The muddled governance of state-imposed forced labour: multinational corporations, states, and cotton from China and Uzbekistan

ABSTRACT Forced labour persists in our global economy despite dedicated attention and eradication efforts from both the public and private sectors. Given the bounded reach and lack of enforcement by states and international organisations, the private sector has been the linchpin for eradication of forced labour globally. Utilising the brand-to-state boomerang model, this paper examines state-imposed forced labour in cotton production in China and Uzbekistan, and grapples with how interests – those of the states and the multinational corporations involved in forced labour – shape private governance outcomes. By investigating state-imposed forced labour in China (specifically, the Xinjiang Uyghur Autonomous Region) and Uzbekistan, we find that multinational corporations are reluctant to work towards eradicating forced labour when their net sales and profit are threatened by doing so. Building on the stream of international political economy research regarding how interests complicate governance effectiveness, we expose a gap in the literature on the impact of state-imposed forced labour on governance outcomes and illuminate global ramifications.


Introduction
Forced labour is prohibited by the International Labour Organization (ILO) 1998 Declaration on Fundamental Principles and Rights at Work, United Nations Guiding Principles, United Nations Global Compact, Organisation of Economic Cooperation and Development (OECD) Guidelines for Multinational Enterprises, numerous domestic laws, and voluntary standards. Yet, forced labour persists in our modern global value chains (GVCs). As defined through the 1930 ILO Convention, forced labour is: All work or service which is exacted from any person under the menace of any penalty and for which the said person has not offered himself voluntarily. (Art. 2.1) The ILO identifies 11 indicators of forced labour: abuse of vulnerability, abusive working and living conditions, physical or sexual violence, restriction of movement, deception, debt bondage, withholding of wages, excessive overtime, retention of identity documents, intimidation and threats, and isolation (ILO 2012). State-imposed forced labour regimes may involve any of these indicators and serve as a form of intimidation to workers on their own (Better Cotton Initiative 2021, ETI 2019).
Forced labour is widespread. According to the ILO, 27.6 million people are in forced labour, of which 17.3 million people are exploited in the private sector, 6.3 million persons in forced sexual exploitation, and 3.9 million persons in forced labour imposed by state authorities (ILO 2022). Governments, academics, and third-party organisations are devoting more and more attention to eradicating forced labour globally and the role of multinational corporations (MNCs) therein (Jones and Tiffen 2020, LeBaron and Lister 2021, Phillips and Mieres 2015. In this paper, we concern ourselves with private governance effortsefforts by MNCs to eradicate forced labour through Corporate Social Responsibility (CSR) policies and multistakeholder initiatives. There is now emerging evidence that forced labour falls within the spectrum of labour exploitation within GVCs, and this paper evaluates forced labour and governance according to that school of thought (Crane 2013, LeBaron 2018, McGuire and Laaser 2018, Phillips and Mieres 2015. Our study specifically examines state-imposed forced labour in cotton production that implicates MNCs. Because cotton comprises a large share of global forced labour incidences (Bureau of International Labor Affairs 2021), studying forced labour eradication efforts in cotton is particularly critical.
In our study, we grapple with how the interests and actions of the states and MNCs involved in forced labour shape governance outcomes. We examine cases of forced labour imposed by two states: China and Uzbekistan. Exploring the unique contexts of these two cases of state-imposed forced labour and the variation between them illuminates unique influences on MNC efforts to eradicate forced labour. Although MNCs have made efforts to end the use of forced labour in their GVCs through CSR policies and multistakeholder initiatives, forced labour remains ongoing in China while it has been eradicated in Uzbekistan. An important gap in the international political economy literature surrounds the uniqueness of state-imposed forced labour and its impact on private governance outcomes, as the emphasis has been on how forced labour arises in and is perpetuated by the market (Crane 2013, LeBaron 2020, Phillips and Mieres 2015. This paper focuses on the particularly novel phenomenon of how state-imposed forced labour can influence the willingness of MNCs to act socially responsible. By examining state-imposed forced labour and MNCs' efforts to eradicate the practice, we apply the brand-to-state boomerang model as developed by . Boomerang politics characterises the impact of activism and the resulting private governance efforts on desired outcomesusually, improvements in human rights or environmental standards (Keck and Sikkink 1998). By testing the brand-to-state boomerang model as developed by  in two cases, we find that the willingness of the MNCs to act socially responsible to eradicate forced labour is affected by states' varying levels of ability and determination to contest the existence of forced labour. When a state with a massive consumer market strongly disputes the presence of forced labour, MNCs are disincentivised to implement CSR measures as their sales and profit in the disputing state are threatened. Thus, private governance outcomes often depend on the degree to which forced labour is contested and the threat that contestation poses to the profit of MNCs.
The remainder of our paper is divided into six sections. The first three sections are the theoretical building blocks of the paper. The first situates our paper in a broader context of forced labour in the global economy; we map out the importance of studying forced labour in cotton GVCs and discuss the role of both the private and public sector in eradicating forced labour or allowing the continuation of forced labour. It will become especially clear that MNCs play an important role in this process, and that their actions to improve labour standards are hugely dependent on how they perceive their profit to be affected, for example by state behaviour. The second theoretical section introduces the model that we use to depict and understand how different public and private actors shape the governance of forced labour, and the third theoretical section outlines the process tracing methods used to examine the two cases. Following these three theoretical sections, we turn to our case-study investigations of China (the Xinjiang Uyghur Autonomous Region) and Uzbekistan. The very last section of our paper is a conclusion, in which we summarise and discuss our main findings.
Forced labour and private governance in the modern global economy

