are decisive for the success of any reform initiative passed down by the central state’ (Heberer and Schubert 2017, 1142). County-level practices are important but cannot always be generalized. For example, a county-level Work Implementation Plan from Qapqal county cited by Murphy and Elimä (2021) and the UN Xinjiang report (OHCHR 2022, 38) states that surplus labourers unwilling to be trained or transferred will have points deducted, which risks them being deemed ‘untrustworthy’. I was unable to identify similar mandates in other regions.

To increase the authoritativeness of XUAR forced labour research, this paper analyses the evolution and current state of labour transfer policies through a multi-decade chronological review of major XUAR-wide planning documents such as Five-Year Plans. This method, first employed by Zenz (2022c, 2023a) and subsequently adopted in the UN Xinjiang report (OHCHR 2022, 38), is found in several studies of Chinese policies. For example, Wang’s (2017) study of discursive governance under Xi Jinping drew on 18 major political texts to highlight how political discourse is imbued with evaluative language, reinforcing discursive steering and related mobilization of cadres through embedded emotive and moral imperatives.

To assess policy implementation, this study integrates lower-level texts and examines internal documents, including the ‘Xinjiang Papers’, ‘Xinjiang QQ Files’ and ‘Xinjiang Police Files’, ranging from central government speeches to county-level policy implementation directives (Zenz 2021c, 2022a). Such multilevel policy analyses are facilitated by the state’s extensive use of ‘[p]olitically binding standard phrases’ or tifas (提法), which allow policy frameworks to be traced across conceptual levels (Heilmann 2017, 313).

Currently, there are no known witness accounts of recent labour transfers. This article publishes the first such witness accounts (second-hand), and the first testimony confirming that Uyghurs refusing such transfers were detained in camps (also second-hand). While representing a very limited dataset, these accounts confirm the overarching analysis.

Since most labour transfers occur within Xinjiang, this paper focuses mainly on intra-regional transfers (XJASS 2018). Pairing Assistance, which pairs developed eastern provinces with poorer western regions, incentivizes eastern Chinese companies to invest in Xinjiang. As part of ‘industry-based poverty alleviation’, it funds factories inside VSETCs and factory parks (Li 2018; Zenz 2019). Pairing Assistance plays only a limited role in transfer coercion, which is primarily rooted in XUAR policy implementation processes.

Labour transfer statistics are challenging to interpret accurately. For example, Murphy and Elimä’s assessment that 2.6 million ethnic persons in Xinjiang are subjected to labour transfers (Murphy and Elimä 2021, 7, 9) is based on the English version of the 2020 White Paper on Employment Rights. This statistic refers only to the first half of 2020, whereas the full-year figure was 3.15 million person-times (Lawinfochina 2020; NEAC 2021). The English translation incorrectly states ‘persons’ instead of ‘person-times’ (人次), meaning fewer persons were transferred as labourers can be transferred multiple times per year (in 2019, the difference between ‘person-times’ and ‘persons’ amounted to 283,000 or 11%). Also, transfer figures include Han. In 2019 Xinjiang reported surplus labourers by person (not person-times) at 2.59 million, of which only 1.65 million were from the four southern Uyghur-majority prefectures of Aqsu, Kashgar, Khotan, Kizilsu (Lawinfochina 2020). Labour transfer statistics are therefore not equivalent to coercive labour estimates.

Based on Xinjiang’s intensification of employment and vocational training policies after 2020, I increased my estimate of ethnic persons at risk of coercion through labour transfers from ‘up to 1.6 million’ to ‘up to 2.0 million’ (Zenz 2022c). Adding a conservative estimate of at least several hundred thousand persons affected by camp-linked labour results in a total coercive labour estimate of up to 2.5 million. Transfer figures are not cumulative as surplus labourers are transferred annually.

Rural surplus labour in China

In the classic dual-sector or Lewis–Ranis–Fei model of economic development, surplus labour is defined as ‘labor [that] can be transferred out of the traditional [agricultural] sector without reducing the volume of farm output’ (Cook 1999, 18). Abundant labour in the agricultural sector is characterized by low marginal productivity (the marginal product of labour is the change in output resulting from employing one added unit of labour). China is widely recognized as having abundant rural surplus labour (Chan and Wei 2019; Cook 1999; Hasmath 2019; Wang and Weaver 2013). Labour transfers move workers from primary to secondary and tertiary sectors, which may or may not involve geographical relocation.

In the 1950s, Mao Zedong imitated Joseph Stalin’s development strategy of promoting industrialization by systematically exploiting a land-confined peasantry through a system of unequal exchange, restricted population mobility and rural collectivization (Chan and Wei 2019, 427). Based on this ‘rural–urban dual system’, rural population shares remained largely stable between 1955 and 1978 at around 84–85%, artificially maintaining large numbers of underemployed surplus labourers (Chan and Wei 2019, 428). With decollectivization in the 1980s, peasants were permitted to seek work in cities, unleashing millions of rural surplus migrant workers who fuelled China’s export-driven development strategy (Chan and Wei 2019, 431).

Rural exploitation and labour transfer policies in China and Xinjiang in the 2000s

By the late 1990s, China’s socio-economic inequality had increased significantly (Heilmann 2017, 259). At the 16th Party Congress in 2002, Jiang Zemin refocused Deng Xiaoping’s original mandate to ‘comprehensively build a moderately prosperous society’ (Heilmann 2017, 52). As part of inequality reduction, the government issued its ‘2003–2010 National Rural Migrant Worker Training Plan’ emphasizing ‘labor transfer of the rural surplus labor force’ (PRC Ministry of Agriculture 2003).

