Introduction

Global Production Networks (GPNs) in the garment industry typically involve fashion brands and retailers as lead firms that subcontract out production to independent manufacturers (Alamgir & Banerjee, 2019). These manufacturers compete for the contracts based on their delivery speed, quality, and cost. Since labour is a major cost component in the industry, countries in the Global South often target the garment industry to bring about labour intensive industrial development. Ideally, this would provide opportunities for economic and social upgrading (Bair & Gereffi, 2003; Rossi, 2013). Nevertheless, with numerous countries competing to get a ‘foot in the door’, in the long run this competition may well add up to a global ‘race to the bottom’ rather than a developmental opportunity (see Anner, 2019; Barrientos & Smith, 2007).

This potential race to the bottom also draws attention to abusive practices, labour repression, and even modern slavery in garment factories as a way to cut costs and to keep wages down (e.g. see LeBaron, 2020; Pinnington et al., 2023; Stringer & Michailova, 2018). Activists’ targeting of fashion brands to force them to take responsibility for abuse by their suppliers led to the introduction of labour codes of conduct (CoCs) as a form of corporate social responsibility (Barrientos & Smith, 2007; Lund-Thomsen & Lindgreen, 2014). Despite the ambition for CoCs to ensure a set of minimum labour standards through private governance, we generally agree with criticism levelled at the ability of the codes to secure such improvements for workers. While suppliers are subjected to CoCs, buyers put them under severe pressure through demands for cost reduction and shorter lead times, thereby proverbially taking away with one hand what they seem to give with the other (Khan et al., 2020; LeBaron, 2020; LeBaron et al., 2022; Locke, 2013; Mayer & Phillips, 2017; Phillips, 2013). Rather, improvements of social and economic conditions in the garment industry require regulatory mechanisms that involve governments, civil society and—importantly for the argument here—trade unions as stakeholders (Anner, 2021; Bair et al., 2020; D’Cruz et al., 2022; Seidman, 2007; Selwyn, 2012, 2013). Anner (2021, p. 6) argues: ‘[G]overnance mechanisms for addressing decent work deficits do not simply emerge out of necessity […] they are the result of struggle and are thus shaped by patterns of worker resistance. How workers protest and what they demand, or what strategies they adopt when they protest all have a strong influence on outcomes.’ Similarly, Mashilo and Webster (2021, p. 525) argue that social upgrading is ‘primarily a result of labour agency, rather than automatically trickling down from economic upgrading’ and that this approach ‘foregrounds workers power as a crucial determinant of social upgrading’.

We take our lead from this GPN perspective, with a focus on workers’ agency and the argument that corporate social responsibility mechanisms (such as CoCs) could become strategic tools for labour organising despite their inherent weakness as a form of private regulation (LeBaron, 2020; Lund-Thomsen & Coe, 2015). However, to successfully intervene in labour rights violations, labour has to deal with the issue of their bargaining leverage in GPNs (Alamgir & Banerjee, 2019; Anner, 2015, 2019; Coe & Jordhus-Lier, 2011, 2023; Graz et al., 2020; Jordhus-Lier & Coe, 2023; Wichramasingha & Coe, 2021). The context for this is power imbalances between lead firms and suppliers in GPNs (Bair et al., 2020), as well as a power imbalance between workers and suppliers as their immediate employers (Alamgir & Banerjee, 2019; Anner, 2015). Based on a comparative case study of two instances where workers seemingly have very little power, namely the garment industries of Lesotho and Eswatini (formerly Swaziland), our primary research question is the following: What are the sources of power that workers and unions draw on in their attempts to combat the violation of their rights? Following on from this, we ask two further questions: Can CoCs, despite their proven limitations, enable workers to strengthen their bargaining position in GPNs? Also, how does the potential leverage provided by CoCs relate to other sources of workers’ power? The study is historical in nature, allowing us to show how union strategies and choices shift over time. In order to answer our questions, we propose a synthesis of sources of workers’ power as analytical framework. We use the term ‘power resources nexus’ to highlight the interplay between different sources of workers’ power. This framework also explicitly links sources of power to the transnational nature of production networks.

The argument unfolds as follows: We provide an account of the literature on sources of workers’ power in GPNs and present our analytical framework. We then discuss the methodological considerations of the paper, before we make a short introduction to the role of garment manufacturing and how codes are contested in the local settings. We discuss how labour has been able/not able use various sources of power to influence the conditions in the industry and conclude the paper with a summary and suggestions for future research.

Analytical Framework: Sources of Workers’ Power in GPNs

In responding to rights violations in conditions of labour repression, workers often face a number of obstacles (Alford et al., 2017; Riisgaard, 2009). First, in an industry such as the garment industry, direct employers themselves often have very little bargaining leverage in relation to lead firms. Even though they are subjected to the lead firms’ CoCs, due to cut-throat competition and low profit margins employers have little room to manoeuvre when facing worker demands (Anner, 2019). Second, in many cases workers can expect little protection from the state, because states in such contexts are often unwilling or unable to enforce minimum standards in the workplace. In this context of labour repression, control over labour is as much about preventing strikes and disruptions to production as it is about the need to keep wage levels down (Anner, 2015; LeBaron, 2020).

