Updated ILO Forced Labor Guidelines Directly Target Uyghur Forced Labor

Executive Summary:

• For first time since establishing its forced labor taskforce in 2001, the International Labour Organization (ILO) has issued authoritative and comprehensive guidance on operationalizing the research and measurement of forced labor, updating its more provisional guidelines from 2012.
• The new ILO Handbook adds a substantial new section on state-imposed forced labor, squarely targeting Beijing’s forced labor in Xinjiang and Tibet and specifically referring to “labor transfers” of ethnic minorities.
• The Handbook adopts the author’s category of “non-internment state-imposed forms of forced labor” in its research guidelines, significantly enhancing the ability to detect the forced labor of
• The Handbook notes that forced labor may be implemented by states for political reasons, including for reasons of “altering the population composition in particular areas.” This language points directly to labor transfers targeting Uyghurs.
• The Handbook’s statement that non-internment forced labor mobilization is best assessed as a risk rather than a specific instance, which it adopts from the author’s recent research, provides strong support for arguments that related legislation should reverse the burden of proof, shifting it from enforcement authorities to companies. This could have implications for the European Union’s upcoming forced labor legislation.

In 2001, the International Labour Organization (ILO) established its Special Action Program to combat Forced Labour (SAP-FL) to develop survey guidelines and indicators for assessing forced labor worldwide. In 2012, this resulted in the first systematic attempt to establish a forced labor measurement framework, a set of Survey Guidelines titled “Hard to see, harder to count” (ILO, 2012). The 2012 Guidelines had several limitations, however. Their technical nature was designed to inform national forced labor surveys, making them largely inaccessible to non-experts. It also came with conceptual limitations, having been primarily designed to counter private forced labor. Finally, the Survey Guidelines were a work in progress, given that the ILO’s efforts were still in their infancy, and therefore did not claim to be authoritative (Journal of Human Trafficking, 2023).

This led to the European Commission using a much more widely known ILO framework, the set of 11 forced labor indicators, in September 2022 draft legislation prohibiting the import of products involving forced labor, widely understood to target forced Uyghur labor in the People’s Republic of China (PRC) (European Commission, September 14, 2022). These indicators were however designed not for formal measurement but to help frontline inspectors recognize potential signs of forced labor at workplaces (ILO, 2012). In conversations with the author, the ILO noted that they had “a communication problem” about the roles of their various indicator frameworks.

The ILO’s 11 indicators are ill-suited to identify what the author has referred to as “non-internment state-imposed forced labor mobilization.” This form of forced labor mobilization, which characterizes forced labor transfers in Xinjiang and Tibet, is a dynamic process whereby states identify a target population for forced work, and then coercively mobilize, recruit, and train them, before transferring them to designated workplaces (Central Asian Survey, 2023; China Brief, September 22, 2020). Workplaces are not necessarily heavily secured, and the process does not involve prisons or labor camps, rendering this type of forced labor less visible and harder to measure. The ILO encountered this challenge when attempting to measure forced labor mobilization for cotton harvesting in Uzbekistan (ILO, 2022). It tried to interview workers in the cotton fields (i.e. at their place of work), but struggled to find specific signs of coercion, besides the fact that workers gave highly standardized answers, suggesting that they had been coached by the state to mislead ILO evaluators (Journal of Human Trafficking, 2023).

In 2023, the author interviewed several senior ILO officials (Journal of Human Trafficking, 2023). The author challenged them about the facts that:

1. their 2012 Survey Guidelines were not designed to capture state-imposed forced labor in the PRC and elsewhere;
2. these Guidelines contained indicators that were only partially suited to capture Uyghur forced labor;
3. these Guidelines were little-known compared to the widely known set of 11 indicators;
4. the ILO’s overall approach to forced labor measurement was unequipped to capture non-internment (i.e., non-prison) forms of forced labor mobilization, which are harder to conceptualize and measure; and
5. the ILO’s global reports on forced labor have been far too optimistic concerning claims of a decline of state-imposed forced labor worldwide.

As a result of this engagement, the ILO told the author in 2023 that they would update their Guidelines. This update was published in late February this year. The results are highly impressive. The ILO appears to have carefully considered all feedback, as well as input by advocacy organizations such as Anti-Slavery International, and adopted improvements on all five points.