In Xinjiang, the transition from state-led to market economy in the 1990s offered limited viable employment opportunities for ethnic groups amid declining public-sector employment (Li 2015; Hasmath 2019). In the early 2000s, the state’s campaign to ‘Open Up the West’ exacerbated ethnic inequalities (Hasmath 2019; Millward 2021, 292–302). Han investors’ growing interest in Uyghur farmland effected an ‘instrumentalization of poverty’ whereby local bureaucrats coerced Uyghur farmers into selling land usage rights (Cappelletti 2020, 234–64). Arable land became concentrated in the hands of local elites (Byler 2022a; Tohti 2015, 30). Uyghur economist Ilham Tohti noted how Uyghur farmers were forced to purchase farm inputs and sell produce at prices fixed by local governments, trapping them ‘on tiny plots of land with no capital or access to credit’ (Millward 2021, 364–65). Such practices are essentially vestiges of Stalinist industrialization, which throughout China enforced unequal exchange between rural and urban regions through mandatory quotas of produce to be sold to the state at fixed and artificially low prices (Chan and Unger 1982). This system, abolished in the mid-1980s with Deng Xiaoping’s economic reforms, may have prevailed longer and in potentially more exploitative forms in Uyghur regions.

Tohti (2015) linked non-Han unemployment in Xinjiang to blatant ethnic discrimination in recruitment, securitization policies that impacted population mobility, and Han dominance over economic and natural resources. He argued that southern Xinjiang’s per capita arable land was so low that ‘vast reserves of surplus rural labor’ needed transferring to other sectors (5). Tohti recommended improved education, training and employment opportunities for non-Han outside of agriculture, and promotion of ‘internal population migration’ through state-led ‘systematic transfer[s]’ of Uyghurs to Xinjiang’s northern industrial regions (7). While Tohti was somewhat careful in his wording, the low land per capita ratio arguably resulted from not only pre-reform policies or rapid Uyghur population growth but also forms of dispossession. Tohti clearly implied that some Uyghurs desired labour migration but lacked skills, structures and opportunities, an assessment shared by Western scholars (Cappelletti 2020; Hann 2014; Hasmath 2019).² Impediments to rural–urban labour migration restricted upward mobility, a phenomenon Chris Hann called the ‘Uyghur variant of the “low level equilibrium trap”’ (Hann 2014, 188).

In the 1990s, some Uyghurs organized themselves to pursue employment in eastern China, for example, a group of rural migrant workers from Khotan (Hotan) prefecture’s Qaraqash (Karakax) county who first left their home regions in 1996 (PRC Government 2010). Related local state-led initiatives did not rapidly develop until 2003 (Sohu 2006). Xinjiang’s 10th Five-Year Plan (2001–05) spoke of ‘transferring the rural surplus workforce’ and set a low annual goal of transferring 150,000 labourers (Beijing Municipal Commission 2016).

Numbers of transferred labourers in Qaraqash county more than doubled from 21,850 in 2002 to 45,160 in 2003, then rose another 76% to 79,681 by 2006 (Yang et al. 2010, 135). In 2004, Xinjiang began to organize dedicated Leading Small Groups to coordinate rural labour transfers, with Khotan prefecture alone establishing 1,327 village-level ‘labor export associations’ for cross-regional transfers (Liu and Yang 2004, 59). A Bayingholin prefecture (2005) policy directive indicates that already then, each region had assigned annual training and transfer quotas.

A 2005 report by China’s National Bureau of Statistics (NBS) described Xinjiang’s labour transfers as directly related to social stability (NBS 2005). Out of 1 million ‘person-time’ surplus labourers transferred in 2004, some 585,200 were from southern Xinjiang. It further complained that transfers were insufficiently state-driven: about 50% were self-initiated, 20% were facilitated by family or friends, and ‘only’ 30% were organized by government entities. The report therefore lamented that the ‘flow of surplus rural labour in Xinjiang was in a semi-disordered state’, rendering state-controlled macro-level steering difficult.

Xinjiang’s 11th Five-Year Plan (2006–10) highlighted ‘employment pressures’ and specified ‘labor transfer training for the rural workforce’ (Beijing Municipal Commission 2016). For the first time, it explicitly outlined ‘labor export’ plans for transfers outside people’s home regions. To boost shares of ‘organized’ (state-directed) transfers, it called for ‘strengthening organization of the rural workforce’. By 2007 transfers rose to 1.45 million person-times, of which 774,200 were linked to southern Xinjiang and 79,500 involved transfers to other provinces (Union China 2008). The authorities provided subsidies, information networks and employment service companies to promote transfers through various channels (Sohu 2006).

While some rural labourers likely welcomed state policies enabling them to earn wage incomes in cities and industrial sectors, transfers already showed evidence of coercion. A 2007 Chinese research report cites a township leader in Kashgar’s Konasheher county: ‘Nowadays, more farmers in the suburbs of cities and towns go voluntarily, but in remote rural areas there is indeed a phenomenon of compulsion. […] Working in the countryside [ … ] it is impossible to [do this] without any force’ (Ma 2007, 35).