CoCs potentially open space for workers and unions to challenge violations of their rights, but we argue that this potential must be grounded in a broader understanding of how sources of workers’ power are linked to GPNs and state formations (Coe & Yeung, 2019, pp. 787–9). Our main theoretical approach is the power resources approach developed in labour studies (Brookes, 2013; Ford & Gillan, 2021; Pun et al., 2020; Schmalz et al., 2018; Silver, 2003; Trif et al., 2021; Webster & Dor, 2023; Wright, 2000). We put this in conversation with the literature that links workers’ power to social upgrading in the GPN literature (Karatepe & Scherrer, 2019; Mashilo & Webster, 2021; Mendonça & Adăscăliței, 2020).

We draw a distinction between three broad sources of workers’ power or bargaining leverage, namely structural power, associational power, and institutional power (for a summary of different approaches to this categorisation, see Ford & Gillan, 2021, p. 309, Fig. 1). Below, we discuss each of these in relation to the garment manufacturing industry. Since power is geographically defined, we link our analytical distinction between these sources of power to the national and transnational scales at which they operate (Alford et al., 2017; Ford & Gillan, 2021; Wickramasingha & Coe, 2021). We use this binary approach to scale bearing in mind that it could be refined further to include a more complex understanding of scale. The scope of this paper does not allow us to do this or to review the literature on scale as such, as well as the vast literature on workers’ agency. The analytical framework is summarised in Table 1 at the end of the outline.

Fig. 1
figure 1

South African Rand exchange rate to the US$, 2000–2022; Source: South African Reserve Bank, accessed on 8 February 2022

Full size image
Table 1 Sources of workers’ power
Full size table

Structural Power

Structural workers’ power is ‘power that results simply from the location of workers within the economic system’ (Wright, 2000, p. 962) and can result from three potential sources:

(a) Workplace bargaining power is power that results from ‘the strategic location of a particular group of workers within a key industrial sector’ (Wright, 2000, p. 962). This ‘accrues to workers who are enmeshed in tightly integrated production processes, where a localised work stoppage in a key node can cause disruptions on a much wider scale than the stoppage itself’ (Silver, 2003, p. 13). Industrial unions whose members are strategically located in assembly lines have traditionally relied on this source of power.

(b) Market place bargaining power refers to ‘the power of workers as individuals that results directly from tight labour markets’ (Wright, 2000, p. 962). This source of power can come from workers possessing scarce skills required by firms (a strategy historically used by craft unions to build their bargaining position), low levels of unemployment and hence a tight labour market, or the ability of workers to survive on means outside the formal labour market (Silver, 2003, p. 13).

(c) Since GPNs rely on transport—roads, ports, railway lines—logistical power as a form of structural power (see Lambert et al., 2012) refers to the opportunity provided by logistical ‘choke points’ (Bonacich & Wilson, 2008) for potential industrial action (see also Mendonça & Adăscăliței, 2020, p. 1066). Global unions and transnational links between unions play a central role in coordinating these types of campaigns (i.e. an element of associational power—see the discussion below), illustrating the relationality of different sources of power (Lambert, 2014).

Associational Power

Associational power refers to the bargaining leverage that accrues from the collective mobilisation by workers. Collective mobilisation becomes possible when unions organise from the local level up to a point where they have sufficient strength to cover entire sectors, industries, or territories. Unions have traditionally used nation states as organising principle and industrial unions usually form national federations to coordinate between them. When unions are organised as powerful national blocks, they can also use their strength as voters and influence political parties through funding and strategic alliances (Silver, 2003; Wright, 2000; some authors, e.g. Brookes, 2013, pp. 191–4, refer to this as coalitional power).

In the context of the rise of consumer campaigns around labour rights, the definition of associational power was expanded to include the claiming of labour rights as justice issues (usually, but not exclusively) in the transnational public domain. This form of associational power is called moral or symbolic power (Chun, 2009; Lambert et al., 2012; Schmalz et al., 2018; Webster & Dor, 2023; Webster et al., 2008 refer to this as ‘societal power’). Moral or symbolic power implies exposing unjust behaviour by powerful actors in a GPN in the public domain.

Institutional Power

Finally, institutional power refers to ‘institutionalised labour rights and institutionalised dialogue procedures’ (Schmalz et al., 2018, pp. 121–122; see also Brookes, 2013, pp. 187–191; Mendonça & Adăscăliței, 2020, pp. 1065–6; Webster & Dor, 2023: 162–167). In contexts of established unions, institutional protections of workers’ rights and their ability to bargain are often results of earlier labour struggles. Even when structural and associational power wane, workers may still have access to some form of protection by national labour law and collective bargaining procedures. In conditions of labour repression, however, the terrain of institutional power becomes a site of struggle to establish the legal and institutional space for unions to represent their members in the first place. These rights include freedom of association, the right to free collective bargaining, and the right to strike. Governments in contexts where low-wage garment factories are located are often hostile to unions and the very idea of labour regulation, due to the fear that higher wages can motivate buyers to move elsewhere (Morris & Staritz, 2016). In addition, some authoritarian states view unions with hostility due to demands for democratisation (Alamgir & Banerjee, 2019; Anner, 2015; Seidman, 2007). Table 1 provides a summary of our synthesis of sources of labour power, linked to how GPNs and CoCs, as well as national and transnational scales.