The New Guidelines

In February 2024, the ILO published a “Handbook,” updating its 2012 Guidelines “Hard to see, harder to count” (ILO, 2024). Here is a rundown of the key changes from a PRC-focused perspective, each of which is then discussed in greater detail:

1. The new guidelines are now presented in the form of an authoritative “Handbook.”
2. This Handbook contains four mentions of “labor transfers” that target minorities for forced relocation. This particular expression is only commonly used in PRC contexts and the way it is defined in the Handbook points directly to forced labor in Xinjiang and Tibet.
3. The Handbook contains an expanded indicator framework, with dedicated sections for indicators of state-imposed forced labor.
4. The most important change from a China perspective is the addition of a dedicated section on measuring state-imposed forced labor (section 9).
5. This new section 9 contains an important subsection titled “General research considerations,” which outlines the challenges of assessing “non-internment state-imposed forms of forced labor.” This subsection draws from and builds on the conceptual work of the author (Journal of Human Trafficking, 2023).
6. Section 9 states that forced labor mobilization may be implemented by states for political or national security (and not just economic) reasons, making it even more diagnostic of Uyghur and Tibetan forced labor.
7. Section 9 also discusses forced labor linked to administrative detention and “re-education” without legal conviction, language that targets Xinjiang’s camps (and similar situations worldwide).
8. The Handbook strengthens the role of desk research, explicitly referring to academic research and NGO reports.

The new Handbook is more authoritative than the previous Guidelines. In the introduction, it presents itself as an integrated, authoritative, and “updated measurement framework and set of tools for the design, implementation and analysis of surveys of forced labor” (p.ix). It highlights the fact that it contains “a new unified set of core indicators of involuntary work and coercion,” representing the new ILO office-level standard for forced labor measurement. The Handbook is a unified update of both the previous Survey Guidelines and the Guidelines Concerning the Measurement of Forced Labour endorsed by the 20th International Conference of Labour Statisticians (ICLS) in 2018, further underscoring its authority (ICLS, 2018).

Previously, it was not clear which ILO document constituted the “standard” for measuring forced labor as defined by ILO Conventions 29 and 105. The adoption of the 2018 ICLS document was an important step in standardizing the ILO’s approach to measuring forced labor. But it rendered the 2012 Survey Guidelines somewhat outdated, further adding to the numbers of documents that evaluators and researchers had to account for. While the new Handbook is an “office document” that has not been formally approved by the ILO governing body, it still constitutes an important development as it represents the world’s most authoritative framework for operationalizing the measurement of forced labor.

Specific References to “Labor Transfers”

The Handbook uses the phrase “labor transfers” four times. While describing a global phenomenon, this particular expression is essentially only used in PRC contexts (Central Asian Survey, 2023). The Handbook’s definition therefore points unambiguously to forced labor in Xinjiang and Tibet. Two uses occur in section 1.3 that outline the updated indicator framework when discussing indicators for coercive recruitment in the context of state-imposed forced labor for economic development (ILO, 2024, p.11):

Other violations involve large-scale labor transfer schemes, where workers belonging to certain ethnic or religious minority groups must—under menace of penalty—relocate to another geographical area to work in a State or private enterprise, sometimes under guise of vocational training or regional economic development.

This phrasing links labor transfers specifically to (1) ethno-religious groups, (2) geographical relocation, and (3) work in enterprises. The advantage of this framing is that it specifically points to forced Uyghur labor. However, there are several disadvantages. First, PRC coercive labor transfers could also target impoverished Han Chinese. Second, transfers also occur through satellite factories in rural villages, in which case they do not involve geographic relocation (see Journal of Political Risk, 2019). Third, labor transfers can involve non-enterprise work destinations (China Brief, February 14). It is a common misconception that transfers must “transfer” people across space, whereas the term “transfer” refers to a sectoral transfer—from farming or herding to wage labor in the industrial sector, the service sector, or wage labor within agricultural processing or seasonal harvesting (CPCS, 2023). A third disadvantage is that labor transfers can involve non-enterprise work destinations.

All ILO forced labor measurement guidelines and documents explicitly state that measurement categories and related indicators must be adjusted to local contexts, meaning that the wording is not a straitjacket. In addition, the Handbook clearly distinguishes labor transfers from forced labor linked to re-education camps (Handbook sections 9.3 and 9.4; both systems of forced labor are discussed in detail in Central Asian Survey, 2023).

