Gender and Forced Labour: Understanding the Links in Global Cocoa Supply Chains

Abstract This paper investigates the gendered patterns and dynamics of labour exploitation and forced labour in the cocoa supply chain. The empirical basis of our analysis is an original primary dataset produced through the Global Business of Forced Labour project, which includes data gathered in Ghana in 2016–2017, comprising 60 in-depth interviews and a survey of 497 cocoa workers across 74 cocoa communities from Ghana’s two largest cocoa-producing regions, the Western and Ashanti Regions. Drawing on this dataset, we show that prevailing business models within the Ghanaian cocoa industry rely on and reinforce labour exploitation and unequal gender power relations. Given that the links between forced labour and gender remain poorly understood, we analyse the factors that render women workers disproportionately vulnerable to severe labour exploitation, underscoring the role of unequal family relations, responsibility for reproductive labour, and social property relations in creating vulnerability to exploitation.


Introduction
Forced labour is widely acknowledged to be a resilient and endemic feature of the contemporary global economy, and policy and civil society actors around the world are dedicating sizable efforts and resources to combat it. Increasingly, government and private organisations are emphasising gendered vulnerability to forced labour and launching programmes to address this. Yet, to date, efforts to understand the gendered patterns and business of forced labour in global supply chains continue to be limited by the lack of rigorous, independent, and in-depth empirical data (LeBaron, 2018b). That so little is known about the root causes of gendered vulnerability to forced labour in global supply chains severely limits the design and effectiveness of anti-slavery initiatives.
Efforts to understand forced labour have also been held back by theoretical and analytical shortcomings, namely, the tendency to study forced labour as an individualised and anomalous phenomenon, isolatable from broader relations of labour in global supply chains. In 2013, the introductory essay to a symposium in this journal highlighted the tendency for scholars to theoretically and empirically 'ring-fence' forced labour. 1 Barrientos, Kothari, and Phillips argued that although there was a burgeoning literature on vulnerable workers and labour standards in the global economy, the most severe forms of exploitation encompassed in the terms forced and unfree labour were 'only rarely integrated into the continuum that informs most research. Instead, they are thought to represent a separate and special category of labour relations and need to be viewed through different theoretical and empirical lenses ' (2013, p. 1038; see also LeBaron & Phillips, 2018). The authors challenged this approach, urging scholars to turn away from abstract theoretical debates grounded in binary conceptualisations of free and forced labour and to embrace a fresh approach to understanding the forms that forced labour takes in the contemporary global economy.
Since that symposium was published, an interdisciplinary wave of scholarship cutting across development studies and international political economy has continued to make steady progress in documenting the nature and role of severe labour exploitation in the global economy. Recent studies have mapped and compared the patterns of forced and unfree labour in the supply chains of various industries (Crane, LeBaron, Allain, & Behbahani, 2017;McGrath, 2013;Mezzadri, 2016;Phillips, 2013) and across a range of geographical settings (Allain, Crane, LeBaron, & Behbahani, 2013;Guérin, 2013;LeBaron, 2018a), contributing to understandings of the dynamics of severe labour exploitation across region and level of economic development. Significantly, this wave of scholarship has moved away from conceptualising 'unfree' and 'forced' labour as discrete and easily isolatable categories of labour relations and increasingly acknowledges that severe forms of exploitation are challenging to disentangle from more minor forms of exploitation (LeBaron, 2018b;Strauss & McGrath, 2017). Theorists have explored and unsettled the links between forced labour and 'regular' exploitation (Bunting, & Quirk, 2017;Chuang, 2014;Morgan & Olsen, 2014), and contested definitions of key concepts and terms, including modern slavery, human trafficking, and unfree labour (Fudge & Strauss, 2013;O'Connell Davidson, 2015;Strauss & McGrath, 2017). While robust empirical data on severe labour exploitation in global supply chains remains rare, the literature is advancing, with several exciting avenues of inquiry.
However, the role that gendered power relations play in shaping patterns of severe labour exploitation within global supply chains is still not well understood (Barrientos, 2013a(Barrientos, , 2013b(Barrientos, , 2014Dunaway, 2014;Joekes, 1999a). While the International Labour Organisation (ILO) and others estimate that women and girls account for the majority of forced labourers globally, even within noncommercial sex sectors (ILO & Walk Free Foundation, 2017, p. 10), scholarship on the links between gender and severe labour exploitation in global supply chains (also referred to as global value chains (GVCs), global production networks (GPNs), and global commodity chains (GCCs)) is at an early stage. We lack detailed understandings of how, why, and under what circumstances gendered power relations can give rise to forced labour across and within various business models, sectors, and geographic locations, and how these dynamics converge or differentiate across contexts. Much of the analysis that has taken place so far on gendered vulnerability to forced labour has focused on feminised industries, such as domestic, care, and sex work. As well, scholarship on labour in global supply chains often bifurcates between analyses of labour exploitation within the 'productive' and 'reproductive' realms, focusing either on household-focused studies of reproduction, or industry, sector, or activity-oriented studies focusing on productive work. To date there has been an overwhelming focus on the latter. As a result, across many sectors and supply chains, there is lacking analysis of the differential labour conditions experienced by men and women, as well as the specific ways that gendered power relations and labour unfreedom intersect and reinforce each other.