Method and Data Sources

The paper is based on a comparative historical case study, drawing on primary and secondary data. Primary data include interviews with government officials, policy makers, business managers, business associations, unions, workers, and industry observers. These semi-structured and in-depth interviews were conducted from 2008 to 2022 (see Table 2). The research was interrupted by the Covid-19 pandemic and a last round of interviews were conducted in 2022.

Table 2 Overview of interviews conducted in Lesotho and Eswatini from 2008–2022, based on category of interviewees and form of interview
Full size table

In-depth interviews (with government officials, representatives from international organisations, business associations and industry experts) were designed to get an overview of developments in the industry regarding size, composition of firms, key policy decisions, relevant laws, enforcement, etc. Further in-depth interviews (with business managers, trade unionists and some government officials and again the industry experts) enabled us to collect information on the impact of the industry on social and economic conditions. These issues related to industrial policy, CoCs, and GPN dynamics, as well as the role of unions and labour in ensuring decent working conditions and wages. These interviews lasted from approximately 30 min to 2 h. In most cases, both authors were present, taking notes and asking questions. Visiting the research sites also allowed us to make observations of factories, government and union offices, as well as the living and working conditions of workers. These interviews, field visits, and documentary evidence form the main source of data for our analysis here, since this provides us with an historical perspective. Our documentary evidence includes academic literature, industry reports from governmental and intergovernmental sources, international organisations, newspaper articles and online media accounts. This allowed us to develop an understanding of the historical situation in both countries prior to our interviews, as well as how it subsequently developed and unfolded up to 2022. This includes updating our overview of the changes in the industry relating to its size, trade, etc. (e.g. see Pasquali et al., 2021), changing government regulations and international interventions (e.g. Pike, 2020), and ongoing union campaigns.

We supplement these main two sources of data with semi-structured interviews with a total of 90 workers at nine factories in the two countries conducted by field workers in 2011, at the early stages of the research. For these interviews factories were selected based on interviews with people familiar with the industry. We included companies perceived to be progressive in their approach to labour and those seen as labour repressive. Fieldworkers ensured that the interviews reflected the gender composition of the workforce (in Lesotho 27 of 40 interviewees were women, in Eswatini 42 of 50). Given that managers often select and coach workers for interviews by CoC inspectors, we wanted our selection process to be free from interference. For this reason, fieldworkers conducted interviews outside work hours and off the factory premises (the settings were public spaces close to transport hubs and/or close to workers living quarters), based on a convenience sample. The questions concerned the knowledge of the CoCs, the wages paid, working hours, role of unions and the importance of CoCs and the working conditions in general. We do not claim a sample representative of the industry in either country and only report on qualitative findings here. These interviews lasted from 30 min to an hour, were conducted in either Sesotho or SiSwati, and were translated into English (see Table 2 above). We may decide to repeat this survey in the future in order to update our perspective on wages and working conditions in the remaining factories. However, the analysis here focuses mainly on union campaigns and we do not see this as a significant limitation. For workplace-based perspectives we have also used other more recent studies (e.g. research conducted by the Better Work programme in Lesotho).

The data analysis followed two paths. The first path addressed the historical situation and development in both countries focused on the garment industry. The second path built on the interview data related to the analytical framework (see above) and included a number of steps going from categorisation of the findings under the three types of power (structural, associational and institutional) to the detailing of the sources of power, including whether ‘traditional’ or ‘new’. Hence, we could assess the relevance and adequacy of the analytical framework and argue for the extension (see below in the Discussion and Conclusion).

Garment Production in Context: Lesotho and Eswatini

Lesotho and Eswatini are two small southern African countries that attracted significant garment manufacturing capacity in the late 1990s (Pasquali et al., 2021) in the context of the United States’ Africa Growth and Opportunities Act (AGOA) (of 2000), which provided access to US markets under the mantra of ‘trade, not aid’. Both Lesotho and Eswatini are landlocked countries, with Lesotho completely surrounded by South Africa and Eswatini by South Africa and Mozambique. More than 50% of Lesotho’s population and close to 60% of Eswatini’s population live below the poverty line. Both countries have largely rural populations, with Lesotho’s urbanisation rate of 29.5% and Eswatini’s rate of 24.1%. The garment industry matters to such localities as the (formal) employment and income generated is important to social and economic conditions. Their currencies (Maloti and Emelangeni) are tied to the South Africa Rand and both countries are part of the Southern African Customs Union (SACU), along with South Africa, Botswana, and Namibia. Both countries’ state budgets are propped up by income from SACU, at times up to 70% (Interview 1). This means that the civil service bureaucracies are insulated from underdeveloped formal economies and the need to raise taxes, but also vulnerable to shocks when SACU incomes decline.