Cappelletti (2020, 149–50) suggests that ‘at least since 2005’, the state was coaxing young rural Uyghur women to work in textile factories in eastern China to prevent early marriages and promote social ‘modernization’. Tohti similarly argued that such transfers ‘started out as a [ … ] worthwhile endeavor’ but employed ‘coercive methods’ (Tohti 2015, 30). For example, after Rabiya³ (pseudonym) graduated from middle school in Kashgar’s Payzawat county in 2007, she and her classmates were taken to the county higher vocational school for Chinese language courses to prepare them for factory work in eastern China. They had not agreed to participate and were locked inside the school. Villages were given mandatory annual transfer quotas. After 15 days of enforced training, a total of 210 girls were transferred to a Tianjin textile factory where they were confined to dormitories, surveilled with cameras and worked excruciatingly long hours. Rabiya could only quit after completing a one-year contract.

By the end of 2010, Xinjiang had 2.2 million rural surplus labourers, 83.8% of them from ethnic groups, mostly in southern Xinjiang (Li 2015, 34). After the 2009 Ürümchi riots, themselves fuelled by growing Uyghur resentment over sociocultural and economic discrimination, labour transfers to other provinces declined drastically as the state rapidly increased securitization and surveillance, forced rural migrants in northern Xinjiang back to their southern homelands, and curtailed free movement (Li 2015, 36; Tohti 2015, 7; Zenz and Leibold, 2017, 2019).

The conceptual evolution of labour transfers, 2014–17

At the Second Central Xinjiang Work Forum in May 2014, central government priorities for Xinjiang shifted from economic development to ‘de-extremification’ (去极端化) and stability maintenance (Zenz 2021c). Anthropological fieldwork confirms 2014 was a watershed year for Uyghurs, who felt increasingly coerced to participate in formal state policies (Steenberg and Rippa 2019, 275). Officials argued that economic development ‘must absolutely be subservient to social and long-term peace and stability’, providing an important supporting role (Central Office Bulletin 2014, 65). Premier Li Keqiang noted that southern Xinjiang’s 3 million surplus labourers posed a ‘particularly prominent’ problem, arguing that ‘people without land, employment or a fixed income have nothing to do and wander all day’ and will ‘be easily exploited by evildoers’ (Central Office Bulletin 2014, 39–40). Similarly, Xi Jinping stated that the unemployed will ‘provoke trouble’, whereas employment is ‘conducive to ethnic interaction, exchanges and blending’ and leads ethnic groups to ‘imperceptibly study Chinese culture’ (20).

By 2017, Chinese companies in the region had adopted this framing. Huafu, which operates the world’s largest textile mill in Xinjiang, stated on its website:

Due to lack of information, lack of courage, and fear of going out, large numbers of rural surplus labourers are idle at home, which increases the burden on their families and brings hidden dangers to public security. Aqsu Huafu actively engaged with government departments, actively absorbed surplus labor [ … ] to gradually transform them from farmers to industrial workers. (Huafu 2017)

The 2014 reorientation of political priorities is reflected in planning documents. While Xinjiang’s 12th Five-Year Social and Economic Development Plan (2011–15) had devoted little space to labour transfers (XUAR Government 2011), its 13th Five-Year Plan (2016–20) gave considerable space to the topic and struck a markedly different tone (NDRC 2016). It verbatim repeated Xi Jinping’s demand from 2014 to ‘systematically expand the scale of [relocating] Xinjiang’s ethnic minorities to other parts of China to receive education, employment, and residence’ (NDRC 2016, 97). It decreed the establishment of 25 industrial parks to create 150,000 jobs. The first Five-Year Plan to set annual targets for labour transfer, it cited a goal of 2.2 million person-times, a target later exceeded by over 30% (Figure 1) (NDRC 2016, 97; XUAR Government 2021a).

Figure 1. Xinjiang Uyghur Autonomous Region (XUAR) labour transfers, 2014–22. Sources: XUAR annual socio-economic progress reports, www.xinjiang.gov.cn

Xinjiang’s February 2016 adoption of the national Decision to Win the Battle Against Poverty established annual targets to achieve Xi Jinping’s mandate of eradicating absolute poverty by 2020 (China News 2016). It specified the use of village-based work teams (officials going door to door) to implement Targeted Poverty Alleviation work, warning that teams must remain villages until poverty alleviation had been achieved. These teams were later used to identify persons for re-education internment and to enforce numerous state policies (Zenz 2019, 2021a, 2023a). The battle against poverty would involve ‘stimulating the inner motivation of the masses’ under the explicit maxim ‘curing poverty means to first cure ignorance and backwardness’. The state’s ambitious transfer goals necessitated increasingly coercive pressures: Chinese academics found that Uyghurs often resisted such transfers even when offered adequate remuneration and free housing (Deng, Mamati, and Wang 2016, 83).

Paternalistic language about improving the ‘quality’ (素质) of ‘backward’ ethnic populations at the expense of traditional knowledge and livelihoods pervades informal official discourse throughout many non-Han regions (Zenz 2013). However, it had been uncommon in formal state texts. Its foregrounding in planning documents from 2016 indicates how officials were to enforce policy outcomes through increasingly overt coercion. Local evidence confirms resulting abuses. An internal disciplinary inspection report states that in several townships in Konasheher county, officials ‘forced’ Uyghur villagers to work in textile and satellite factories for minimal pay (Shufu County Commission 2017).