Forced labour in cotton GVCs
According to data on forced labour from the U.S. Department of Labor, cotton is the product produced with the highest incidence of forced labour; following cotton are gold, sugar, cocoa, coffee, and bricks, among others (Bureau of International Labor Affairs 2021). Because it makes up a significant share of global incidences of forced labour and because it flows through complex GVCs from raw material to final product, cotton is highly relevant for study on efforts to eradicate forced labour. Most contemporary trade in the global cotton industry takes place between lead firms and suppliers in GVCs (Anner 2018). In 2015, somewhere between 40 and 50 per cent of global trade flowed through GVCs (World Bank 2020).
GVC analysis is central to understanding the different actors and their interaction, value capture, and governance in a given sector. Specifically, GVC analysis of governance depicts the level of control a MNC, or lead firm, has over the production processes of its suppliers toward the beginning of the GVC (Gereffi 1994, Gereffi and Fernandez-Stark 2016. GVC analysis is not the only framework offering a useful way to think about economic networks in the global economy. For example, the global production network (GPN) approach (Coe and Yeung 2015, Henderson 2002, Hess and Yeung 2006 has emerged as a notable branch of the GVC literature, defining networks to encompass more actorssuch as geographical territories and statescompared to the firm-centric focus common to the GVC approach (Behuria 2020, Castañeda-Navarrete et al. 2021. However, our analysis is mostly grounded in GVC analysis as this literature has emerged as tremendously rich and useful for understanding governance structures in the global economy, the power of firms in dictating standards in their GVCs, and the business interests of MNCs. MNCs' business interests are defined on performance and image dimensions, such as minimising risk to profitability and reputation . GVC analysis is especially relevant for cotton, as the framework provides useful illustrations of how cotton production is organised globally from farmer to final productyarn, fabric, or garmentsin a complex web of intermediaries. See Figure 1 for a visualisation of a typical cotton GVC. As depicted in Figure 1, a typical cotton GVC has many different stages from farm to garment, and in our contemporary global economy, each stage can take place in a different state. With regard to cotton produced with forced labour in China and Uzbekistan, the presence of forced labour is at the beginning of the GVCparticularly at the farming, aggregating, and ginning stages (Lehr andBechrakis 2019, Schweisfurth 2020). For this study, GVC analysis shows how MNCs selling garments are implicated in forced labour and illustrates the sizable control lead firms have over suppliers. In particular, garments have been widely studied and have been shown to fall into the 'captive' governance type, meaning lead firms wield much power and control over their suppliers and inputs (Mayer and Gereffi 2010). However, this power and control manifests in conflicting pressuresfor efficiency and for better labour standards; these conflicting pressures constitute a limitation of private governance in achieving better labour standards.

Forced labour and business pressures
Much of the literature on labour rights and standards in GVCs details the lead firm leverage over suppliers, which in turn leads to labour exploitation (Crane 2013, LeBaron 2021a, McGuire and Laaser 2018, Selwyn 2019, Sturgeon 2001. Examples of competitive business pressures that yield to lead firm leverage and harm to the workers are cost-minimisation, value capture, short delivery time, and subcontracting, among others (ILO 2019, LeBaron 2020, Milberg andWinkler 2011, Sturgeon 2001). Recent econometric research has found that changing the efficiency-seeking strategies of cost minimisation in the GVC in emerging markets may help improve labour standards (McGahan and Distelhorst 2019). Further, multistakeholder initiatives comprised of any coalition of businesses, non-governmental organisations (NGOs), and/or investors, are found to be susceptible to the same pressures as MNCs, namely cost-minimisation, profitability, and financial conflicts-of-interest (Lund-Thomsen 2021).
Because this paper considers forced labour on the spectrum of labour exploitation, these business pressures can lead to conditions sustaining forced labour (Crane 2013). Meanwhile, linking up to GVCs can be an important feature of economic development and growth, so many states are working to make their business environments attractive for these linkages (Chang and Andreoni 2020, Gereffi and Fernandez-Stark 2016, Graham and Woods 2006, Hauge 2020. Considering both these business pressures and economic development goals is important when thinking about global cotton production, as they showcase some limitations of private governance. GVCs contain contradictory elements. They offer useful avenues for development through growth in trade and investment among countries that take part in them as highlighted well by World Bank (2020). Conversely, the expansion of GVCs have enabled MNCs to strengthen their power in the global economy, and often involves labour exploitation and immiseration at the bottom rung of GVCs through downward pressure on prices and value capture (Selwyn and Leyden 2022). In this paper, we do not dismiss the value of GVC participation for development, but certainly try to highlight the labour exploitation that characterises many GVCs.