The Kingdom of Lesotho is a parliamentary democracy with a constitutional monarch, but the country is prone to military coups and severe political instability due to contestation between fractious political parties. The government derives its legitimacy from its ability to mediate access to powerful external actors and funding. A history of external intervention goes back to British colonialism and the apartheid era, when South Africa sponsored a military coup (Maliehe, 2021; Turkon, 2008). The country’s labour regime is based on dated labour laws, which were drafted during the era of military rule and attempts to redraft labour laws are hampered by political instability. On paper, though, labour laws comply with core ILO conventions (Mosito, 2020). Eswatini (formerly Swaziland), is Africa’s last remaining absolute monarchy and a labour repressive regime. Unions have historically been one of the main civil society blocks opposing absolute rule (Levin, 1997). The country’s government treads a fine line between internal repression and external appeasement and unions often have to negotiate their own role in this dynamic. Eswatini is marketed as an investor friendly location, with business-friendly laws and investment incentives (EIPA, 2022). Given the country’s history of labour repression and status as an absolute monarchy, being included in the AGOA faced strong criticism in the US and it took considerable political manoeuvring from the Swazi state before it succeeded, only to be excluded in 2015, and re-admitted yet again in 2017.

The garment industry emerged in Lesotho and Eswatini in the 1970s and 1980s, mainly due to South African investments directed at exports to South Africa. From the late 1990’s, garment manufacturing grew substantially (Pasquali et al., 2021) when Asian (mainly Taiwanese) companies opened manufacturing operations to take advantage of comparatively low wages and factory shells in industrial parks set up by government agencies, namely the Lesotho National Development Corporation (LNDC) and the Swaziland Investment Promotions Agency (SIPA). These investments got a substantial boost with the promulgation of the AGOA by the US in 2000. This was assisted by the comparatively low value of the South African Rand (to which Lesotho and Eswatini’s currencies are tied) to the US dollar, which meant that already low wages were even more competitive (see Fig. 1). At the industry’s peak in the early–mid-2000s, approximately 50,000 workers, mostly young women, were employed in these factories in Lesotho, with 40,000 employed in Eswatini at the same time. These numbers were significant, due to the small size of the two countries (Lesotho has a population of 2.2 million and Eswatini a population of 1.1 million). Also, both countries have small formal economies, with the majority of the population relying on subsistence agriculture (Morris & Sedowski, 2006; Morris et al., 2011).

Following this first wave of new factories and job opportunities, the industry was in for a shock. The South African Rand appreciated significantly to the US$, mainly due to platinum exports (up from less than ZAR11.00 to the US$ in 2002 to nearly ZAR6.00 2 years later—see Fig. 1). This appreciation more than doubled the relative cost of labour. Also, in January 2005, the Agreement on Textiles and Clothing (ATC), which limited exports from China to US and European markets, expired. Several companies moved out again, seeing employment figures diminishing. In both Lesotho and Eswatini, remaining companies redirected some of their production to South African markets and additional South African manufacturers relocated to take advantage of lower wages and the fact that three countries were members of SACU. Up until the Covid-19 pandemic, employment in garment factories eventually stabilised at around 45,000 in Lesotho, and 22,000 in Eswatini (Pasquali et al., 2021).

As could be expected, the Covid-19 pandemic impacted negatively on the industry in both countries. Lesotho’s continued reliance on exports to the US, as well as the sourcing of significant inputs from China, hit the country’s industry hard. This was aggravated by South African border closures, also because South African factory managers were not able to get to their factories across the border. During a field trip in 2022, several interviewees informed us of large firms that had either downsized operations or moved out of Lesotho altogether. We also observed first hand a number of closed factories, while, somewhat paradoxically, two additional industrial parks had been opened. The closure of a significant number of manufacturing operations caused speculation over whether this was related to Covid-19 lockdowns, or whether the country’s reputation as a ‘sweat free’ sourcing destination had been damaged by a high-profile campaign around gender-based violence (GBV) at some of the factories. This campaign happened from 2018 onwards and involved an international NGO that involved local unions. Several of the people we interviewed in 2022 speculated about whether the fall-out of this campaign explained the decline in orders from international buyers (Interviews 2, 3 and 4). Hence, these observers estimated that only about 25,000 persons are employed in Lesotho. We provide more detail on dynamics around this campaign later in the paper. Because the industry in Eswatini had already reoriented it exports towards South Africa (due to the country’s temporary expulsion from the AGOA), it was less hard hit than in the case of Lesotho. Nevertheless, supply lines from China were severely disrupted and South African border closures also impacted negatively on the industry. Some factories temporarily switched to the manufacture of protective clothing such as face masks during the lockdown period (see Pasquali & Godfrey, 2022 for more detail on Eswatini and Covid-19). Tragically, a South African factory manager was not able to seek medical attention in South Africa and died from Covid-19 as a result. In both countries, the industry continues to be based on the foreign ownership of the garment factories, with little success in growing local producers.

In both countries, wages and working conditions in these garment factories are controversial. In Lesotho, which mainly exports to the US, unions and international NGOs protested poor conditions in the factories from the onset of the early 2000s (Gibbon, 2003). In response to this, Lesotho’s government and manufacturers promoted the importance of CoCs in marketing the country as ‘sweat free’ (Seidman, 2009). Different interventions were staged to show the effects of the codes. From 2010 to 2016 the International Labour Organisation (ILO) ran the Lesotho Better Works Programme (BWP) (Pike, 2020), an example of what is known as a multi-stakeholder initiative (MSI), involving the state through public governance, unions, non-governmental organisations (NGOs), as well as business associations (Barrientos & Smith, 2007; Lund-Thomsen et al., 2021).