Increased labour transfer coercion under Chen Quanguo, 2017–19

A decisive turn towards more coercive mobilizational approaches coincided with Chen Quanguo’s appointment as Xinjiang’s party secretary in 2016 and the mass internments from early 2017.

Xinjiang’s 13th Five-Year Poverty Alleviation Plan from May 2017 adopted the new central government concept of the Poverty Alleviation through Labor Transfer programme (Yecheng Government 2017). Labour transfers of older persons and ethnic minority women were now to be accelerated through satellite factories. The plan noted that poor people’s ‘labor and employment willingness and abilities are insufficient’. Relying heavily on employment creation and labour transfers, it repeated five times that the ‘inner motivation’ of locals is insufficient and must be ‘stimulated’. People’s outdated mindset of ‘waiting, relying, wanting’ must be ‘eradicated’, a phrase not found in previous major planning texts. The plan reiterated that ‘curing poverty means to first cure ignorance and backwardness’. The rapid acceleration of state-led labour transfers coincided with an even more drastic curtailment of voluntary migration and self-chosen work. Between late 2016 and early 2017, Uyghur rural–urban migrant workers were forced to return to their home regions (Tynen 2020, 314–315). Evidently, the goal was not any employment but state-controlled work. Between 2014 and 2018, the government’s aggressive securitization and expansion of state-led policies gradually displaced private economic relations with state employment and more formalized state–society interactions (Steenberg and Rippa 2019; Zenz and Leibold 2019).

In mid-2017, the XUAR published a Regional Development and Poverty Alleviation Implementation Plan specifically for southern Xinjiang (Alashankou Government 2017). Involving a ‘strengthening of gratitude education’, it specified all-out social mobilization efforts to ‘stimulate [people’s] drive and determination to change [their] situation of poverty [and to] change [people’s attitudes] from “I am wanted to get rid of poverty” to “I want to get rid of poverty”’. Governments at all levels were to ‘more proactively implement employment policies’.

In concrete terms, this means that cadres in village-based work teams would ‘deeply penetrate’ households and perform ‘thought work’ until they ‘cause a transformation in the way farmers think about choosing their employment’ (Nilka County 2020). An internal December 2018 work summary from a village-based work team in Khotan county described how they entered each home. Poor families were to be made aware of their deficiencies. They were subjected to strengthened ‘motivational education’, and households considered to be poor because of ‘laziness’ were sent to dedicated ‘education’ activities (Khotan County 2018). An internal June 2019 Yarkand county poverty alleviation report describes the public shaming of ‘lazy people’ (at public meetings about village mobilization and beautification campaigns) to stimulate the motivation of the masses: ‘[L]et lazy persons speak [in front of all], let them blush and sweat, stimulate the inner motivation of the whole village, especially the inner motivation of poor households’ (Shache County 2019).

Mobilization tactics include large-scale transfers of agricultural and pastoral land usage rights, whereby farmers receive payments per unit of their land entered into government trusteeship schemes. Following ‘vigorous promotion’, 58,000 households in Aqsu had transferred 154,500 hectares of land, resulting in the labour transfer of 73,300 persons (Aqsu Propaganda Committee 2020). This led to unprecedented transfer shares: in September 2018 a village in Altay prefecture reported a labour transfer rate of ‘over 95 percent’, boasting that it had ‘realized a village without idlers’ (China News 2018). Internal documents show how amid an apparent absence of men, a Tekes county village was ordered to round up ‘all women and other surplus labourers’ – 500 persons from only 391 households – to work in neighbouring cities (Qiaolake 2018). Such drastic livelihood changes, dictated by long-term political goals and brought about through mass mobilization campaigns, involve a high risk of coercion. In 2022, labour transfers through transfers of land usage rights continued, with state organs citing examples of rural Uyghur couples in Kashgar being sent to factory parks while their land and herds were managed by state cooperatives (NDRC 2022).

In January 2018, Xinjiang initiated a special plan to transfer 100,000 labourers from 22 poor counties in southern Xinjiang to other regions. It ended up transferring more than 221,000 workers and contained the explicit mandate to ‘train all who should be trained’ (应培尽培), emphasizing intensive political indoctrination, ‘gratitude to the party’, Chinese language skills, work discipline and military drilling (Lawinfochina 2020; Xuehua 2018). Centralized state-led transfers involved accompanying officials and police guards (Chen, Zhang, and Shi 2019; Zenz 2021a). Between 2004 and 2018, the number of labour transfers in southern Xinjiang grew threefold, from 585,200 to 1,736,000 person-times (NBS 2005; Xinhua 2019).

Unlike with VSETC-linked labour, firsthand witness testimony has not emerged from rural surplus labourers. However, Anar Sabit, a Kazakh camp survivor whose testimony featured in The New Yorker, noted that between July and October 2017, many street cleaners from southern Xinjiang appeared in her northern city of Kuytun.⁴ A relative with a government job explained that these were transferred workers who work during the day and at night are ‘locked’ in dormitories. This type of worker in the same city (Kuytun) featured in undercover footage obtained by The New York Times, depicting transferred surplus labourers from Khotan and Kashgar in a secure work compound with security checkpoints (Buckley and Ramzy 2019). These workers stated in the video that they could not leave at will. Similarly, a Kazakh witness stated that in 2021 the state transported many southern Uyghurs to his city of Karamay.⁵ Six transferred labourers worked in his restaurant. One of them, a Uyghur man from Kashgar, applied multiple times to the state for permission to return home, which was eventually granted.