Public and private governance of forced labour
Labour governance is the public and private regulations, standards, norms, rules, and actions that influence labour standards in the global economy (LeBaron 2020). As noted above, international organisationssuch as the OECD, United Nations, and ILOand states prohibit forced labour. However, because international organisations focus on decentralisation and state regulation of labour, definitions about how liable lead firms are for labour violations in their GVCs are indeterminate (Černič 2018). For example, the ILO delegates power to the state to regulate their labour markets, while the structure of the global economy has changed to largely consist of dispersed production through GVCs. Further, the ILO largely constructs its remediation programmes in cooperation with the state involved (Phillips and Mieres 2015). Thus, states can prevent ILO programmes from operating, if desired. In other words, the reach of public labour governance are the boundaries of the domestic market, which is out of line with the modern global economy in which capital moves freely and the production of a single good takes place in many different countries through GVCs (Alexander 2019, Phillips 2013, Ponte and Sturgeon 2014. Meanwhile, states have mostly acted through due diligence legislation, or legislation that codifies self-regulation by MNCs to ensure goods imported into the domestic market are not tainted by forced labour (Feasley 2015, Greer and Purvis 2016. There are increasing varieties of due diligence legislation with notable examples in the UK, US, EU, and Brazil (Phillips et al. 2018). The general consensus in the literature is that non-mandatory due diligence legislation does not help to eradicate forced labour, as there are no clear guidelines for what disclosures MNCs must provide regarding efforts to eradicate forced labour, no enforcement or penalties for non-compliance, and no inclusion of rules for labour recruitment intermediaries in GVCs (LeBaron 2020, Phillips and Mieres 2015, Smit 2021). Thus, MNCs have strategic control over the design and implementation of their due diligence measures (LeBaron 2020). So, within a broader institutional environment of soft international and national policies, rules, and regulations, private governance through CSR is the linchpin of labour governance through the coordination and self-regulation of the GVC by the lead firm (perhaps pressured by consumer or human rights activism) (Stringer and Michailova 2018).
Thus, the focus of this paper is on private governance in the form of CSR, as mobilised alone and/ or through multistakeholder initiatives and third-party programmes, and particularly MNC's structural power to impact outcomes befitting our use of the brand-to-state boomerang model. Structural power confers 'the power to decide how things shall be done, the power to shape frameworks within which states relate to each other and to people' (Strange 1994, p. 24). While an institutional view suggests the eradication of forced labour occurs because of regulation and normative values, forced labour has structural inertia that deflects these institutional forces (Crane 2013). MNCs' structural power tends to promote soft law, rather than binding international agreements (LeBaron 2020). In sum, MNCs not only exercise instrumental power and agenda-setting power to bolster their own business practices (Fuchs 2005), but also act as governing institutions themselves, both alone and collectively (May 2015). This is particularly evident in the area of private governance of GVCs by MNCs, multistakeholder initiatives, and third-party programmes, and has ramifications for the eradication of forced labour globally.

Private governance effectiveness
There is mixed evidence regarding whether MNCs governing their GVC through self-regulation actually improve labour standards, given the conflicting business pressure for efficiency (Alexander 2019, LeBaron and Lister 2021) and the distance from lead firm to supplier in a GVC (Locke 2013).   developed a framework to illuminate the process and conditions under which MNCs pressure a state to have better labour standards through private governance. In this study, the focus is less about how better labour standards are achieved, but rather how interests shape private governance, how private governance is influenced by states, and the corresponding influence on the outcome. Broadly, the question becomes how does stateimposed forced labour impact CSR efforts and outcomes. Through case studies of state-imposed forced labour in China and Uzbekistan, a political dilemma facing MNCs regarding their net sales and profit (occurring when a state has the ability and determination to contest the presence of forced labour) is shown to be particularly influential on the outcome of eradication of forced labour.

State-imposed forced labour and its influence on eradication efforts by MNCs
The brand-to-state boomerang model A boomerang model depicts the influence public, private, and social actors have on governance outcomes. The original boomerang model developed by Keck and Sikkink (1998) specifically considers how governance actors, particularly civil society and MNCs, influence state-to-state negotiation for certain outcomes. A modified form, the brand-to-state boomerang, explains the influence brands or MNCs can have on a state to attain certain outcomes (den Hond andde Bakker 2012, Seidman 2007). We apply and build upon this boomerang to study MNCs' willingness to eradicate forced labour from their cotton GVCs in the Xinjiang Uyghur Autonomous Region (XUAR), China and Uzbekistan. 1 The brand-to-state boomerang can be conceptualised through a three-stage model, as developed by . In the framework, issue salience is the negative impact on brand reputation and sales (because consumers will shop elsewhere) that the labour standards violation brings to an MNC. However, in our application, the contestation by the host state of the labour violation adds nuance to issue salience because of conflicting pressures from home and host states, as will be elaborated in the next subsection. The issue salience then informs the deliberateness of the MNC response. The following collectiveness of the private governance action is enabled through the mobilisation structure, i.e. the number of MNCs, the multistakeholder initiatives, and the third-party organisations taking action. Collective action is necessary for influence on the outcome because of the large number of MNCs with cotton GVCs and MNCs taking on the higher costs of remediation alone harm their competitiveness, i.e. a collective action problem. As necessary for boomerang politics, MNCs must have GVC operations in the state and take on a governance role, 'Brand advocacy requires the presence of brands willing to assume a political role vis-à-vis government' (Oka 2018, p. 99). The political context of the state and resource dependence 2 of the host state on cotton production determines the power relationship between the MNCs and the host state. Overall that political context mediates the MNC response to the labour violation and the collective action to remediate the labour violation. These factors influence the private governance outcome, as visualised in Figure 2. Through our case studies, we test and tease out nuance with regard to the level of state contestation to the three-stage brand-to-state boomerang model and specify the conditions for MNC leverage and influence on outcomes. This three-stage model informs the structure of our case studies regarding governance efforts by MNCs in the XUAR and Uzbekistan. First, we discuss the political context and resource dependence, as this provides necessary background on the state-imposed forced labour regimes and their contestation of the presence of forced labour. Second, we discuss the issue salience and MNC response, and third, the mobilisation structures and collective action. Given that forced labour has been eradicated in Uzbekistan and remains ongoing in the XUAR, utilising and building upon the boomerang model expands the literature on the circumstances under which private governance is effective, as well as illuminates the significant impact of state contestation.