In this paper, we investigate the links between gender and forced labour within the global cocoa supply chain. We draw on original data gathered through the Global Business of Forced Labour project (LeBaron, 2018a), including data collected in Ghana in 2016-2017, comprising 60 in-depth interviews with cocoa workers and a survey of 497 cocoa workers across 74 cocoa communities within the two largest cocoa-producing regions in the country, the Western and Ashanti Regions. Our data sheds light into the different working conditions and experiences of men and women workers; why and how workers become vulnerable to forced labour and overlapping forms of labour exploitation; workers' broader experiences of labour and work within the cocoa sector; and how cocoa industry business models are configured to profit from exploitation, including forced labour. Although the literature on cocoa tends to foreground the experiences of men, grounded in the assumption that cocoa is a 'male crop,' our research finds that both men and women working within the global cocoa supply chain experience widespread labour exploitation, and some workers experience forced labour, the key indicators for which include: non-payment, underpayment, and withholding payment; physical violence and verbal abuse; threats of dismissal; deception; and nonphysical coercion (especially food deprivation); sexual violence; and nnaho, a form of involuntary labour. We find that women workers tend to experience more severe forms of labour exploitation within the cocoa industry than men, and that business models are configured to profit from women's unequal position within the industry and society more broadly.
We seek to advance two particular points in relation to the literature on labour standards in global supply chains. First, in line with extant studies (Barrientos, 2013a(Barrientos, , 2014(Barrientos, , 2019Dunaway, 2014;Guérin, 2013;Mezzadri, 2016), we hold that gender plays an important role in shaping patterns of work and exploitation in global supply chains. We identify five key mechanisms through which gender relations structure patterns of labour exploitation and forced labour within our case study of the cocoa industry: (1) gendered norms and divisions of labour; (2) payment practices and inequalities in income; (3) unequal land access; (4) lacking redress and access to justice; and (5) social reproduction, the household, and family/kinship practices. We argue that these dynamics come together to render women workers disproportionately vulnerable to severe labour exploitation. Not only did we find that women earn less, have less stable employment, and experience more significant barriers to accessing stable employment, as has been previously emphasised in the literature, but our data also suggests that women are more vulnerable to the severe forms of labour exploitation associated with forced labour.
Second, we build on existing studies of chain and network dynamics in the global cocoa sector (Fold, 2004(Fold, , 2008(Fold, , 2014Neilson, Pritchard, Fold, & Dwiartama, 2018) to argue that prevailing cocoa business models rely on and reinforce unequal gender relations, including a regressive mode of social reproduction. We show that women's exploitation in the realm of paid, economically productive labour operates through and alongside patriarchal gender norms and a gendered division of reproductive labour that leaves women with unequal responsibility for childbearing, care work, and domestic labour, such as cleaning, cooking, and washing. We draw on social reproduction theory (SRT) developed by feminist political economists to show that women's exploitation in the productive sphere cannot be understood in isolation from their experiences within the realm of reproduction. We underscore the role of unequal family relations, responsibility for reproductive labour, and gendered social property relations in fuelling women's vulnerability to severe labour exploitation in global supply chains. Our key theoretical contribution is to extend the insights of SRT to a global South context and to severe labour exploitation experienced by workers within the bottom rungs of the global labour market. In spite of some promising recent contributions to the literature (for example, Mezzadri, 2016), both of these remain largely new terrain for SRT analysis.
Our paper consists of five parts. Following this Introduction, we elaborate our perspective on gender and forced labour in relation to existing theories. The subsequent section discusses our methodological approach. We then present our empirical findings, including our analysis of the gendered patterns of exploitation within the cocoa industry and the root causes of these patterns. A final section concludes, arguing for the need for future research to centralise and prioritise the relationship between gender and forced labour and to adopt an integrative, non-binary view of 'productive' and 'social-reproductive' labour in the study of global supply chains.

Understanding gender and forced labour in global supply chains
Scholarship on global supply chains, also called GVCs, GPNs, global commodity chains (GCC) (Gereffi, Humphrey, & Sturgeon, 2005;Gereffi & Kaplinsky, 2001;Henderson, Dicken, Hess, Coe, & Yeung, 2002;Kaplinsky, 2000), and most recentlyglobal wealth chains (Seabrooke & Wigan, 2017 and poverty chains (Selwyn, Mosiolek, & Ijara, 2019)has become a prominent lens into global political economic dynamics, and increasingly, into global labour relations (Anner, 2012;Anner, Bair, & Blasi, 2013;Fransen, 2012;Fransen & Burgoon, 2017;Locke, Rissing, & Pal, 2013). 2 Recent interventions in the literature have argued for the importance of gender relations in shaping the dynamics of global supply chains. Notable contributions include Stephanie Barrientos' (2013a, Gender and forced labour in global supply chains 1097 2013b, 2014, 2019) work on gender in global value chains and global production networks, Wilma Dunaway's (2014) edited volume on 'gendered commodity chains', and on global value chains as gendered structures. These studies demonstrate that gender relations play a key role in shaping patterns of work and employment, wealth distribution, and dynamics of inclusion and exclusion in global supply chains. These scholars have argued that gender should be integrated as a fundamental part of the methodological and analytical framework for inquiry into global supply chains, rather than a conceptual 'add-on' (Dunaway, 2014; see also Mezzadri, 2016;Strauss, 2013).
Taking forward these insights, our study seeks to integrate gender into the framework for analysing severe labour exploitation in global cocoa supply chains. Although the feminist literature on global supply chains has only rarely investigated forced labour, it nevertheless offers several important insights that helped to guide our research. Drawing on this body of work, we propose five key mechanisms through which gender shapes patterns of labour exploitation within our study: (1) gender norms, types of work, and the gender division of labour; (2) payment practices and gendered inequalities in income; (3) unequal land access; (4) lacking redress and access to justice; and (5) social reproduction, the household, and family/kinship practices. In this section, we briefly describe each mechanism and their centrality to understanding gendered patterns of forced labour. In Section 4, we present our empirical findings related to each mechanism, analysing how and why each facilitates labour exploitation in the Ghanaian cocoa industry. First, we first briefly set out our understanding of social reproduction, which we draw from the feminist political economy literature known as SRT (Ferguson, LeBaron, Dimitrakaki, & Farris, 2016; see also Bakker, 2007). 3

Social reproduction theory and global supply chains
Understanding how gender shapes patterns of labour exploitation in global supply chains requires an integrated understanding of labour that stretches across the productive and reproductive realms. Feminist political economy literatures have sought to achieve this through the concept of social reproduction, which refers to 'the activities associated with the maintenance and reproduction of peoples' lives on a daily and intergenerational basis' (Ferguson et al., 2016, p. 4; see also : Bakker, 2007;Bhattacharya, 2017;Ferguson et al., 2016;Katz, 2001;LeBaron, 2010;Vogel, 1983). While theorists within the social reproduction literature have different theoretical orientations, broadly, social reproduction theorists note that gendered divisions of labour in the reproductive realm (eg. childcare, housework) shapes patterns within the labour market, as well as women's employment opportunities, contractual relations, and conditions of work. These dynamics are especially important in relation to workers at the bottom rungs of the labour market, who may not have the means to outsource responsibility for reproductive labour to others and have especially limited flexibility to arrange reproductive labour around paid work in the productive sphere.