The two other instances of “labor transfer” are found in section 9.4, which discusses state-imposed forced labor for the purpose of economic development:

Such [labor transfer] schemes can result from the combination of various methods of compulsion to work: measures of general nature involving compulsion in the recruitment, assignment and transfer of labor, used in conjunction with other restrictions on freedom of employment, such as preventing workers from terminating their employment contracts or compulsorily extending contracts, penal sanctions for breaches of contract or as a means of maintaining labor discipline, restrictions on freedom of movement or on the possession and use of land, or abusive application of vagrancy laws (ILO 2007, para 107). (p.168)

This links labor transfers not only to coercive mobilization but also to retention mechanisms designed to maintain employment, such as Unemployment Monitoring and Early Warning and other poverty-fallback prevention systems. These have become increasingly important in recent years as the region has transitioned from the campaign-style approach of Xinjiang’s former CCP Party Secretary Chen Quanguo (陈全国) to a focus on institutionalizing coercive transfer policies by his successor Ma Xingrui (马兴瑞) (China Brief, June 5, 2022). In addition, the reference to restrictions on the “use of land” (taken from a 2007 International Labor Conference report) is highly significant for assessing PRC coercive labor transfers, which frequently enforce transfers of land-use rights from targeted ethnic groups to large operators or government cooperatives, subjecting the now landless Uyghurs more readily to labor transfers (China Brief, March, 2021; CPCS, 2023).

The New Indicator Framework

The Handbook provides an updated indicator framework. The previous indicator framework consisted of a six-field matrix: two columns for the two dimensions of forced labor per definition in ILO Convention 29, and three rows for the three employment cycle phases: recruitment, work conditions, and ability to leave work (see Journal of Human Trafficking, 2023). The 2024 Handbook simplifies the framework to a four-field matrix, eliminating the phase distinctions for the coercive dimension.

The 2012 version of the framework was only designed for private (company-based) forced labor. The Handbook now adds sections and indicators for state-imposed forced labor (SIFL) for each of the three employment cycle phases both for involuntariness and for the menace of penalty (coercion) dimension. This addition is crucial. However, the new framework imposes an unnecessary limitation by restricting SIFL indicators to particular forms of state-imposed forced labor, forms that differ for each phase. For example, indicators linked to prison labor are relegated to the second phase (“employment”), while those linked to forced labor mobilization for economic development—the type of forced labor affecting Uyghurs and Tibetans—are limited to the first phase (“recruitment”). While forced labor transfers are best measured at the recruitment stage, Xinjiang is now enforcing Uyghur work quotas by preventing them from leaving work, using mechanisms such as the new Unemployment Monitoring and Early Warning system (China Brief, June 5, 2022). The association of employment cycle phases to particular forms of state-imposed forced labor therefore imposes an unnecessary restriction.

The coercive dimension adds indicators that are essential for capturing forced labor from coercive labor transfers in Xinjiang and Tibet, namely the imposition of punishments such as detention or imprisonment for refusing state-assigned work.

The New Section on State-Imposed Forced Labor

The Handbook’s most important improvement is the addition of a 23-page section on state-imposed forced labor (section 9). After defining the aspects of state-imposed forced labor following ILO Convention 105, section 9 presents a crucial innovation—a section titled “General research considerations.” This section notes that:

In contrast to most forms of forced labor, state-imposed forced labor operates through a pervasively coercive wider social context marked by a general lack of civic freedoms and a state apparatus that generates powerful coercive pressures through an extensive grassroots apparatus consisting of state and non-state institutions. Non-cooperation entails a systemic risk that is often more implicit than overt (p.149).

This conceptual framing of state-imposed forced labor, adopted from the author’s recent research, encapsulates why forced labor is difficult to conceptualize and even harder to measure (Journal of Human Trafficking, 2023, p.21-22). By making this a research consideration for all forms of state-imposed forced labor, the ILO ensures that the elusive character of this unique form of forced labor—ubiquitous in Poverty Alleviation through Labor Transfer policies in Xinjiang and Tibet—receives due attention (Central Asian Survey, 2023).