These accounts indicate that transferred labourers are not free to leave, consistent with a Yanqi county (Bayingol prefecture) directive regarding labour transfers from southern Xinjiang. It required authorities to ‘ensure that stable employment rates reach 95 percent’ (Yanqi Government 2018). Transferred labourers who temporarily leave their workplaces were to be tracked so that ‘their whereabouts and actions are clear’. Obtaining approval for such leave involved multiple government entities.

The 2019 poverty alleviation rectification drive

Internal government documents from the Arslanbagh Township Poverty Alleviation Work Group (Yarkand county), obtained as part of the ‘Xinjiang QQ Files’ (Zenz 2022a), show how state efforts to compel Uyghurs into poverty alleviation and employment measures intensified after 2018, including through a ‘rectification’ campaign targeting perceived shortcomings in implementing such policies. Yarkand’s July 2019 document on ‘Recent Key Work in Poverty Alleviation’ put officials under extreme pressure, mandating that ‘lazy persons, drunkards and other persons with insufficient inner motivation’ would need to be subjected to ‘repeated [ … ] thought education’ (Jinqi tuopin 2019, 3). If this failed to produce ‘obvious results’, they were to be dealt with ‘according to the document’ (while the exact meaning is unclear, the context suggests coercive measures). Students and elderly persons over 60 years were to pick crops such as cotton, vegetables, tomatoes, and peppers to instil a view that ‘as long as one is able to work, [even] the weak part of the workforce must stir up their inner motivation and create value’ (4). By late 2019, Yarkand was compiling lists of ‘lazy persons’, ‘drunkards’ and persons ‘without sufficient inner motivation’ (Qi cun shachexian n.d.). One list labelled individuals as old as 77 as ‘lazy’.

Arslanbagh Work Group documents include a Targeted Poverty Alleviation Information Collection Form for 2018–20 used by state work teams to evaluate reasons for poverty (Kashi diqu n.d.). Among them is the category ‘Insufficient Development Motivation’, subdivided into three checkboxes: (1) ‘a mindset of waiting, relying, wanting’, (2) ‘influenced by religious thinking’, and (3) ‘low cultural level’ (implying ‘backward’ mindsets). Persons whose ‘insufficient development motivation’ was attributed to ‘religious thinking’ would have been potential internment targets.

Developments since 2020: unemployment monitoring and early warning

Since 2020 and especially under Xinjiang’s new party secretary Ma Xingrui, a technocrat from Guangdong experienced in economic development, Xinjiang has been shifting from Chen Quanguo’s highly mobilizational, campaign-style labour transfers to a more normalized and institutionalized strategy that emphasizes maintaining labour placement achievements through intensified monitoring. Xinjiang’s 13th Five-Year Poverty Alleviation Plan first specified the creation of an ‘Unemployment Monitoring and Early Warning System’ to ensure achievement of employment targets. Xinjiang’s 14th Five-Year Plan for Employment Promotion (2021–25) mandated, in the critical ‘guiding thought’ section that outlines the plan’s core principles, that not just one person per household but ‘every single person who is able to work is to realize employment’ (XUAR Government 2021c). Governments at county and township levels must ‘comprehensively analyse the specific reasons for the decline in [a particular household’s] income’, and the first listed countermeasure is labour transfer. Xinjiang’s 14th Five-Year Plan for Socioeconomic Development (2021–25) reiterates these goals (XUAR Government 2021b). Through this intensification, labour transfers reached a record 3.17 million person-times in 2021 before levelling off at above 3 million in 2022 (Figure 1).

Whereas in the 2000s transferred Uyghurs often left their assigned workplaces, recent labour transfer implementations enforce worker retention. A September 2020 report about transferring 51,154 southern Xinjiang labourers specified a target of ‘95 percent stable employment’ (NRRA 2020). Retention was enforced through close monitoring by state agencies in both home and work regions. Only 61 labourers left their work (due to illness), resulting in a ‘stable employment rate of 99.88%’. In 2021, Xinjiang sent 400,000 cadres to monitor the income situations of 12 million rural households through an ‘early prevention, early intervention, early assistance’ campaign that identified 774,000 households for ‘real-time monitoring’ (China Daily 2022). This further prevents transferred Uyghurs from leaving workplaces without state approval. In 2023, the new ‘Southern Xinjiang Employment Promotion Project’ aimed to ‘broaden employment channels outside the home’ as part of ‘strengthening the poor masses’ inner development motivation’ (Xinyuan Government 2023).

VSETC-linked labour placements

The labour transfer and poverty alleviation documents analysed so far make no mention of VSETCs (internment camps) or their detainees (‘persons in re-education’ 教转人员). The VSETC system represents an evolution of Mao Zedong’s Re-Education Through Labor system which achieved neither effective psychological transformation nor profitable production (Zenz 2023b). VSETCs create a camp-to-labour pipeline where camps focus on re-education internment (without labour), followed by a gradual release process of short-term camp-based skills training, job training in nearby factories alongside evening re-education, and then coerced work placements in factory parks or further afield. Eschewing Mao’s model of re-education through manual labour, the VSETC system leverages non-camp factories to increase profitability and therefore the system’s long-term sustainability.