As mentioned, within the scholarship on global supply chains to date, there has been a tendency to bifurcate analysis of labour across the realms of production and reproduction, with an overwhelming focus on production. This has created a skewed perception of the root causes and manifestations of labour unfreedom, particularly for women workers who tend to bear disproportionate responsibility for reproductive work. Especially lacking is analysis of how labour conditions in the productive realm link to those in the reproductive realm within global supply chains, even more so where work in the productive realm is forced or severely exploitative. These shortcomings are perhaps understandable from a methodological point of view, given the difficulty of researching women's 'invisible' or 'hidden' contributions in contexts that are already marked by informality, illicitness, and illegality. In analytical terms, however, this lack of holistic investigation works to reproduce dualist understandings of the spheres of production and reproduction (a dualism that has been long problematised by feminist scholars). The importance of breaking down the production-reproduction binary is highlighted by Dunaway (2014, p. 4), who enjoins scholars to reject the 'false analytic divide' between productive and reproductive spheres and to pay greater attention to the site of the household in commodity chain analysis. Drawing on SRT insights about labourand especially, SRT scholars' insight that labour is not a narrow economic category, but is practical human activity that spans both households and paid worksites (Ferguson et al., 2016)our analysis seeks to explain how relations of power and oppression extant beyond the productive sphere operate and bear upon patterns of exploitation amongst the labour force.
In addition, we draw on SRT's insights about how gender relations in the economy and society can shape the patterns of labour exploitation in global supply chains. As Strauss (2013) notes, SRT highlights a wider set of social power relations through which patterns of exploitation are embedded and produced, including those of gender, race, and sexuality; these, too, operate and shape experiences across both the productive and reproductive realms. Such relations are especially important in relation to analysis of vulnerable workers, who are often located at the nexus of several systems of oppression and labour subjugation (Mezzadri, 2016(Mezzadri, , 2017Strauss, 2012Strauss, , 2013. Although SRT has rarely been extended to understand forced labour in global supply chains, we harness its insights, alongside overlapping bodies of literature on gender and commodity chains and agriculture in Ghana, to identify five key mechanisms through which gender shapes patterns of labour exploitation in the global cocoa supply chains. While this literature highlights multiple dimensions of gender relations and their implication in processes of production, reproduction, and value-generation, we concentrate on five here, since these were the most relevant to our research questions and, importantly, to our empirical material. The remainder of this section outlines these mechanisms and Section 4 demonstrates how these operate to create gendered vulnerability to forced labour within our study.

Gender norms, types of work, and the gender division of labour
The first key mechanism through which gender and forced labour are linked in global supply chains relates to gender norms in work and the division of labour. As documented within the feminist GCC/GVC literature, gender has implications for the type of productive activity women carry out in supply chains and, importantly, for the types of productive activity they are excluded from (Barrientos, 2019;Joekes, 1999a). Gender relations and norms therefore hold explanatory power as to why certain groups of workers may be concentrated in particular nodes within a chainfor example, in sections that are predominantly feminised or masculinisedand the differential and frequently unequal conditions and rewards these workers experience (Barrientos, 2019). A study of a farmed fish supply chain in Bangladesh (Kruijssen, Rajaratnam, Choudhury, McDougall, & Dalsgaard, 2016), for example, found that women are prevented from accessing and participating in paid productive work by a combination of patriarchal gender norms and the gendered division of labour, which work to restrict women's activities to 'hidden' and unremunerated tasks such as feeding, pond management, fertilisation, and fish preparation.
As this suggests, in addition to waged labour, global supply chains integrate and rely on casualised, informalised, and non-waged forms of labour, to which households make a significant contribution. This is crucial for understanding the nature of women's work and exploitation in supply chains, since the division of labour within households is profoundly gendered. In terms of our case study of the Ghanaian cocoa industry, agricultural research on West Africa highlight gender norms in the distribution of labour, whereby women's work typically focuses on household-oriented subsistence activities, such as the cultivation of food for consumption within the family, and men's work is concentrated in market-oriented productive work in cash crops, such as cocoa (Koopman, 1991;Okali, 2009). For this reason, cocoa has been traditionally seen as a 'male crop', with women's productive involvement in cocoa generally understood to fall within two main categories: as family labourers or, less frequently, as owners of cocoa farms (Barrientos, 2013a, p. 16; see also Joekes, 1999b). Women's work as family labour in these settings is not typically remunerated (although women may receive indirect compensation for helping out on a spouse or male family member's farm in the form of housing or food) (Koopman, 1991, p. 152). More recent scholarship on the cocoa sector also highlights the growing role played by migrants (Knudsen & Fold, 2011) and women (Barrientos, 2013a) as waged labourers on Ghanaian cocoa farms, drawing attention to the 'hidden' Gender and forced labour in global supply chains 1099 character of women workers' contribution to cocoa farming, whether as unpaid family workers or as waged casual or seasonal labourers (Barrientos, 2013a, p. 7).