State contestation, net sales, and the boomerang model
The applicability of the brand-to-state boomerang stems from the fact that the MNCs implicated in the forced labour in cotton in each case are large garment retailers, private governance efforts have been taken, and there is variation in the outcome. Additionally, garment MNCs are known to be particularly sensitive to negative press because their business, in a large part, relies on brand reputation (Barney 1991, Dunning 2001. In our application of the brand-to-state boomerang, we add the contextual nuance of disputed state-imposed forced labour. The contestation of the presence of forced labour by a state that organises the forced labour impacts both the issue salience component and changes the political context informing the MNC-state relationship. Note that issue salience is high when brand reputation and sales may be negatively impacted by a labour standards violation, and thus, MNCs are motivated to act in a deliberate manner . Therefore, MNCs are most likely to act collectively when the issue is salientin terms of a benefit to both reputation and salesand strong influence for improved labour standards is most likely when MNCs act collectively . However, when forced labour is contested by the imposing state, we argue the benefit to brand reputation and sales for remediating labour violations in one state can be negated by the impact to reputation and sales in the disputing state. Namely, MNCs face a trade-off in securing profits in one state versus another state, and MNCs must make a political decision to remain silent or exert their authority in GVC governance. This argument is shortened to 'net sales.' We argue this net sales juxtapositionthe political dilemma a MNC faceslikely happens when a market of a state that is contesting the existence of forced labour is important to the MNC's interests, i.e. their profit.
Thus, in instances of state-imposed forced labour, the private governance efforts and corresponding outcomes likely depend on the ability and determination of the state to contest the presence of forced labour and how important that disputing state is to an MNC's net sales. While this argument does not require a reconfiguration of the three-stage model, it does warrant additional empirical examination of the level of contestation and issue salience in each case. In the cases of China and Uzbekistan, as we will show through a rigorous process tracing methodology outlined below, the opposite governance outcomes display the significant ramifications of state-imposed forced labour on MNCs' willingness to exert their governance authority.

Methodology
Forced labour in cotton production in China and Uzbekistan implicates MNCs through their GVCs. These case studies exhibit the nuances of state-imposed forced labour and its influence on eradication outcomes. Case studies are essential empirical work in investigating and disentangling a phenomenon and its real-life context when those boundaries are opaque (Yin 2009). China and Uzbekistan, as the second and sixth largest producers of cotton in the world, 3 respectively, provide a useful comparison. These two cases' suitability for the empirical examination of private governance effectiveness is emphasised through the following characteristics: (1) both involve instances of state-imposed forced labour, (2) both implicate MNCs through GVCs, (3) both were addressed by international activist campaigns and face regulatory action in MNC home/buyer states, (4) both are cases involving cotton production (as forced labour characteristics can differ by sector). 4 Thus, given the similarities of the cases and the different governance outcomes with regard to continuation and eradication of forced labour, studying the contextual variation is significant to infer beyond the immediate, observable information (King et al. 1994). A systematic, structured procedure for data collection and analysis is described below.
First, in examining the particularities of state-imposed forced labour in each case, we analyse the political context influencing private governance constraints and opportunities, as well as explain resource dependence on cotton production to characterise the power relationship between the MNC and the state. The political contextincluding geographic isolation, domain maintenance, and moral legitimisationshape and sustain forced labour, and examining the variation in state-imposed forced labour can 'refine the relevant contextual specificities and boundary conditions' (Crane 2013, p. 63). Threats of retaliation against MNCs for private governance efforts and the size of consumer markets shape the power dynamic between the state and MNCs. This is mainly carried out through analysis of descriptive statistics regarding cotton production as a percentage of each state's Gross Domestic Product (GDP). Further, relevant documentary evidence from governments and investigative reporting are analysed for information on MNCs implicated and the level of state contestation.
Second, we examine the private -MNC and multistakeholder initiativegovernance activity to define and analyse the issue salience and MNC response in each case. We research whether there is sizable news coverage and the number and weight of activist campaigns heightening MNC reputational damage in the home and host states. Further, we analyse a few examples of prominent MNCs and their governance activity in their CSR policies or code of conduct guidelines to analyse the reach and oversight of their GVCs. The sampling criteria used for choosing certain MNCs (so that they are representative of the larger population of MNCs implicated in forced labour) is their prominence in cotton GVCs and their global brand recognition, as showcased in their prominence in the media, their internationalisation in more than 30 markets, and their membership in garment retailer associations. It should be noted that researching company responses to forced labour is challenging because voluntary reporting is unreliable and unverified, and much of the MNC-supplier relationship is proprietary (Rühmkorf 2018). Nevertheless, important evidence is found to draw out the impact of varying state contestation of the presence of forced labour on the willingness of the MNC to remediate.
Third, we analyse the mobilisation structure and collective action, including the number of MNCs active and the collective effort made, and the limitations of these governance initiatives in each case. As third-party programmes, such as ILO programmes, contribute to the mobilisation structure, their activity is described in each case. Conducting simple text analysis on two garment retailer association public statements emphasises the difference in wording on forced labour governance efforts and illustrates the variation in collective action in each case. We draw information directly from campaign/multistakeholder initiative websites that are particularly active in cotton and forced labour eradicationretail associations, the Ethical Trading Initiative, the Better Cotton Initiative, the Responsible Sourcing Network, and the Cotton Campaign.
To supplement the document-based research and empirical analysis, which includes accounts from local stakeholders, we also conducted semi-structured interviews with experts on forced labour and cotton GVCs, including actors involved in governance efforts in Uzbekistan, to solicit perspectives on the ability of private governance efforts to eradicate forced labour. Topics of discussion were governance initiatives in the XUAR and Uzbekistan, CSR effectiveness, and unique characteristics of stateimposed forced labour that can lead to political dilemmas for different governance actors.