The earliest policy document on VSETC-linked labour placement is the ‘Notice on Further Improving the Autonomous Region’s Policy Regarding the Textile and Apparel Industry’ from April 2018 (Khotan City 2020). Here and in most other documents on the subject, the transfer of camp detainees into factory work is codified through the tifa of ‘Education Training Centers +’ (教育培训中心+). The plus symbol after VSETC terminology indicates camp-linked placement, nearly always into labour-intensive industries such as garment-making. By December 2018, Xinjiang’s Development and Reform Commission (XJDRC) noted that VSETC-linked labour had become an important ‘carrier’ for attracting ‘a large number of [eastern Chinese] enterprises to invest and build factories in Xinjiang’ (XJDRC 2018).

Numerous witness accounts confirm that VSETC-linked employment can constitute a severe form of forced labour, where workers are paid negligible wages, have no freedom of movement, labour for long hours under close surveillance and supervision, and are threatened with further internment for non-compliance (Amnesty International 2021, 126–29; Byler 2021b, 113–15, 119; Zenz 2023b).

While documents addressing VSETC-linked labour placements do not reference ‘transfer’ of ‘surplus labourers’, at the implementation level they are often presented in the context of poverty alleviation. A 2018 document issued by the Poskam (Zepu) county (Kashgar) Poverty Alleviation and Development Office lists ‘Poverty Alleviation Workshops’ as well as ‘VSETC Poverty Alleviation Workshops’ (Zepu Government 2018). While construction of regular poverty alleviation workshops was supervised by township governments, VSETC workshops were managed by county VSETC Departments. Camp ‘graduates’ were to be placed in workshops separate from those for transferred labourers, potentially due to greater perceived security needs. One camp witness placed in the same factory park as transferred labourers said she worked in a separate, more securitized factory with other former detainees, who slept in locked dormitories and worked in locked cubicles (Byler 2021b, 111; 2022b, 9).

The relationship between detainee and non-detainee work placements is complex. In some cases, detainees’ family members had to work in factories so that detained husbands could ‘get out quickly’ (Amnesty International 2021, 128). Based on witness interviews (albeit providing very limited details), Byler suggests that ‘some new industrial parks host a mix of detainees […] and “rural surplus labourers”’ through deliberate arrangements (Byler 2022b, 7). To demonstrate that co-placement constitutes a wider policy, he cites an article from the Institute for Economic Research at the XUAR Development and Reform Committee. While Byler assumes this article constitutes a policy, the institute is merely an advisory entity whose expert author was making policy suggestions only. The article suggests a potential collaboration framework between VSETCs and rural collective economic organizations (RCEOs), a special form of a wider rural economic organization category that can employ surplus labours (Institute of Economic Research 2019). While RCEOs could theoretically facilitate co-located employment of VSETC detainees and surplus labourers, neither this nor other documents I reviewed indicate this. The article does not mention surplus labour, RCEOs are rural and typically not located in factory parks, and planning documents discuss RCEOs and labour transfers in unrelated sections (XUAR Government 2021b).

An August 2018 Kashgar government notice states that 100,000 VSETC ‘trainees’ would be ‘shifted’ to training bases or industrial parks (Kashgar Government 2018). This was alongside separate skills trainings for 100,000 family members relatives of detainees, classified as ‘groups with socioeconomic difficulties’. For each county, these family members were to receive training and work placement in accordance with the numbers of released detainees. However, rather than confirming that both groups are placed in the same factory parks, as Byler suggests, the document indicates different work settings and sectoral allocations. VSETC detainees are to be placed into employment training bases, usually in factory parks, with 50,000 of them assigned jobs in textile and garment industries. By contrast, their family members are primarily sent to work in satellite factories near their homes, with only 20,000 assigned to textiles and garment-making. Whether this policy extends beyond Kashgar is unclear. Overall, the document affirms that former detainees and their poor rural (surplus) family members constitute distinct target groups, trained and placed according to distinct priorities.

In sum, both labour systems follow very different policy and implementation schemes and pursue at least partially different aims (Zenz 2023b). While both create a docile and inexpensive workforce, VSETC-linked work placements facilitate re-entry of detainees from diverse vocational backgrounds into society, whereas labour transfers convert farmers and pastoralists into industrial workers. The fact that at least some detainees can return to their former workplace underlines how camp-linked labour placements focus more on coercive integration of detainees into society than on sectoral transfers (Amnesty International 2021, 127). While former detainees are broadly assigned work in their home regions and usually face severely curtailed mobility, surplus labourers can be transferred to northern Xinjiang or eastern China. Whereas Xinjiang’s March 2019 white paper on counterterrorism defines VSETCs as dedicated ‘de-extremification’ vehicles, labour transfers and other socio-economic measures serve an auxiliary preventive function of limiting the soil in which ‘extremism’ can take root (SCIO 2019). No new policy or implementation evidence on camp-linked labour has emerged since early 2019, suggesting that new coercive labour victims are now likely linked to labour transfers or forms of prison labour (Zenz 2023b).

Conceptualizing labour transfer coercion as a risk continuum

While the coercive pressures inherent in VSETC-linked work placements are evident, Poverty Alleviation through Labor Transfer programme creates independent systemically coercive dynamics. Xinjiang and Uzbekistan present comparable cases of non-internment state-imposed forced labour (Zenz 2023a). My comparison shows that in both regions, coercive recruitments and transfers into state-mandated work placements operate through a centralized authoritarian state apparatus that steers economic policy and incentivizes or commandeers relevant economic actors. Both systems create a coercive social environment among targeted populations, and leverage substantial local human resources and institutions to develop mobilizational pressures at the grassroots level. My research further suggests that coercive mobilization consists of six phases: (1) identification of employment targets/quotas, (2) recruitment, (3) training, (4) transfer, (5) worker management and (6) worker retention (Zenz 2021a, 2023a).