The section then continues to list several considerations for research on what it refers to as “non-internment state-imposed forms of forced labor,” a phrase based on one coined by the author to capture the nature of Xinjiang’s forced labor transfers (Journal of Human Trafficking, 2023). Perhaps the most important of these considerations is that this form of forced labor may be measured by assessing “evidence of a state policy” linked to coerced work (p.150; Journal of Human Trafficking, 2023, p.22). In the absence of direct witness testimony or the ability to conduct on-site research in regions such as Xinjiang or Tibet, policy analysis is the only way to demonstrate the existence of coercive labor practices and is how the first systematic research demonstrating the existence of Uyghur forced labor was conducted (Journal of Political Risk, 2019). In addition, the Handbook agrees with the author’s conclusion that non-internment state-imposed forced labor is best assessed during mobilization, given that coercion may be much less visible at workplaces (p.150; Journal of Human Trafficking, 2023, p.22).

The Handbook notes that political aims may motivate forced labor policies, as is primarily the case in Xinjiang and Tibet. When reviewing the 2012 Survey Guidelines, the author had noted that the ILO’s measurement frameworks were mainly geared toward capturing forced labor motivated by economic exploitation, which is the most common motivation for coerced work. The Handbook takes this critique into account, which the author had also voiced in encounters with ILO officials. It now states that state-imposed forced labor may be assessed through:

evidence of a state policy that instrumentalizes employment or work for political objectives such as aligning political views with those of the established political, social or economic system, altering the population composition in particular areas or enhancing national security.

Especially noteworthy is the reference to “altering the population composition.” This refers to state policies to “optimize” ethnic population ratios, especially in southern Xinjiang’s Uyghur heartland by ending the “dominance” of Uyghur population groups and increasing numbers of Han Chinese settlers (Central Asian Survey, 2021). In 2017, the central government mandated an increase in southern Xinjiang’s settler population by 300,000 by 2022, a strategy that largely relied on attracting Han Chinese from other parts of the PRC through promises of free land, housing, education, and government jobs (China Brief, March, 2021). PRC scholar-officials have confirmed that labor transfers to other regions in Xinjiang or to other PRC provinces serve to reduce the density of concentrated Uyghur populations (China Brief, March, 2021). Recent research has found that transfers to other provinces were set to increase by 38 percent in 2023 (China Brief, February 14). In conversations, ILO officials had cited the author’s article in Central Asian Survey (2021) as an example of how labor transfers directly relate to other aspects of the atrocity. The inclusion of this observation in the Handbook is very significant, as it points to major non-economic reasons for Xinjiang’s forced labor systems, and to ways in which state-imposed forced labor can reinforce other oppressive policies.

The new section 9 discusses four forms of state-imposed forced labor, each of which is complemented by a section discussing related research considerations:

1. Abuse of military conscription (section 9.1)
2. Exaction of work beyond normal civic obligations (section 9.2)
3. Abuse of compulsory prison labor (section 9.3)
4. Compulsory labor for the purpose of economic development (section 9.4)

Section 9.3 of the Handbook contains language that points directly toward forced labor associated with Xinjiang’s re-education camps (The China Journal, 2023):

Compulsory labor cannot be imposed on people in administrative detention. […] A common violation is the imposition of compulsory labor on those detained by law enforcement authorities but never meant to be tried by an independent judiciary because they have committed “minor offences”—those that are not serious enough to be subject to criminal prosecution but deemed errant enough to qualify for “re-education” (p.160).

The Role of Desk Research and NGO Reports

The strengthening of the role of desk research and an increase in the list of recommended sources to assess forced labor is another important change in the Handbook for identifying Uyghur forced labor. In its “desk review” section, the 2012 Survey Guidelines referred to preparing forced labor research based on a “review of court cases, law enforcement data and other relevant sources” (ILO, 2012, p.45). In contrast, the 2024 Handbook advises consulting “academic articles, reports from prior national or sectoral surveys, programme evaluations, qualitative studies by NGOs or civil society groups, and investigative journalism reportage” (p.25). These additions strengthen the contribution made by non-official sources, which in the case of Uyghur forced labor have played an essential role.