Political context and resource dependence
The case of forced labour in the XUAR offers perspective into state-imposed forced labour, the unique context of CSR efforts in China, and also highlights the dilemma a MNC faces when a state contests the presence of forced labour. The forced labour of Uyghurs and other ethnic groups living in the XUAR by the regional government under the control of the Chinese Communist Party has been widely documented by governments, NGOs, human rights organisations, and thinktanks (CECC 2020, Foreign & Commonwealth Office 2020, Lehr and Bechrakis 2019, Xu 2020).
The political context mediates the influence a social campaign can exert to achieve an outcome (Meyer and Minkoff 2004). After initial denial of the existence of detention facilities where forced labour occurs (Maizland 2021), Chinese government officials began to allege that the facilities in the XUAR are vocational training centres, with the aim of poverty alleviation and de-radicalisation of terrorist, separatist, and extremist views (Zeng 2020). The Chinese government maintains strict control of any foreign journalists entering the region, censors these visits, and rebuts any negative reporting (Kang 2021). Therein, the state disputing the presence of forced labour marks a phenomenon worth studying with regard to private governance. Moral legitimisation sustains and shapes capabilities for forced labour to continue. Specifically, forced labour is accepted through practices such as 'storytelling and other forms of communication as well as broader forms of socialization and culture management' (Crane 2013, p. 61). In the Chinese government's direct management of the situation, forced labour becomes politically sensitive, and this likely alters the motivations for MNCs to act as they may face retaliation.
Recently, the Chinese government has retaliated; they have reacted to allegations of forced labour in the XUAR by limiting MNCs' sales, sanctioning Western government officials and private sector workers, and stirring a consumer boycott (Indvik 2021). On 23 March 2021, the US, UK, EU, and Canada, in their first coordinated effort, imposed sanctions on Chinese officials, calling the activities in the XUAR human rights abuses. On 24 March 2021, after the joint sanctions, H&M, Nike, Burberry, and others disappeared from product searches on major Chinese online retailer's websites and mapping applications (Indvik 2021). The Chinese Communist Youth League also established an online protest, 'Support Xinjiang Cotton' against H&M, Nike, and other brands. Numerous Chinese brands and celebrities came out in support (Cheng and Chan 2021). These instances show the state contestation and the resulting effect on MNCs' bottom lines in China and increasing politicisation of forced labour in the XUAR.
With regard to resource dependence, the importance of cotton as a commodity is important to evaluate the power asymmetry between the MNC and the state. Cotton is particularly important to the economic output of the XUAR, but less significant overall to China. Over 80 per cent of China's cotton is sourced from the XUAR (Lehr 2020), and 22 per cent of the world's cotton comes from China. 5 Compared to Uzbekistan, whose cotton production is a large share of GDP as will be detailed in the next subsection, China's cotton, yarn, textile, and apparel exports, not counting production and internal consumption, comprise only 1.9 per cent of China's GDP (Lehr and Bechrakis 2019). With many other important commodities contributing to China's GDP, China is relatively less resource dependent on the MNCs who source cotton from China, and the power dynamic further leans towards the state. This minimal resource dependence alters the political context, which according to the three-stage model mediates the MNC response and collective action.
The forced labour in the XUAR remains ongoing and is sustained through state-imposition and control, in a large part due to the power dynamic in favour of the state. Furthermore, the conflicting perspectives of the MNC home statemostly Western statesand China heightens the net sales risk the MNC faces in what governance efforts to take as the MNC weighs which consumer market is more important to their bottom line if one market or the other is cut off.

Issue salience and MNC response
The issue salience of the labour violation (largely brought on by potential damage to brand reputation) informs MNCs' response to instances of labour violations ). Many MNCs have been implicated, and many MNCs respond saying it is impossible to know their GVC in its entirety (BEIS Committee 2021), while also saying they can effectively self-police their GVCs (Lowry 2021).
There is an inherent contradiction. The MNCs implicated in forced labour in the XUAR include Marks & Spencer, H&M, and Coca Cola, among 82 others (Xu 2020). News reports have increased dramatically since 2017 on the issue, increasing the notoriety of the case. Since 2016, H&M was named in 190 articles and Nike in 144 articles. 6 This media attention brings reputational damage to MNCs, as exemplified in a briefing paper released by the Ethical Trading Initiative, which specifically focuses on how businesses are negatively impacted by state-imposed forced labour and how they can mitigate that reputational damage (ETI 2019).
Additionally, the reputational damage to MNCs is very strong in the MNC home states, as many states are conducting oversight of MNCs' ability to carry out oversight of their GVCs. Many MNCs are being explicitly named in government documents, thinktank reports, and activist campaigns (BEIS Committee 2021, CECC 2020, End Uyghur Forced Labour 2020, Xu 2020). For example, in a 10 March 2021 report, the UK Parliament's Business, Energy and Industrial Strategy Committee found that many companies are complicit in forced labour in XUAR because they could not guarantee that raw cotton did not originate there (BEIS Committee 2021). Further, the U.S. House Ways and Means Committee has conducted multiple Congressional Hearings 7 on forced labour in GVCs, and even specifically on the XUAR.
Whereas MNCs face pressure from many states in the West and human rights organisations to audit their GVCs and end sourcing from the XUAR, they also face risk to their sales and profit in China, given China's contestation of the existence of forced labour and their economic retaliation. H&M serves as an illustrative example of the political dilemma a MNC faces with regard to net sales and its impact on issue salience. On 24 March 2021, H&M took down their 'Statement on Xinjiang' due to a spike in negative Chinese press coverage, store removal from large Chinese online retailers JD.com and T-Mall, and store removal from mapping apps Baidu and Gaode (Indvik 2021). Moreover, Hugo Boss and Asics posted on the Chinese social media website Weibo that they would continue to source cotton from the XUAR (Business and Human Rights Resource Centre 2021), but later deleted their posts (Indvik 2021). Note that China comprises 10 per cent of Hugo Boss's sales globally, and China comprises 12 per cent of Asics' sales (Business and Human Rights Resource Centre 2021). This emphasises the key importance of China's market to MNC sales and heightens the net sales risk a MNC faces when the evidence of forced labour is contested by the government of a key market.
As private governance through CSR is normally beneficial for both an MNC's reputation and sales, this set of circumstances requires particular attention. This dilemma concerning net sales illustrates the perils of CSR when political contestation exists; the MNC faces misaligned incentives with regard to acting socially responsible with respect to consumers in their home marketthey must risk sales in one state or risk sales in another. In 2020, with a population of 1.4 billion 8 and an average income per capita of 10,610 USD, 9 China is an important market. In sum, issue salience is low for MNCs given the massive consumer market in China greatly impacts the business interest of MNCs in terms of net sales. Thus, net sales are an important consideration when determining the issue salience and the resulting MNC response through the three-stage model. In this case, the issue salience has been shown to be considerably low due to the annulling effect of state contestation of forced labour.