Payment practices and gendered inequalities in income
Gender and labour exploitation further intersect in global supply chains through payment practices and gendered inequalities in income. Feminist scholars have documented significant gender disparities in wage and income levels in agricultural and horticultural supply chains in Africa (Barrientos et al., 2013; see also Dolan, 2001;Morgan & Olsen, 2011). As above, these disparities are understood to reflect gender norms and the division of labour, which shape the types of work that women undertake in supply chains, constrain their income-generating activities, and/or restrict them to household-based activities. Gender disparities in pay and income can also be traced to biases and stereotyping in hiring and payment practices, whereby women tend to be: seen as more 'compliant' and therefore cheaper workers; hired to do manual, repetitive or otherwise labour-intensive activities that are less well-paid; or paid lower wages for equivalent work due to patriarchal power relations and forms of subordination.
These dynamics are borne out by findings from Barrientos et al.'s (2013) study of cocoa workers in India, which documents a substantial pay gap between women and men workers in Andhra Pradesh and Tamil Nadu. In the context of Ghana, Vigneri and Holmes (2009) also note that women cocoa farm owners tend to have lower incomes than their male counterparts. They attribute this to a range of structures, norms, and practices stretching across the productive and reproductive spheres, including women's unequal access to productive resources and assets, such as credit and land, and intrahousehold gender power asymmetries.

Unequal land access
Women's unequal access to land has been identified as a factor that limits upward economic mobility and income generation, and increases vulnerability to poverty (Agawaral, 1994;Deere & Leon, 2003;Deere, Oduro, Swaminathan, & Doss, 2013). Customary laws pertaining to land inheritance and tenure in Ghana are complex, and vary according to, inter alia, local norms and practices, ethnicity, and geographic location (Higgins & Fenrich, 2011). Patrilineal and matrilineal descent systems are common across the different ethnic groups in the country, with both systems tending to privilege male inheritance (although women may inherit land matrilineally, but not usually as wives or children) (Sarpong, 2006, p. 7). Other drivers of unequal land access include women's exclusion from decision-making processes over land, which are typically controlled by Chiefs and other local (male) opinion leaders and household heads; their relative lack of information about land access and tenure issues; and restrictions in usufructuary land usage (Sarpong, 2006). This means that women's access to, and rights over, land are often dependent on their relationship with husbands or other male family members, which acts as a powerful constraint on their autonomy and reinforces broader patriarchal norms.
Studies (Jacobs, 1996; see also Agarwal, 1994Agarwal, , 2003 highlight how patriarchy is supported and sustained not just through the site of the household, per se, but through the interrelationship of women's subordination across small-scale agricultural practices, systems of land tenure, and the family. These multi-sited inequalities and power relations shape and undergird women's subordinate position in society, including in work and the labour market.

Lacking redress and access to justice
Scholars of labour standards (Awumbila, Deshingkar, Kandilige, Teye, & Setrana, 2018;Awumbila, Teye, & Yaro, 2017;Elias, 2010) have documented how women's vulnerability to labour exploitation is intensified by a combination of informality, precarity, gender inequality, and a lack of legal rights and access to justice. Focusing on the experiences of migrant female domestic workers in Malaysia, for instance, Elias (2010) reveals how the invisibilised and informal character of domestic work and women's position as migrant workerswhose visas are frequently tied to an individual employermake it especially difficult (and in some cases impossible) to access labour protections and to challenge exploitation and injustice. These dynamics work in concert with severe constraints on workers such as the confiscation of identity documents by employers to profoundly constrain women's agency (Elias, 2010). Similar patterns are identified by Awumbila et al. (2017) in their study of men and women domestic workers in Accra, Ghana. The authors examine how gender mediates experiences of labour exploitation among domestic workers, noting that male domestic workers tend to have clearer duties and job descriptions than female domestic workers. This gives them greater grounds for, and legitimacy in, challenging unfair treatment on the part of employers, such as the imposition of additional working hours or tasks beyond their agreed role. Female domestic workers, on the other hand, tend to have no contract and no clear job description, which restricts their ability to contest exploitation and abuse. These dynamics are reflective of and compounded by broader gender asymmetries in education and qualification levels, the gender division of labour, and patriarchal gender stereotyping and norms (see also Awumbila et al., 2018). 4 2.6. Social reproduction, the household, and family/kinship practices SRT has long documented women's 'hidden' contribution to economic production, particularly through their disproportionate labour in producing and sustaining people as well as commodities. SRT scholars have advocated for a non-binary approach that captures the 'interrelationship of paid and unpaid work and of economically productive and socially productive labour' (Strauss, 2012, p. 145;LeBaron, 2010) and captures productive and reproductive activities as a unified socio-economic process (Bezanson & Luxton, 2006, p. 37). This is essential to understanding labour standards in global supply chains, since reproductive labour shapes the condition and character of the worker when they enter into the 'sphere' of productive work, as well as patterns of surplus value extraction in GVCs (Dunaway, 2014).
In relation to our case study, the need for a holistic approach to understanding women's labour is confirmed by the literature on agricultural practices in Ghana, which stresses the importance of intercroppinga practice that blurs the lines between production and reproduction. Intercropping involves the planting and cultivating of other arable crops, such as cocoyam, pepper, cassava, and maize, among the cocoa trees, an activity that is commonly carried out by women on cocoa farms (Agyare-Kwabi, 2009). These crops have a dual purpose: they provide a source of food for the household while the cocoa trees are growing, and also boost the productivity of the cocoa trees themselves (Ofori-Bah & Asafu-Adjaye, 2011). This activity therefore blurs the lines typically demarcated between reproductive, subsistencebased activity and productive activity linked to global commodity production. In addition, because women's labour in both productive and reproductive spheres is often shaped by intra-household power dynamics (and vice versa) (Kabeer, 1997) these are critical to understanding women's experiences of exploitation in cocoa supply chains.