In Uzbekistan, which ended forced labour in 2021, scholars described the resulting coercive environment as ‘structurally forced consent’, where the ‘absence of freedom to actively decide for or against taking part in [state-mandated work]’ is embedded and enacted from within the sociocultural context (McGuire and Laaser 2021, 560). This results in a ‘continuum […] of forms and degrees of unfreedom’ (Phillips 2013, 177). This framing is applicable in Xinjiang, where labour coercion is now much higher than it was in Uzbekistan. Interviews with predominantly female Uyghur self-organized rural–urban labour migrants between 2014 and 2017 showed that, while increasingly oppressive rural securitization campaigns drove them to leave, many also moved to urban centres to escape religious conservatism in traditional village environments (Tynen 2020; cf. Hann 2014). Urban labour incomes, though meagre, were higher than rural subsistence levels.

Income statistics confirm this. Income–earning potential by itself is not proof that work was chosen through free and informed consent, and even well–paid work can be forced labour. Rather, it differentiates labour transfers from camp–linked labour. In 2018, Khotan prefecture’s transferred labourers earned 7833 RMB on average (Tianshan 2019). Since these persons typically work for six to 10 months each year, this translated into monthly wages below Xinjiang’s 2018 minimum wage level of 1820 RMB per month (Xinhua 2020), despite being 2.6 times higher than that year’s national poverty threshold of 2995 RMB per year (Xilingol Government 2020). State claims about incomes of transferred labourers are often exaggerated and fail to reflect wage garnishments. Even so, official statistics show that over the period 2015–20, Xinjiang’s rural disposable per capita incomes grew much faster than urban incomes, especially because of drastic increases in wage income and other monetary transfers, such as social security payments and migrant worker remittances. Whereas per capita urban incomes grew by 24.9%, the rural equivalent grew by 88.8% (XUAR Statistics Bureau 2022a). Rural transfer payments, which exclude social security and pensions and therefore more directly reflect wage remittances, grew by 220.1% during that period (XSY 2016; XUAR Statistics Bureau 2022b). Moreover, intensified labour transfers have coincided with significantly expanded social security coverage for such workers (XUAR Government 2021c).

Clearly, transferred Uyghurs constitute cheap and easily exploited labour amid soaring nationwide labour costs. Chinese academics recommended that ‘eastern and central regions should have mandatory annual quotas of arranged Xinjiang labourers, which […] appropriately reduces labor costs’ (Chen, Zhang, and Shi 2019). Even so, poverty alleviation remains a significant, if subordinate, labour transfer goal: Xi argued that development is a prerequisite for, albeit not a guarantor of, social stability (Central Office Bulletin 2014, 6–7).

The ILO defines forced labour as work that is both (1) involuntary (without free and informed consent) and (2) enforced through a menace of penalty (ILO 2012, 14). While in Xinjiang the menace of penalty through internment or other sanctions is pervasive, the opportunity to leave the countryside and increase incomes means that involuntariness could theoretically vary. In practice, however, Xinjiang's extremely coercive environment means that choices are not ‘free’, transferred labourers are unable to freely leave employment, and none of these dynamics can be objectively evaluated through local audits or interviews, given that Uyghurs cannot speak freely. The resulting pervasive risk of state-imposed forced labour can only be effectively addressed through measures such as the UFLPA‘s rebuttable presumption.

Labour transfer coercion is therefore best conceptualized as a risk continuum. Those who actively choose against transfers are at highest risk, while those opting for them do so in a context of multiple unfreedoms. One study of labour transfer in Kargilik county (Kashgar) found that 43.6% of rural surplus workers were not yet employed in 2017 (AiSan 2019, 22). Of these, some 63% were women. The main reasons cited for unemployment included caring for young children or the elderly, or a general ‘unwillingness’. In response, the state instituted centralized caretaking programmes for children and the elderly (Chen, Zhang, and Shi 2019). Full employment is achieved not only by transferring Uyghurs to industrial sites, but also through village satellite factories with nurseries that ‘liberate’ women so they can work. In November 2017, Kargilik county planned to establish a satellite factory in every other village (United Front 2017). A poverty alleviation directive from Yarkand county mandates that village workshops be supervised by ‘at least’ one government official to ensure smooth operation and attendance (Shache County n.d.).

In sum, coercion along the risk continuum increases when employment measures target groups that had actively decided against full-time wage employment, and when they remove non-Han from organic community structures. These structures are targeted by ‘population optimization’ policies designed to ‘crack open solidified [Uyghur] society’ (Zenz 2021a, 14–15; 2021b). The 14th Five-Year Plans’ full employment requirements intensify these groups’ coercive risks.

Labour transfers and the risk of internment

Scholars have speculated that refusal to participate in employment campaigns renders Uyghurs liable for internment, although the evidence was tenuous (Byler 2021a, 18). A government list of 75 ‘expressions of religious extremism’ included ‘refusal to accept government management’, a vague term that likely includes poverty alleviation and employment measures (Phoenix Information 2014). Now, there is authoritative new evidence from the Xinjiang Police Files (Zenz 2022a).