Mobilisation structures and collective action
In the brand-to-state boomerang, issue salience informs the MNC response, and the MNC response influences the collectiveness of the private governance effort through mobilisation structures (see Figure 2). In this case study, the low issue salience leads to minimal mobilisation and collective action on the part of MNCs and multistakeholder initiatives as described below.
First, MNCs passed risk and liability onto multistakeholder initiatives and third-party programmes instead of enhancing the auditing of their own GVCs. Therein, the collective action is lacklustre because of the threat of sales and profit harm and lower issue salience. For instance, on 10 March 2020, a coalition of four trade associations wrote a joint statement that iterated their concern over reports of forced labour, that it poses a 'profound challenge' to the integrity of the GVC, and that they are considering all available approaches but ultimately urge state-to-state engagement on the issue. 10 This statement does not say member MNCs will end sourcing from the XUAR nor increase their oversight and transparency regarding their distant suppliers. Rather, they call for the situation to be handled diplomatically. This exemplifies the finding that MNCs' CSR policies on forced labour are 'more aspirational and less stringent' than other reputation-damaging behaviour, such as bribery (LeBaron et al. 2021, p. 11). Therein, we see the skirting of responsibility, the lack of political will for collective action led by MNCs when CSR is misaligned with profit interests, and the call for public governance when forced labour is disputed.
Furthermore, two multistakeholder initiatives, whose members are MNCs, human rights organisations, and suppliers, made more of an effort, but still faced blowback for making statements on the XUAR. The Ethical Trading Initiative released a statement only after negative media coverage in mid-March 2021. Ethical Trading Initiative urged its members to withdraw from XUAR entirely due to the lack of access for independent auditors (ETI 2021). Similarly, the Better Cotton Initiative suspended licensing cotton in the XUAR in March 2020, and suspended all field-level activities, such as assurance and capacity-building work, in October 2020 (Better Cotton Initiative 2021). In the West, the Better Cotton Initiative faced negative press after both taking down their statement on the XUAR from their website for a period of time and, simultaneously, faced a consumer boycott in China because of that statement (China Daily 2021, Woo 2021. This depicts the dilemma facing MNCs and multistakeholder initiatives when the situation of forced labour is disputed, as well as shows why collective action was minimal in this case of state-imposed forced labour. Other multistakeholder initiatives and third-party programmes have also exhibited minimal collective action. The Responsible Sourcing Network, which includes investors, companies, and human rights advocates in their coalition, does not have a XUAR Cotton Pledge, while it does for Uzbekistan and Turkmenistan. 11 Instead, Responsible Sourcing Network references the Uyghur Forced Labour Coalition, that is largely supported by human rights organisations, with only seven MNCs signing on to their strongly worded Call to Action that requires MNCs to cut ties with all suppliers and sourcing of cotton from the XUAR. 12 Further, the UK Parliament has reached out to the ILO in a letter to inquire why they have not commented on the situation in the XUAR. The ILO, with a membership of states, largely constructs its programmes with the cooperation of state governments (Phillips and Mieres 2015). The ILO responded to the UK Parliament saying it cannot make any public statements until its Committee of Experts have examined all evidence available from different actors (ILO 2021b). 13 In sum, few MNCs are acting, the mobilisation structure is comparatively stunted, and the collective action of MNCs is minimal. Given the size of the market in China, market access is essential for MNC sales and profit. Thus, it is clear that MNCs have not acted in a collective manner to end the state-imposed forced labour in the XUAR, likely because of net sales concerns. This case study highlights the nuance of state contestation, which lowers the issue salience and affects private governance efforts. Next, the state-imposed forced labour regime in Uzbekistan is examined.

Political context and resource dependence
The case of forced labour in Uzbekistan offers perspective into state-imposed forced labour, as well as the lesser political dilemma a MNC faces when implicated in cases where the forced labour is not severely disputed and CSR activity does not imperil access to a key market. Though the Constitution of Uzbekistan prohibits forced labour, there is widespread acknowledgement and research into the state-imposed forced labour regime that existed there (Atayeva and Belomestnov 2010, Evans and Gill 2017, McGuire and Laaser 2018, Schweisfurth 2020. Reports of state-imposed forced labour began in 2004 by Uzbek journalists, and later the Environmental Justice Foundation published a report White Gold to document the use of state-imposed forced labour (EJF 2010). The Uzbek Forum for Human Rights (formerly Uzbek-German Forum for Human Rights, UGF) has monitored and reported annually on child and forced labour in Uzbekistan's cotton fields since 2009 (Schweisfurth 2020).
The power dynamic between the state and MNCs sourcing cotton from Uzbekistan is nuanced. It is important to note the vestiges of the Soviet legacy, especially in state-controlled production, in Uzbekistan. 14 Following the death of President Islam Karimov who ruled from 1991 to 2016, the former Prime Minister Shavkat Mirziyoyev became President and continued the state-imposed forced labour for cotton harvests (MacFarquhar 2016, Schweisfurth 2020). In Uzbekistan, thirdparty monitoring and auditing was challenging particularly because of the climate of fear and repression; monitors and auditors were met with arbitrary arrest, threats, travel restrictions, surveillance, and confiscation of research materials (Evans and Gill 2017). Similar to the case of stateimposed forced labour in the XUAR, the government ascertained control and management of the narrative around forced labour to legitimise it (Crane 2013). This shows that there is a level of contestation and unwillingness to acknowledge and work to eradicate forced labour, but this is not as significant as in the XUAR wherein numerous government documents adamantly deny the existence of forced labour.
With regard to resource dependence, the state is relatively more dependent on cotton, as it is integral to Uzbekistan's economy. Uzbekistan is the sixth largest producer of cotton, comprising 3 per cent of the world's cotton production. 15 While cotton production is integral to both China and Uzbekistan, cotton makes up a significantly larger share of GDP in Uzbekistan, approximately 25 per cent of GDP in 2016 (Evans and Gill 2017). Further, the mere fact that the ILO was allowed to monitor harvests starting in 2013 shows that the state was relatively more pliant. The state's significant dependence on cotton mediates the influence of MNC efforts in private governance, as shown through the three-stage model.
As of January 2021, the ILO had declared that forced labour has ended in Uzbekistan (ILO 2021a). Next, issue salience and MNC response are discussed to highlight the contextual differences of stateimposed forced labour in Uzbekistan and analyse how these differences enabled more robust and influential private governance.