These five mechanisms are not typically foregrounded within the study of forced labour. Yet our research found that they play a crucial role in creating gendered vulnerability to forced labour and exploitation in the cocoa supply chain. In Section 4, we document how each mechanism operates. First, we provide an overview of our research methods.

Background and case study selection
The empirical material for this paper is derived from a larger study called The Global Business of Forced Labour project, which examines the business models of forced labour in global agricultural supply chains, based on case studies of cocoa and tea (LeBaron, 2018a). This study consists of various strands of empirical research, including the mapping and analysis of cocoa and tea supply chains, extensive field research in Ghana and India, and elite interviews across Europe and North America, conducted between 2016 and 2018. In Ghana, the purpose of the field research was to generate original, ground-level data on how the business of forced labour operates within the cocoa sector and specifically, to understand: why and how workers become vulnerable to forced labour and overlapping forms of labour exploitation; workers' broader experiences of working within the sector; how business models are configured to profit from exploitation; and why prevailing initiatives to combat forced labour are falling short. Field research was conducted in two regions, the Ashanti and Western regions, which were selected because of their significance to both national and global cocoa production. Data focused on workers' experiences was collected using a mixed methods design, and included: a digital survey (N = 497); ethnographic research (including participant observation, informal, and group interviews); and semi-structured interviews (N = 60).
Studying labour standards in Ghanaian cocoa sheds light into a strategically important national export crop, as well as the dynamics within a valuable global supply chain. Over 60 per cent of the world's cocoa is grown in Ghana and Côte d'Ivoire (International Cocoa Organization, 2018). Cocoa production is essential to the Ghanaian economy, accounting for approximately 10 per cent of the country's GDP. Ghana's cocoa sector has expanded substantially over the past ten years, and it is estimated that 25-30 per cent of the population rely on the industry for their livelihood (around 6 million people) (Verité, 2018). This figure includes nearly 800,000 smallholder farmers (Vigneri & Kolavalli, 2018).
3.1.1. Semi-structured interviews. In the Ashanti Region, semi-structured interviews were conducted in two districts: Ejisu-Juaben and Bosomtwe district. These districts were selected because of their substantial contribution to regional cocoa production and because of their high population of migrant workers, who are documented within the secondary literature to be particularly vulnerable to forced labour (Mohammed & Apusigah, 2005). In the Western Region, two districts were selected as sites for the semi-structured interviews, Aowin and Bia West, following the same criteria as the Ashanti Region. Prior to recruiting participants, our team conducted 'community entry' processes, in which we met with community leaders, normally the Chief or his spokesperson, to explain the purpose of the project and to obtain permission to carry out research in the area. The Chief did not play a role, however, in selecting our research participants.
In terms of the demographic profile of participants, 56 out of 60 were migrants. The research team sought to recruit as many women cocoa workers as possible (although access to women workers was, at times, hampered by gender relations on the farms, discussed in more detail below). In total, 12 women cocoa workers were interviewed, of whom 10 were migrants. Similarly, the research team sought to ensure some variation in age distribution among the sample, with the youngest worker interviewed aged 19 and the oldest aged 60. Finally, participants were recruited across the four different categories of cocoa worker (see section 4.2.1): in total, eight leaseholders, forty-six caretakers, and two farm owners were interviewed, as well as four workers who worked in both day and contract labouring. This categorical breakdown is complicated, however, by the fact that some of the cocoa workers interviewed were engaged in multiple types of cocoa work. We discuss these dynamics in more detail below.
3.1.2. Digital survey. The digital survey sample was constructed in a structured way. 5 Because there was no pre-existing list of cocoa communities in the Western and Ashanti regions, the research team worked with agricultural extension officers from Cocobod, the cocoa authority of Ghana, and with data from the Ghana Statistical Service to create our own comprehensive list of cocoa districts in these regions. We then conducted an accessibility evaluation which demarcated the districts accessible to our research team. Within this, we used purposive sampling to select districts that were of varying proximity to the major cities in each region, which allowed us to incorporate diverse demographics. We further sought to encompass workers of varied age, migration status, type of employment arrangement, and gender in our sample.
3.1.3. Overview. Whilst our dataset represents a relatively small sample of women workers, it provides sufficient leverage to inform our exploratory analysis of gender and forced labour in the cocoa industry. The sample of women workers should be seen in light of the wider difficulty in accessing cocoa workers due to their remote geographic location, sparse distribution, and in some instances, the proximity of the farm owner, who either lived in the same village or, occasionally, compound as the worker. Access to women workers was further constrained by gender norms and practices, in that it was expected that the research team should prioritise men over women when interviewing respondents. As a result of this, some interviews with women workers had to be abandoned when men (often husbands or other family members), returned to the farm. Collecting and triangulating data from interview and survey sources gives us insights into the character and incidence of labour exploitation amongst cocoa workers across significant cocoa communities in Ghana.

Findings
As depicted in Table 1, our research suggests that there are four primary forms of labour exploitation experienced by cocoa workers in Ghana: extremely (and illegally) low incomes; indebtedness; wage deductions; and economic coercion (specifically, food deprivation). In terms of indicators of forced labour, experiences of non-payment, underpayment, and withholding payment were very common among workers. A significant number of workers also experienced nnaho, a form of involuntary labour, physical violence and verbal abuse, deception, and nonphysical coercion (namely food deprivation). Less common but still reported by workers were threats of dismissal and/or financial penalties and, in one instance, the threat of supernatural retaliation. Finally, one worker reported an incident of sexual violence. These forms of labour exploitation cut across both men and women workers in our study. However, we found that the precise character and patterns of labour exploitation, and their impact on workers, had a strongly gendered dimension, which we set out and analyse below.