A classified internal directive from Kashgar’s Prefecture Stability Maintenance Command in February 2017, immediately before the mass internments began, outlined criteria for subjecting key population segments to ‘strike-hard’ detention. These included persons who (1) ‘without reason are unwilling to receive various types of welfare-related policies’ (this would include most poverty alleviation measures); (2) ‘do not participate in grassroots organizational arrangements’ (these broadly include labour transfer programmes); or (3) ‘repeatedly refused employment opportunities provided by resettlement assistance institutions’ (resettling former prisoners) (Kashgar Stability Maintenance 2017). Such criteria are consistent with the 2018 revision of Xinjiang’s De-Extremification Regulation, which defines extremism, a label that makes one liable to re-education internment, as ‘speech and behaviour [that … ] rejects or interferes with normal production and life’ (Xinjiang People’s Congress 2018, art. 3).

Non-participation in state employment programmes also directly increased a person’s internment risk. An internal Tekes County directive (n.d.) to ensure full implementation of mass internments exhorted officials to ‘resolutely round up all who should be rounded up’. Everyone’s personal information was to be closely examined so that prospective detainees are ‘sifted out’ and ‘every single untrustworthy person is detained’. Among the filtering criteria were people’s ‘occupation and living conditions’, implying that those without stable or state-designated employment would face higher risk of detention. Uyghur informants confirmed that in 2017, participation in state policies was a strategy for escaping internment (Steenberg and Rippa 2019, 289).

Documented instances show local authorities wielding the internment threat in the context of poverty alleviation programmes. For example, a 2018 work inspection in Konasheher county uncovered that 659 poor households were accidentally given excessive resettlement benefits (Shufu County n.d.). To conceal their error, local authorities illegitimately ‘demanded immediate repayment of subsidies by threatening [beneficiaries] with being sent to the re-education center’, a tactic that ‘caused panic’ among affected households (7).

For the first time, this internment threat can be confirmed through direct local testimony. According to Gulzia (pseudonym), a Uyghur woman who was detained in a VSETC in southern Xinjiang between late 2017 and early 2019, two of her cellmates were detained for refusing to accept state-mandated work assignments.⁶ One of them, a woman from a rural township in Kashgar City, had two small children, a husband who worked in a factory all day, and was helping elderly in-laws with farming. When she refused a city factory job arranged by the local government, authorities detained her for harbouring ‘extreme religious thoughts’. Another woman in Gulzia’s cell was told by camp staff upon her arrival that she had been detained due to ‘non-cooperation with [government] arrangements’ (likewise a factory work assignment). By 2020, local authorities may have enforced such labour transfers without immediate internment, through a combination of mandatory centralized childcare, elderly care, and land transfers, as the structurally coercive capacities of poverty alleviation through labour transfers grew ever stronger.

The threat of internment does not alter the conceptual distinction between Xinjiang’s coercive labour systems. The Poverty Alleviation through Labor Transfer programme generates systemic coercive pressures without this threat, but such a threat significantly enhances them.

Conclusions

This paper has argued that Xinjiang’s two coercive labour systems are implemented under different policy frameworks, focus on different target groups, and achieve different (albeit complementary and partially overlapping) aims. Xinjiang's surplus labour transfers evolved from an in itself legitimate national socio-economic policy and now constitute part of a multi-faceted atrocity. After in 2014 Uyghur underemployment came to be framed as a threat to China’s national security, these transfers became highly coercive, concurrent with mass internments in 2017 and the poverty alleviation rectification drive in 2019.

The policy of placing VSETC detainees into work began in 2018 and likely concluded by 2020. By contrast, from 2021 and under the new party secretary Ma Xingrui, Xinjiang is consolidating the mass transformation of ethnic (non-Han) agriculturalists into factory workers through intensified monitoring, transforming a mobilization-driven labour transfer campaign into an institutionalized long-term mechanism (Zenz 2022c). This is accompanied by a fundamental strategic shift away from institutions of the state’s coercive (domestic security) apparatus, focused on re-education and internment, and towards entities overseen by the Development and Reform Commission, focused on poverty alleviation and economic development. Such transformation is at the heart of Beijing’s goal to enforce lasting sociocultural change. State-sponsored forced labour is difficult to detect at the company level, and instead requires a broader, society-wide analysis of recruitment and transfer mechanisms. The widely used 11 ILO forced labour indicators are ill-suited for this, making it easier for Beijing to sign ILO agreements and pretend that coerced non-Han employment is now a normalized and acceptable arrangement (Zenz 2023a). This is further reinforced by the fact that the camp-to-labour policy appears to no longer process new victims, whereas prison labour likely increased (Zenz 2023b).

Meanwhile, cheaper Uyghur labour helps offset rising nationwide labour costs. To prevent global consumers from becoming complicit in the atrocity through supply chain linkages, global policymakers need to understand the complexities inherent in Xinjiang’s coercive labour systems if they are to design accurate and effective countermeasures.

Acknowledgement

The author thanks the peer reviewers.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Byler (2021b, 110–15) basically adopts this tripartite scheme of Xinjiang’s forced labour systems and related material, albeit without citing my work (Zenz 2019).

2 During the period 2010–14 while conducting fieldwork in Kashgar, Rune Steenberg met Uyghurs who in his view voluntarily participated in labour transfers (conversation, April 2023).

3 Interview conducted by my research team in Turkey, early 2021.

4 Written communication, September 2022.

5 Interview by Rune Steenberg, May 2022, used with permission.

6 Interviews, September–October 2022.