In the early 2000s, Eswatini received less international attention compared to Lesotho, despite the main bulk of garment exports being divided between the US and the EU market (and a minor part at this stage going to South Africa). Still, campaigns from international unions, the Congress of South African Trade Unions (COSATU) and Swazi unions called for actions against labour repression in the country. Eventually, as mentioned, the country was expelled from the AGOA, but reinstated again. These contextual factors shape the extent to which workers and unions can draw on sources of power to respond to violations of their rights.

Workers, Unions, and Sources of Power

We now analyse how workers and unions have been campaigning to improve working conditions and labour rights in the garment industry in the two countries. We structure our presentation along the three main categories of workers’ power, namely structural, associational, and institutional power. In each case, we discuss Lesotho first, followed by a discussion of Eswatini as point of comparison.

Structural Power

As could be expected, workers in both Lesotho and Eswatini’s garment manufacturing sector have little structural power (Anner, 2019; Silver, 2003). In both Lesotho and Eswatini, there are differences between production networks directed at exports to the US market (mostly Taiwanese owned firms) and production networks directed at the South African market (a combination of South African and Taiwanese firms). The fact that Eswatini got expelled from the AGOA meant that most of its production is now directed at the South African market (see, e.g. Pasquali & Godfrey, 2022; Pasquali et al., 2021). In both countries, production networks directed at the US market are locked into high-volume, low-cost production. Manufacturing operations in the countries typically produce T-shirts, sweaters, undergarments, etc. In both countries there are lower tier manufacturers who are subcontractors to other more established operations who got contracts from overseas buyers. This further reduces the workplace bargaining power of workers, since production can be shifted between different operations. Workers in these factories had very little workplace bargaining power due to the cut-throat competition between employers for contracts from fashion brands. As the Taiwanese owner of a factory in Eswatini explained: ‘In 2000, Wal-Mart paid 82 US dollars for a dozen of jeans, in 2007 the price was 57 US dollars. With costs going up, I was losing my breathing space… The unions talk about inflation, but in the market for jeans it’s about deflation!’ (Interview 5). When we revisited this company in 2016, it had moved out of the US market and focused on protective wear, with 85% of their production destined for the South African market. The same factory owner explained: ‘Industrial clothing is also a tight market, but at least there is a little breathing space’ (Interview 5).

Production networks directed at South Africa tend to be for the higher end of the consumer market, either fast fashion, or work-wear (Pasquali et al., 2021). In both countries, South African-owned manufacturers tend to service a lower volume and higher value-added segment of the retail market. A South African operation in Lesotho manufactures work-wear (coveralls, protective clothing, uniforms, etc.) for export to the South African market and beyond, and in Eswatini a South African company has invested in machinery that can handle higher-end fabrics such as chiffon, as well as the skills of their workers to operate these machines. In theory, at least, these workers have more bargaining leverage than suppliers to the US market, due to higher levels of capital investments by manufacturers. However, South African retailers have lower levels of interest in CoCs than their US and European counterparts. We found that there was an awareness of the structurally precarious position of workers among government officials and unions alike in the two countries. Eswatini’s (then Swaziland) monarch famously once described foreign investors as ‘birds’ that can easily ‘fly away’ (Fransman, 1981, p. 75). During our interviews with unions, we were also struck by how concerned labour activists were about workers losing their jobs due to manufacturers withdrawing their operations from the country.

When it comes to marketplace bargaining power, we should mention that formal sector employment only accounts for 10% of total employment in Lesotho and 15% of total employment in Eswatini. Due to the initial surge in employment levels in the early 2000s and the subsequent decline, there is an oversupply of skilled workers in both countries. This is especially true for manufacturers in production networks directed at high-volume, low value-added production networks. This means that workers employed in a labour intensive sector such as garment manufacturing have very little bargaining leverage. In both countries, the majority of the employed or economically active adult population work in the informal sector, mainly in subsistence agriculture on ‘trust land’, or land that resort under the authority of traditional leaders (Government of Lesotho, 2018, p. 5; Schubert, 2021). Since unemployment levels in both countries are extremely high, it is relatively easy for firms to replace workers in the context of a labour surplus economy. When interviewed, workers cited examples of companies using replacement (‘scab’) labour during strikes and in the case of Eswatini workers who worked during a strike even got police protection. A 22-year-old woman, who worked in the cutting department, explained: ‘With the last strike… which lasted 2 weeks, we were told not to take part and security was provided by the government to help us go to work. At work there was police. Later, after the strike, the minister came to the factory and he thanked us for not taking part in the strike action. He said it showed that we appreciate[d] our working conditions’ (Interview 6). In both countries workers, we interviewed feared losing their jobs and some chose not to join unions.