Gendered experiences of labour exploitation
4.1.1. Gender norms, types of work, and the gender division of labour in cocoa. Our research finds that gender norms and divisions of labour are key to understanding women's vulnerability to forced labour and exploitation in the industry. Cocoa farming in Ghana operates on a sharecropping system, using two main customary arrangements, abunu (share in two) and abusa (share in three). Under the abunu arrangement, the farmland is divided into two, with the tenant and farm owner each taking one half. All the produce from the divided portion of the land then belongs to the tenant. In the case of abusa, the land is not divided, but the produce (and therefore proceeds) from the land are shared between the farm owner and tenant, with the farm owner taking two thirds and the tenant taking one third. 6 In cocoa, the type of customary arrangement also has implications for the character of the tenancy or leasehold, with those tenants sharing the land under an abunu arrangement typically having leaseholder rights over the land 7 (this arrangement is also sometimes referred to as dɔ meyɛnkyɛ, which literally translates to 'weed and let's share' in English), and with those operating under an abusa arrangement typically working as caretakers on a seasonal tenancy, also known as nhwɛsoɔ or 'caretaking'. In addition to these two types of workercaretakers and leaseholdersour study identified two other types of farmworker: 'kɔ dɔ paa' (literally 'go weed well') or contract labourers, who carry out a prescribed set of tasks over a particular period of time for an agreed price, and 'by-day' or day labourers, who are paid a fixed fee on a daily basis. As mentioned (see Section 2.2), cocoa is commonly perceived to be a 'male crop'. However, we found that this perception is inaccurate. We found that women are also working as caretakers, contract labourers, and day labourers in cocoa farming and, as such, are extensively involved in cocoa production in both the Ashanti and Western Regions of the country. Moreover, women moved between the different types of work in cocoa farming, both simultaneously and over time, for example, by working as both caretakers and contract labourers, or as both contract labourers and day labourers. None of the women who participated in our study were leaseholders, which can be understood to reflect gendered inequalities in access to land, discussed in more detail below.
Findings from the interview and observational components of our study point to a gendered division of labour in cocoa farming. However, this was not clear-cut, with working patterns and arrangements proving highly variable and dependent on context. Both women and men are involved in harvesting, carrying the pods, breaking the pods, weeding, and drying and fermenting the beans, with women most commonly undertaking carrying, pod-breaking, and harvesting, and drying activities. Activities such as spraying pesticides, pruning, and thinning the cocoa tress, on the other hand, are most commonly carried out by men, a finding that largely mirrors the division of labour described by Barrientos (2013a). Yet, beyond these broad trends, there is considerable variation. Women hired as contract, day labourers, or as the sole caretaker of a farm carry out pruning and thinning tasks, the removal of parasitic plants, and the spraying of pesticides. Where women work as family labour on the farms, the division of labour appears to be more pronounced, with women tending to weed, carry pods, break pods, harvest, and fetch water for the mixing of the pesticides (although, again, there were a number of exceptions to this).
As well as their mode of employment, the type of activity undertaken by women in cocoa production depends on factors such as the time of year, the availability of other family labour, the financial viability of hiring labourers, the age of the worker, and their relationship with the farm owner. For example, one woman caretaker, an indigene of the region who worked for her grandmother, was able to hire day labourers to spray pesticides. Another woman caretaker, a migrant and single mother, carried out all the tasks on the cocoa farm by herself, including spraying pesticides, thinning, pruning, and harvesting. In this instance, kinship and family structures coupled with access to, and availability of, inputs shape the character of work carried out by women cocoa workers. This sheds some light on the complex interplay of class, gender, and migrant status in this context, which is explored in more detail in Section 4.2.3 The types of work women carry out on cocoa farms has implications for how, when, and if they are paid, how much they are paid, and how vulnerable they are to particular forms of exploitation. Our research suggests that women are concentrated in the more precarious forms of employment within the cocoa industry. Within our in-depth interviews, women cocoa workers noted that it is difficult to get work as permanent workers because farm owners prefer to hire men as caretakers: "Most of the time the women are not given the opportunity to be permanent workers on the cocoa farm because the kind of work that someone needs to perform is quite difficult, so they hardly give it to women. Most of the time they prefer men … you may find some women involved in cocoa farming as permanent farm workers but they are only a handful. (Cocoa worker 49, female caretaker) The preference for male workers was attributed to the perception that men will be able to undertake more arduous physical labour on the farm, and also reflects historical gender norms in Ghanaian agriculture, as discussed in Section 2. Due to these gender norms and biases in hiring and employment practices, women looking for paid work in cocoa farming typically take on daily or contract positions, rather than caretaking work, which leaves them open to other forms of exploitation. In particular, we found that workers concentrated within more precarious work tend to experience higher levels of underpayment and non-payment, which we found to be key indicators for forced labour. Workers reported that farmers seek to minimise costs by underpaying them, including by making deductions for equipment or inputs (eg. cutlass, fertiliser), food, or transportation, as well as Gender and forced labour in global supply chains 1109 by imposing fees and fines, such as for securing a job within the industry. At times, such dynamics are used to create situations of debt bondage, wherein workers are made to pay off fees through unpaid labour. Workers explained to us that they often tried to recover wages lost due to non-or under-payment, but rarely succeeded.
4.1.2. Payment practices and gendered inequalities in income in cocoa. Analysis of gender differentials in payments and income is complex due to the variability and fluidity of working patterns in cocoa. Overall, results from our digital survey indicate that women are earning less in cocoa farming than their male counterparts. For those involved in caretaking work, for example, women workers' average yearly earnings are 1610 Ghana cedis ($298.15), compared to an average of 2356 Ghana cedis ($436.30) for male workers. Findings from the in-depth interviews offer further insight into this gender differential and the ways in which exploitative working practices operate through (and are shaped by) unequal gender power relations and norms, particularly in terms of limiting incomes for women cocoa workers.