The section “General research considerations” for state-imposed forced labor is of further significance. It specifically points out how information collection in such settings can be a major challenge: “As the State itself imposes this category of forced labor, States may have little incentive to collaborate with or facilitate the work of researchers wishing to shed light on it. Information access can be problematic” (p.151).

Implications for the Proposed EU Forced Labor Ban

Non-internment state-imposed forced labor mobilization is especially challenging to conceptualize and evaluate. The 2024 Handbook accounts for these difficulties in the “General research considerations” for assessing this type of forced labor, which provide crucial context in line with the unique properties of coercive labor transfers. Of particular significance is the statement that “non-internment state-imposed forms of forced labor … [are] more readily assessed as a systemic risk than a specific instance, given that this form of forced labor creates an environment that renders its victims much less likely to speak freely” (p.150; compare Journal of Human Trafficking, 2023, p.21).

This research guidance directly suggests that this form of coerced work creates a society-wide systemic risk. This framing, which the ILO adopted from the author’s work, effectively means that the burden of proof of forced labor rests on those who participate in the affected economic system or are connected to it through their supply chains.

This could have interesting implications for the European Union’s forced labor regulation. The version of this regulation proposed by the European Parliament contained a crucial provision to reverse the burden of proof of the existence of forced labor, shifting it from the authorities to the importing entities, akin to the US Uyghur Forced Labor Prevention Act or UFLPA (European Parliament, 2023). This reversal is essential for capturing goods made in the context of state-imposed forced labor, as without it, any legislation would result in severe underenforcement (Journal of Human Trafficking, 2023). However, during the negotiations between the Parliament and the Council, this crucial provision was removed, in line with the version adopted by the Council which privileged the economic interests of member states (European Council, March 5, 2024; European Council, March 13). [1] As a result, it is unlikely that the new measure can effectively counter Uyghur forced labor (UCANews, March 6).

The 2024 Handbook could play a major role here. To trigger new investigations into suspected forced labor cases, the proposed legislation already suggests a so-called “risk-based approach.” This is centered around a database of known cases of forced labor, or in the case of state-imposed forced labor, of affected regions or sectors. It uses “substantiated concern” as the evidentiary threshold (European Council, March 13; Crowell, March 26). However, in the course of such investigations the burden of proof remains on the investigating “competent authority,” in contrast to the UFLPA (European Council, accessed April 17).

Under section 47, the agreed version of the legislation text states:

Where in response to a request for information from a lead competent authority, an economic operator or a public authority refuses or fails, without a valid justification, to provide information requested, provides incomplete or incorrect information with the objective of blocking the investigation, provides misleading information or otherwise impedes the investigation, including when a risk of forced labor imposed by state authorities is identified, the lead competent authority should be able to establish that the prohibition has been violated on the basis of any other relevant and verifiable information gathered during the preliminary phase of the investigation and the investigation (European Council, March 13, p.25).

Without free and unfettered access to Xinjiang’s factories and affected Uyghur workers, the Commission’s ability to demonstrate forced labor is severely limited. This is where the new ILO Handbook becomes relevant. By arguing that specific instances of forced labor in state-imposed contexts cannot be reliably measured because of the very nature of this type of coerced work, the Handbook’s “General research considerations” provide further authoritative guidance for the interpretation of this crucial section of the legislation, which has tied itself to ILO standards that the Handbook operationalizes. Specifically, the EU’s investigating authority could easily interpret section 47 in tandem with the new ILO Handbook to argue that the presence of forced labor is established by the existence of a relevant state policy, its enforcement on the ground, and the general inability to assess specific instances of forced labor, as outlined in the Handbook’s “General research considerations.” Even though this would not reverse the burden of proof, the investigating authority could then determine the presence of forced labor simply based on (1) a relevant database entry that documents the prevalence of forced labor in Xinjiang based on existing research reports, and (2) a demonstrated connection of an imported good with supply chains linked to Xinjiang.

In so doing, the EU forced labor ban could operate in a more similar fashion to the UFLPA in regard to Xinjiang (although in contrast to the UFLPA, it lacks a preventive detention mechanism). Europe could then potentially avoid its current fate of being a dumping ground for goods made with forced labor from Xinjiang (SCMP, March 21). 

Notes

[1] After the Council approved the new text in March, the Parliament is scheduled to vote on it on April 22 (European Parliament, April 11).