Issue salience and MNC response
In the case of state-imposed forced labour in Uzbekistan, the issue salience to MNCs (brought on by brand reputational damage) is high, and no significant political dilemma faces MNCs. Though some contestation is present (in the fact that the state continued to orchestrate forced labour through the presidential transition), it is not significant enough to influence the governance outcome. For example, MNCs feel pressure from consumers to implement CSR policies and end sourcing in Uzbekistan due to media coverage of the issue. News reports on the forced labour in Uzbekistan increased steadily since 2003, and news peaked in 2020. 16 Most sources were government documents, human rights organisations, and international institutions. H&M was named in five articles and Nike was not mentioned. The news reports that were cited the most were in Uzbekistan Daily, Reuters, Central Asia News, and The New York Times. This media attention served to heighten awareness and reputational damage. Further, the fact that Uzbekistan's news sources were even able to discuss forced labour shows lesser state contestation. In the case of the XUAR, the US-China rivalry may have amplified media attention and policy conflict regarding GVC governance more so than with other states (Gereffi 2020). Nevertheless, the size of the market and the potential impact on net sales are central to understanding the issue salience fully.
MNC home state sales would be harmed if they continued to source cotton from Uzbekistan because garment MNCs rely on brand reputation (Barney 1991, Dunning 2001. For example, the Ethical Trading Initiative specifically mentions state-imposed forced labour in Uzbekistan as harmful to MNCs' brand reputation (ETI 2019). Meanwhile, decreased sales in Uzbekistan would likely not be significantly harmful because there are few consumers. With a population of just 35 million 17 and an average income per capita of just 1,670 USD, 18 Uzbekistan is not a large market for MNC garment retailers. Further, the government of Uzbekistan did not make any similar threats to MNCs' profit like in China (i.e. their stores were not being banned from online platforms, nor did their state media spur a consumer boycott). In other words, issue salience is high for MNCs given that CSR activity brings benefit to both reputation and sales, and the net sales interest is weighted heavily toward the home states of the MNCs -Western statesgiven the minimal counterbalancing force from Uzbekistan's consumers. In sum, the issue saliencethe reputational damage forced labour in Uzbekistan brings to the MNCis high, and MNCs responded concertedly.

Mobilisation structures and collective action
In contrast to the lack of collective action in the XUAR, MNCs, multistakeholder initiatives, and thirdparty programmes have made a collective and credible threat to end cotton sourcing from Uzbekistan because of state-imposed forced labour. From the beginning, coalitions mobilised to act collectively. On 15 August 2008, a coalition of four trade associations wrote a letter to President Karimov asking him to end the use of forced labour in Uzbekistan. In contrast to the hesitant industry association joint statement on the XUAR, this letter directly urged the government of Uzbekistan to end forced labour in cotton, and threatened to withdraw from sourcing cotton from Uzbekistan entirely if improvements were not made. 19 Clearly, given the mediating political context of relative power of MNCs over the state, the associations and their MNC members faced less of a political dilemma in this case and acted collectively.
Further, the mobilisation structures enabling collective action were more robust in Uzbekistan. For example, the Cotton Campaign, a multistakeholder initiative consisting of human rights, labour, investor, and business organisations including the Responsible Sourcing Network, Uzbek Forum for Human Rights, and others, has focused on building awareness and recruiting MNCs to use their collective power to end forced labour in Uzbekistan since 2012. Further, the Responsible Sourcing Network maintained the Uzbek Cotton Pledge, a pledge to not source cotton from Uzbekistan, and its membership of 328 brands from 2010 to 2022. Therein, the mobilisation structures were highly integrated. Additionally, the Better Cotton Initiative recently relaunched their programme in Uzbekistan and has been sharing their open-source method for sustainable cotton production with the ILO and the Deutsche Gesellschaft für Internationale Zusammenarbeit. They are developing vertically integrated farm-to-garment production systems in order to export textiles and garments rather than raw cotton and move Uzbekistan up the GVC (GIZ 2020). 20,21 This is remarkable evidence of the structural power of MNCs. Taken together, there were numerous campaigns, multistakeholder initiatives, and third-party programmes operating in Uzbekistan with the aim of forced labour eradication. This mobilisation structure enabled a strong and credible private governance effort. In sum, the minimal contestation that existed meant the state-imposed forced labour regime was more pliant (presenting an uncontested political opportunity for MNCs), and MNCs mobilised.