When it comes to logistical power as a form of structural power, both Lesotho and Eswatini are landlocked countries that are dependent on mainly South African (and Mozambican, in the case of Eswatini) transport infrastructure for imports and exports. This raises a potential pressure point for unions in both countries to use but would depend on cooperation and solidarity from South African unions in the transport sector. In both cases, this has not been explored in any depth by unions. We should, however, mention that Eswatini’s pro-democracy movement, of which the labour movement is a part, regularly blockades the country’s border in association with South African unions. Eswatini’s pro-democracy movement has offices located in the head-office of COSATU and maintains strong links to their South African counterparts. In fact, this has become a recurring protest repertoire. Nevertheless, these actions tend to be sporadic and directed mostly at the Eswatini state, and not companies and collective bargaining campaigns for better wages and working conditions (Tati, 2020).

In the case of both countries, we see that structural power is limited on all three dimensions. Workers and unions are generally forced to explore other forms of power, to which we now turn.

Associational Power

Our comparison between the two countries shows that unions use both collective mobilisation and symbolic leverage as sources of associational power. However, unions in Lesotho tended to draw on the symbolic leverage linked to CoCs, whereas the unions in Eswatini tended to rely on the traditional organisational leverage based on collective bargaining with employers. In both countries unions have directed some of their demands at the state, but the ability to use links to political parties remain constrained. Also, in both countries union leaders tend to be male, whereas the membership of unions is predominantly female (see Gibbs, 2005; McCarthy et al., 2021).

In Lesotho a significant share of the workers reported that the unions visited the factories every week, still, union membership turned out to be somewhat limited, in our case about a quarter of the workers that we interviewed. Given that the labour law stipulates that at least 50% of the workers need to be members of one union for the union to receive recognition and engage in collective bargaining with the factory owners, this was a problem. In addition, rivalry amongst the unions, in particular between the two major unions at the time, namely the Lesotho Clothing and Allied Workers’ Union (LECAWU) and the Factory and Allied Workers’ Union (FAWU), undermined the opportunities of receiving such recognition (see also Gibbs, 2005). A 33-year-old male sewing machine operator explained why he was not a member of a union: ‘I have only one reason when I consider that the type of service is not good. I have been a member in the past, but this time around, I decided I will not join. I was a member when I was retrenched, and I paid for nothing.’ He also mentioned inter-union rivalry between FAWU and LECAWU at the factory: ‘In this company there are two unions, FAWU and LECAWU. They always compete for membership and this has created problems within the unions themselves’ (Interview 7).

While we noted the presence of the unions at a weekly basis at the factories, the workers informed us that they mainly experienced the unions as being interested in collecting membership fees. Few of those we interviewed were union members. ‘I have no interest’, said a 27-year-old trimmer, ‘I am just concerned with my job’ (Interview 8). A 26-year-old male supervisor negatively described unions as ‘people coming from outside’ (Interview 9). Some workers we interviewed were quite cynical about unions. As one, a 30-year-old machinist, remarked: ‘It’s the same either way, you will go if they expel you. So there is no need to waste money on a subscription’ (Interview 10). This lack of direct interaction between union officials and their members may be related to their strategy of engaging buyers directly and not combining this form of leverage with workplace mobilisation. Union officials told us that problems could be handled more effectively by copying their communications with local companies to the buyers directly. In the local parlance this strategy was described as ‘faxing LA’—or sending communication to Los Angeles, where the head-office of the GAP, one of the major buyers, is located (Interview 1). Our initial interviews with workers were conducted in 2011, but we should mention that this sense of distance between union officials and their members remained an issue that the Better Work Programme also commented on (see Pike, 2020).

To be sure, unions tend to rely on symbolic pressure, rather than real workers’ organisation. Compared to doing the groundwork of organising at the factory level, the codes seemed to have provided the unions with a much ‘easier’ route to make their voices heard, namely using the fax numbers to the international buyers to send information about non-compliance.

In May 2015, after considerable effort from IndustriALL’s regional structures, FAWU and LECAWU, with another smaller union merged to form the Independent Democratic Union of Lesotho (IDUL). It was hoped that this would bring about much needed unity in the industry and limit union rivalry. However, already in November 2014, the National Clothing Textile and Allied Workers Union (NACTWU) was set up as rival, and some of the organisers of the merged unions joined the smaller United Textile Employees (UNITE), which had been formed in 2008. This means that formal workers unity in Lesotho remains elusive, although the three unions at times jointly campaign on common issues.

We mentioned one such campaign earlier on in the paper, namely a high-profile campaign around gender-based violence and harassment at a specific textile manufacturer that happened from 2018 onwards (see Worker Rights Consortium, 2019). The campaign was spearheaded by a US-based NGO called the Worker Rights Consortium, supported by the AFL-CIO’s Solidarity Center. The NGO involved local unions in a campaign around sexual harassment of workers at the Nien Hsing Textiles Company in Maseru and put pressure on the Taiwanese company’s buyers (including Levis Strauss) to hold them to account. After the publication of a research report that detailed the harassment of women workers by mainly male supervisors, a code of conduct was introduced to address the issue. This code, which came into effect in 2021, allowed for the local monitoring of steps against the problem and included penalties from buyers in the form of reduced orders if the stipulations of the code was contravened (Worker Rights Consortium, 2021). Before the campaign and the Covid-19 lockdown, Nien Hsing Textiles employed approximately 10,000 workers. Between 2020 and 2022, the company closed several of its operations in the country and laid off 6900 employees, blaming a reduction in orders from the US (Mphatsoane & Mokhopi, 2022). In this context, speculation was rife about the role of the campaign in decisions by US buyers to reduce their orders for denim garments from Lesotho (Interviews 2, 3, and 4).