Firstly, we found that women's engagement in cocoa farming is frequently restricted to less profitable types of activities (although not necessarily to activities that are less crucial to quality). Where day labourers are paid, on average, around 20 Ghana cedis ($4.23), this can vary according to activity, with workers receiving 15 Ghana cedis ($3.17) for weeding and other land preparation work and 20 Ghana cedis ($4.23) for pruning and pesticide-spraying tasks. In keeping with the broad gender division of labour outlined above, the women engaged in day labouring work in our study were more likely to be hired to carry out weeding, a pattern that acts as a constraint on their incomes. In addition to this, some respondents reported that women are paid less than men for contract and day-labouring work irrespective of the activity, a wage differential that was again attributed to the assumption that women will be able to carry out less work, or to do it less competently, than men. As one male worker put it, 'the women won't be able to weed like a man' (Cocoa worker, 21). Finally, these limits on income are compounded by women's reproductive responsibilities, as they are unable to devote as much time to (paid) productive work in cocoa. As a number of the women interview participants noted, it is the 'double burden' of their domestic work and their responsibilities on the farm that makes working in cocoa particularly difficult and constrain their earning: 'The timing is very difficult, because I have small children. I am a mother. I have to wake up early look for food for my children, prepare food for them, so by 8am I set off and by 1pm I have to return back home.' (Cocoa worker 40, woman caretaker) Furthermore, our research found that intra-household power dynamics are critical to understanding women's experiences of exploitation in cocoa supply chains. Where women were working with their husbands on the farm (rather than as the sole caretaker or as day or contract labourers), payment was typically received through their spouse, who is paid by the cocoa Purchasing Clerk or farm owner. This means that, as one participant ruefully remarked, female workers do not always receive an equitable share of the proceeds from the sale of the beans, as it is their husbands who determine how much they receive: 'There are differences, [laughs] when a man gets paid he takes some out and gives it to the wife to buy what she needs.' (Cocoa worker 40, woman caretaker) In short, a range of factors limit women cocoa workers' incomes and earning power within the cocoa industry. Women's unequal access to economic opportunities and assets is significant given that poverty is a key source of vulnerability to forced labour. For instance, one of the key vulnerability factors for forced labour in the cocoa industry is very low wages, which in cases of emergencies, cause workers to borrow money, often at high interest rates, which creates situations of debt bondage. Gendered inequalities in income are therefore significant in shaping vulnerability to more severe forms of abuse and exploitation. 4.1.3. Unequal access to land in cocoa. As mentioned in Section 2.4, women tend to be excluded from decision-making processes over land, men typically inherit land over women, and women encounter additional barriers to accessing land through tenure systems. Our findings suggest that these gendered disparities in land access shape women's experiences of exploitation in cocoa farming. Among the cocoa workers we interviewed, becoming a leaseholder was frequently cited as a key aspiration, reflecting the perceived increase in autonomy and security in the abunu arrangement and the higher share of the produce taken by the tenant. However, as noted above, none of the women who participated in our study worked as leaseholders and only four were directly hired as caretakers, with the other women working alongside husbands or other family members on the farm, and/or as day-labourers. The difficulty of accessing land for women cocoa workersas a leaseholder, caretaker, or farm ownerwas raised frequently by our interview participants, particularly in relation to the constraints this placed on their earning power, the size and productivity of the farms they were able to work on, and the types of work and contracts they were engaged in. In this way, our research suggests that gendered patterns of exclusion and inequality in land access are a key factor that blocks women workers from entering more permanent and less precarious modes of work in cocoa farming and renders them vulnerable to forms of exploitation associated with other types of work in cocoa, such as non-and under-payment of wages and forms of involuntary labour or nnaho. 4.1.4. Lack of redress and access to justice in cocoa. In addition to the forms of exploitation experienced by women cocoa workers in our study already mentionedsuch as underpayment, nonpayment, and the withholding of payment by farm owners, which were sometimes combined with verbal abuse, threats, and coercionthey also experienced involuntary labour, referred to as nnaho. Workers reported that they were required to perform unpaid and additional labour as a condition of their employment, such as working on the farm owner's land for a set period of time (typically ranging between one week and three months). As well, workers reported being compelled to take on unpaid tasks such as weaving drying mats for cocoa beans. Failure to comply would result in penalties, including being fired, deductions from their wages, and the imposition of fines. Whilst these dynamics broadly parallel the key forms of exploitation experienced by male workers, our findings suggest that gender mediates their pattern, character, and effects, here in terms of a lack of redress and access to justice.
Women reported a number of incidents where farm owners had refused to pay them for day labouring work, both in weeding and spraying pesticides. When asked how they reacted to these types of practices, women workers reported that they typically did nothing, since their position as women and migrants made them effectively powerless to challenge exploitative or coercive practices on the part of the farm owner. As one worker explained: "There are some farm owners, they are very greedy. Because of their greediness, assuming that you harvest three bags of cocoa beans as a farm worker you are entitled to one and he is entitled to two. There are some who in addition to their two will still keep your money […] It happened to me, I went to do the daily farm work type and I was never paid. I never asked the person up to date." (Cocoa worker 27, female migrant caretaker and day-labourer) The primary means of redress available to cocoa workers in case of unfair treatment is to gather a meeting of village elders, usually including the Chief, to discuss and adjudicate on the issue. As well as the tendency for these power structures to be male-dominated and to exclude women, workers reported that calling a community meeting carries a cost of several hundred cedis, which has to be paid by the worker. That said, even in the event that a worker could afford to pay this, many workers emphasised that it would be pointless to do so, since their lack of kinship ties and status within the community (and the fact that these leaders are often themselves cocoa farm owners), means that the outcome of the meeting is unlikely to go in their favour. Again, these constraints on challenging exploitative practices is compounded by the intersection of gender, migrant status, and class, since women are less structurally able to challenge unfair treatment and encounter even greater material constraints due to their lower incomes.