Conclusion
Forced labour persists in our contemporary global economy, and this study has shown the mixed effects of MNC self-regulation, multistakeholder initiatives, and third-party programmes on the eradication of state-imposed forced labour, largely due to behaviour and context of the states in question. Essentially, private governance breaks down under the influence of state contestation due to the threat of lost business sales and profit for MNCs in the host state. This builds on the international political economy literature in that MNCs are a structural force that shape the agenda around governance (in this case, the eradication of forced labour), and that their private authority has limits: MNCs will likely be socially responsible only when it is in their business interest.
In the XUAR case, state-imposed forced labour remains ongoing. Although some MNCs and multistakeholder initiatives have responded, they have not acted concertedly. This is largely due to the threat of economic blowback on their sales in China. Notable exceptions are the Ethical Trading Initiative and Better Cotton Initiative statementsbut even these have been saddled with controversy. At the zenith of politicisation, some MNCs and multistakeholder initiatives took down their public statements on the XUAR following China's retaliation on their market access. As Shepherd (2021) aptly put, 'Multinational companies have been forced to walk a tightrope to ensure they are not complicit in human rights abuses in Xinjiang while avoiding Beijing's ire as they seek to operate in the world's second-biggest economy.' As argued by Lehr and Bechrakis (2019), implementing a voluntary industry ban on cotton from the XUAR similar to the one on cotton from Uzbekistan 'would undoubtedly prove challenging for the industry, given the importance of China as a source country' (p. 18). As such, the political context of a major economy whose government vehemently disputes the existence of the state-imposed forced labour regime contravenes the issue salience, given net sales and profit concerns.
In the Uzbekistan case, state-imposed forced labour has been eradicated. A broad and concerted effort by many MNCs and multistakeholder initiatives to end sourcing of cotton from Uzbekistan was shown to be particularly influential on the eradication of forced labour because the MNCs did not face a significant political dilemma with regard to their net sales. Further contributing to this influence was the political context of a resource dependent state that was likely to succumb to outside pressures, granting relative structural power to MNCs. As  argues, there needs to be a credible threat of withdrawal from the state for MNCs to influence the government, and this was clearly present in Uzbekistan. See Table 1 for a summary of the two cases.
Then how does state-imposed forced labour impact private governance? The impact is likely mediated by the degree to which the forced labour is contested by the imposing state and the threat that contestation poses to an MNC's net sales and profit. In the case of the XUAR, the state's ability and determination to contest the presence of forced labour lowers the issue salience the negative impact on brand reputation and sales in the Westbecause of the credible threat of loss of sales and profit potential for MNCs selling to consumers in China. In the case of Uzbekistan, MNCs and multistakeholder initiatives acted collectively likely because they did not feel a large enough threat to their sales, given the consumer market in Uzbekistan is small and Uzbekistan did not retaliate. In sum, private entities can play a role in eradicating state-imposed forced labour, but the impact on profit and sales determines actors' willingness to act socially responsible. In other words, if a disputing state's market is important to the MNCs' sales and profitability, then there is a strong disincentive for MNCs to interfere. Broadly, private governance is not entirely ineffective in cases of state-imposed forced labour, but when the presence of forced labour is severely disputed and politicised, private governance and the conditions necessary for effectiveness are muddled.
A limitation of this study is the different durations of the governance activities in each case, with the state-imposed forced labour in the XUAR being more recent and ongoing. However, as argued, the conditions are significantly varied insofar as the length of time governance has been ongoing matters less, as it is already clear that the private governance activities in the XUAR are fewer and less concerted. Further qualitative and quantitative research on state-imposed forced labour is needed. Specifically, examining private governance in other cases such as Turkmenistan through the boomerang model would be insightful. Additionally, researchers should quantitatively examine and determine the exact size of the consumer market that would tilt the scales and pose a political dilemma in terms of the business interest for net sales over social responsibility. Finally, as described in the literature review, as forced labour falls on the spectrum of labour exploitation, examining state contestation and issue salience in cases of poor labour conditions tacitly imposed (or not effectively deterred by the state) could be tested. This study has policy implications because it indicates that the structure and governance of GVCs need to be comprehensively examined and structurally changed to eradicate forced labour globally. Policy practitioners are increasingly discussing and implementing unilateral import bans to prevent entry of goods produced with forced labour. 22 In cases of state-imposed forced labour, these public governance interventions might prove more effective as they may alleviate the political dilemma facing MNCs in terms of acting against the disputing state because MNCs are required to act. However, these interventions can be implemented inequitably across products and states and for geopolitical reasons, and this politicisation can be detrimental to the actual end goal of eradication of forced labour if a MNC decides to bifurcate production into 'clean' and 'dirty' GVCs. Recent studies on private governance of labour standards have discussed enforceable duties on lead firms through the Bangladesh Accord (Anner et al. 2013) and the potential for more effective private governance when multistakeholder initiatives are unioninclusive (Ashwin et al. 2020). However, policy practitioners may want to consider to what extent these more robust private governance initiatives will be effectively implemented by MNCs under conditions of state contestation.
In sum, this application and modification of the brand-to-state boomerang in China and Uzbekistan contributes to the international political economy literature on private governance, and particularly, to the rather limited studies of private governance effectiveness in cases of state-imposed forced labour. When states deny the existence of forced labour and are willing to enforce that belief, it complicates MNCs' willingness to act socially responsible given the credible threat of harm to their bottom line. MNCs can take refuge in unclear and complicated GVC structures and non-mandatory disclosure of the furthest suppliers, and call for the case to be settled diplomatically, rather than exercising their considerable structural power to eradicate forced labour. MNC shirking due to politicisation, mixed evidence regarding private governance efforts to eradicate forced labour globally, and potential bifurcation of production because of conflicts of interest between one state and another all detail the muddled governance of forced labour.

Disclosure statement
No potential conflict of interest was reported by the author(s).