In Eswatini, the option of political access to the state is not available, since the country is ruled by an anti-union, absolute monarch (though the unions have been and continue to take part in the pro-democracy movement). Rather, unions here have focused on building organisation and establishing collective bargaining with employers. Despite internal divisions like those in Lesotho, as well as state labour repression, Swazi unions have managed to forge higher levels of worker unity than in Lesotho with the formation of the Amalgamated Trade Unions of Swaziland (ATUSWA). Also, in contrast to Lesotho, unions in Eswatini were initially reluctant to draw on the symbolic leverage afforded by CoCs. Their reluctance is explained by events in the early 2000s, when a trade union contacted Wal-Mart directly about the abuse of workers’ rights at the factory of one of their suppliers. According to these unionists, Wal-Mart’s response was to cut the contract with the offending firm and all the workers employed there lost their jobs as a result. The union subsequently decided not to contact buyers, but rather to organise at the factory level (Interview 11).

At the time when we conducted our first round of interviews (in 2008), the most active union in the garment industry was the Swaziland Manufacturing and Allied Workers’ Union (SMAWU). SMAWU was affiliated to the Swaziland Federation of Labour (SFL), described by Simelane (2007) as ‘workerist’, meaning that they focused on shop floor organisation, rather than politics. Put differently, whereas the competing Swaziland Federation of Trade Unions (SFTU) openly supported the pro-democracy movement, the SFL was more pragmatic about its engagement with the state. Nevertheless, this shifted somewhat during the 2008 garment industry strike, which was supported by 93% of workers in a strike ballot. SMAWU’s rivals were the Swaziland Processing and Refining Allied Workers Union (SPRAWU), affiliated to the SFTU. As a result of SMAWU’s initial success in increasing workers’ wages through centralised collective bargaining, SPRAWU lost many of its members to SMAWU, which clearly had wide popular appeal among workers due to the strike. As in the past, police met the strike with violence. The strike ended when the government agreed to address transport and accommodation grievances. When they failed to do so, the union threatened to resume the strike. Employers then collapsed their own collective bargaining association, thereby effectively putting an end to centralised bargaining. SMAWU relied on associational power—its ability to mobilise workers but could not sustain their action. When the bargaining structure collapsed, eventually SMAWU also collapsed.

Though it took time to regroup after their defeat in the strike of 2008, the unions continued their efforts on collective mobilisation. In 2012, Eswatini’s two major trade union federations, the SFTU and the SFL decided to merge into a new federation called the Trade Union Congress of Swaziland (TUCOSWA). Eventually the deliberations concerning AGOA (see the next section) also brought garment sector unions together as they (together with seven other unions) formed ATUSWA in 2015.

As mentioned, associational power also refers to pressure by unions on the state to enforce labour rights. In both countries, this possibility is constrained, but Lesotho allows for more scope in this regard than Eswatini. Political parties often use working conditions in Lesotho’s garment manufacturing industry as an election issue and at times prominent union leaders also seek political power. For example, in the run up to the elections in May 2012, a coalition of opposition parties promised workers that they would work towards a ‘living wage’ in the industry should they come to power. Their campaign was successful and immediately after the election unions started to put pressure on the new government to increase drastically the legislated minimum wage. In 2021, workers in Lesotho went on strike due to a failure of the government to implement the agreed minimum wages. Unions initially jointly organised the strike, but when the strike lost steam unions publicly blamed each other for this. Protest in Eswatini also continued. In 2021, the pro-democracy movement in Eswatini staged several protests against authoritarianism in the country, also resorting to blockading the border as part of the established protest repertoire. These attempts by unions to put pressure on the state brings us now to a more detailed discussion of institutional power.

Institutional Power

In the context of states that are either unwilling or too under-resourced to provide for the legal and institutional framework labour dispensation that enables unions to represent their members, the very terrain of institutional power becomes contested. This is the case in both Lesotho and Eswatini, where unions have drawn on international pressure to open this space. Both countries have ratified the core ILO conventions, but the devil is often in the detail of labour law, as well as in its regulation and application (Mosito, 2020).

Again, we start our discussion in Lesotho, where the state initially ‘tolerated’ the unions, but this had changed over time. The Lesotho Labour Code of 1992 and subsequent amendments technically meet ILO obligations, but in practice, is ‘ill equipped to deal with modern developments’ (Pike, 2020, p. 923; see also Mosito, 2020). Dispute resolution institutions are overwhelmed by backlogs and the labour inspectorate is under-resourced. In contrast to the flashy office building that houses the LNDC in Maseru, the Department of Labour’s office is in a run-down part of town, in a building that is equally run-down. When we interviewed the Labour Commissioner in 2008, the inspectorate had one vehicle for the entire country, which also had to be shared with other departments (Interview 12). In addition to a lack of resources to enforce labour rights, the state routinely responds to strikes violently, and resorts to turning off cellular phone networks during protests (Pike, 2020; Salm et al., 2002).