Women's lack of redress creates additional barriers to exerting rights and reinforces their inability to exit labour relations, no matter how exploitative they may be. Workers reported being unable to leave due to debts to employers, unpaid wages, and violence (including sexual violence) and threats of violence from employers. Our research suggests that while both men and women in the cocoa industry endure unfair treatment because they have no viable economic alternative, this is compounded for women due to the additional gendered barriers they face in accessing justice or redress. 4.1.5. Social reproduction, the household, and family/kinship practices. Finally, our research found that women's social reproductive labour (largely unremunerated work classified as 'family labour') compounds their vulnerability to exploitation in paid work. In line with SRT's insight that productive and social reproductive activities are often integrated and inseparable, our research indicates that the line between 'paid' productive and 'unpaid' reproductive labour in cocoa is extremely difficult to demarcate. On a number of occasions during our fieldwork, for example, the research team observed women carrying out 'reproductive' and 'productive' labour simultaneously, for example, by going to fetch firewood at the same time as carrying water to use in the mixing of pesticides, or by carrying out weeding on the cocoa farm, while also harvesting crops to use in the preparation of food for the family. As the sites of reproduction and production are not even spatially separate in this context, the integrated character of these processes is, in a sense, writ large.
Elsewhere, the overlapping categorisations of cocoa workers presented another challenge to the productive/reproductive labour binary. For example, ten of the women we interviewed identified as ahwesofɔ, or caretakers. However, only four of these women had been directly hired by the farm owner as the (sole) caretaker of the farm. The other women worked alongside their husbands or other male family members, who had been hired as the primary caretaker. To complicate this further, some of the women engaged as family labour also worked, on occasion, as contract and day labourers on other farmlands.This indicates that patterns of women's productive labour cannot be understood in isolation of their family relationships and responsibilities for social reproduction. This interrelationship is evident in this worker's account: "I am a working mother, working on the farm and at the same time a mother. So when I go to the farm I do it to the best of my strength […] As a nursing mother what happens is by 9am I am on the farm, by 12noon to 1pm then I am back home." (Cocoa worker 36, female caretaker) A number of participants in the semi-structured interviews reported that farm owners prefer to hire male caretakers with wives and a family to live and work on their farmlands. The reasons offered for this preference varied, including that it would make it harder for the worker to leave the farm and that it would make the worker less likely to create trouble for the farm owner. This suggests that farm owners anticipate that it will be easier to exploit workers if they have greater ties to the job, here in the form of dependants, and that they may benefit from the additional family labour that this guarantees. Put otherwise, farm owners are, in a sense, cognisant of the importance of social reproduction, that is, of the essential and inextricable role of reproductive labour in value extraction.
4.1.6. Gendered vulnerability to forced labour in cocoa. The gendered power relations and inequalities documented within this section are not distinct, but overlapping, with women workers experiencing multiple or all forms of exploitation. In the worst cases, these dynamics were compounded by gender-based violence. Some women farm workers reported experiencing verbal abuse, threats, and coercion at the hands of the farm owner, including threats that they would be fired from their job, insults, and criticisms of their work. In addition to this, one worker recounted how a farm worker's wife had been raped by the farm owner, a man who was known to have a history of sexual violence towards the wives of workers he employed. These findings are indicative of a need for further research that takes into account social relations, power asymmetries, and forms of subordination existing both within and beyond the site of production.

Conclusion
In this paper, we have documented the key role that gendered power relations and inequalities play in shaping cocoa workers' vulnerability to labour exploitation. We have demonstrated that the severity and prevalence of exploitation is shaped by workers' gender, as well as other intersections, such as migration status. We have argued that to fully grasp the role of gender in shaping patterns of workers' vulnerability, there is a need to examine not only women's conditions with respect to paid employment, incomes, and earning power as is typical in the literature, but also to investigate: broader gendered disparities in land access and tenure, including women's dependence on male family members for rights over land; gendered power asymmetries within the household; women workers' balancing of social reproductive labour and responsibilities with paid labour in the 'productive' sphere, and how the former shape their participation and conditions within the latter, including by creating pressures towards daily or contract positions rather than caretaking work; and the impacts of gender norms and biases in hiring and employment practices. Only a broad and integrated understanding of women's work in production and reproduction can yield an accurate understanding of women cocoa workers' disproportionate vulnerability to exploitation within global supply chains. Indeed, ignoring social reproduction within studies of forced labour gives a skewed perception of the causes of and potential solutions to women's exploitation in supply chains.
This underscores that lingering dualist conceptualisations of the reproductive and productive spheres within GVC analysis is not only limited, but, in the case of cocoa, premised on a flawed ontological understanding of the role and functioning of labour in this context. Put simply, there is a need to recognise the interrelationship of the two, as part of a wider set of social relations and processes that underpin and shape profitability in the cocoa industry. In our case study, we have mobilised this insight from SRT to demonstrate that the extraction of surplus value in cocoa is dependent on a regressive configuration of social reproduction, which leaves women especially vulnerable to exploitation, precarious employment, low incomes and poverty (and which is sustained by and reinforces patriarchal gender norms, power asymmetries, and other forms of gender inequality). These findings indicate that gender (and other forms of oppression) must be relocated from the margins to the centre of GVC analysis, since these relations are structurally and systemically implicated in the workings of GVCs as a whole.
Our analysis also underscores the importance for SRT scholars to expand their focus to include global South contexts, forced labour, and the specific forms of labour unfreedom experienced by those at the bottom rungs of the global labour market. Doing so elucidates the diverse forms that gendered power relations can take within the labour market, and underscores that even in severe forms of labour exploitation such as forced labour, workers are balancing responsibilities across productive and reproductive realms. In other words, households and social reproduction do not go away or cease to be relevant where forced labour is present, but rather, socially reproductive relations can be a crucial force underpinning and shaping severe labour unfreedom.