In the garment industry, labour conditions were so dire that international campaigns were launched around the issue. Major brands that sourced garments from the country introduced CoCs. These had some impact, but primarily on health and safety conditions, rather than providing for collective voice. Relationships between Basotho workers and mostly Taiwanese supervisors remained conflict ridden and strained. Most of the workers we interviewed in 2011 were aware of CoCs, but a common refrain was that ‘the target is the most important.’ A 32-year-old trimmer said: ‘All they [management] care about is that they reach the target for the day’ (Interview 13). Also, workers often did not know what CoCs entailed. A 23-year-old machinist related how codes inspections were carried out: ‘Those who come are here maybe once a year and these are buyers. When they come all of us have to put on our nose bags and when you don’t have it you have to go home that day, otherwise it might be a bad reflection to the visitors’ (Interview 14). A decade later, our findings from 2011 were confirmed by the ILO’s BWP’s baseline study when similar patters were reported (Pike, 2020).

As a way to address the abuses of workers’ rights, the ILO launched the BWP in December 2010 and formally ended it in June 2016. The intention was for it to assist the industry to improve working conditions and labour standards as it had done in Cambodia (Polaski, 2006). BWPs are 5-year initiatives aimed at strengthening social dialogue between government, industry, and unions, in an attempt to enhance labour conditions without impacting negatively on competition. The programme was initially funded by the US Department of Labour and the intention was for it to become self-sustaining, with funds contributed by manufacturers themselves (Pike, 2020).

There was some evidence that the programme succeeded in improving workplace relations, ironically because it provided for direct worker participation in the structure that bypassed unions. This was due to union weakness and rivalry (Pike, 2020). But there were tensions between the programme and the Department of Labour, as some of the officials at the BWP confused their role with those of labour inspectors (Interview 15). Some of the managers we interviewed felt positive about the programme since disputes were resolved in a more pro-active manner while the programme was running (Interview 16).

Manufacturers locked into production networks directed at US markets were supportive, in part due to pressure from their buyers. In the end, the programme closed because the Lesotho government was reluctant to make participation to the programme mandatory and South African manufacturers were not interested in contributing financially in order to keep the programme going. Also, manufacturers expected their orders to increase due to participation. In reality, they thought the opposite was true (Interview 17). According to Pike (2020, p. 922), the Lesotho government’s reluctance was ‘in part attributable to frequent changes in government and low prioritisation on the issue, but also to fear that added pressure on employers would lead them to shift production to lower-cost locations.’

Towards the end of the programme evidence pointed to much of the gains made during its existence being eroded and the adversarial approach to supervision that was prevalent before the introduction of the programme returning (see Pike, 2020). The BWP’s closure meant that the potential to impact on institutions was not realised. The Department of Labour remains under-resourced and unions divided (Interview 18). This is underscored by the fact that a matter such as the sexual harassment of women workers by male supervisors became the focus of the campaign discussed earlier in this paper from 2018 onwards. It is noticeable that the state was one of the signatories to the agreement made in 2019. Furthermore, the subsequent reduction in orders from US-based buyers (also signatories to this campaign) and the resulting job losses of among those women who were supposed to be the beneficiaries of the agreement underscores the importance of the need for state involvement and upgrading in matters related to basic rights in the workplace.

In Eswatini labour relations are often at the heart of challenges to the political regime. As in the case of Lesotho, the country’s labour laws meet ILO requirements at the technical level, but industrial action is difficult due to cumbersome processes that must be followed before strikes are legal. Also, police routinely respond to strikes violently. The terrain of opening space for institutional power remains part of labour’s struggle. Unions in Eswatini, though, have been able to build up a unified front in confronting the matter.

The Department of Labour in Eswatini is also better resourced than in Lesotho, with a former Labour Commissioner who took an active interest in the social causes of low-wage employment (he even did his Master’s dissertation on the topic at a South African university) (Interview 16). However, due to the role of the monarchy in the state bureaucracy, lines of authority are often ambiguous. While company visits by labour inspectors seemed to take place regularly, the content of the visits remained unclear and the impact negligible, if not invisible. The US State Department wrote in a report on Swaziland: ‘[T]he Labor Commissioner’s office conducted few safety inspections because of staffing deficiencies and an alleged desire not to “scare off foreign investors”. Workers neither had legal statutory rights to remove themselves from dangerous work places without jeopardising their continued employment, nor did any collective bargaining agreements address the matter’ (Solidarity Center, 2006, p. 31). Only 12% of the workers we interviewed were aware of government inspectors visiting their factories to inspect working conditions. The financial manager of a major Taiwanese operation said of government labour inspectors: ‘They come maybe once a year, and only when there is a major issue in the newspaper.’ According to him, government takes the approach of ‘no news is good news’ (Interview 20).

The lack of institutional power—indeed, the active undermining of an established system of labour rights by the state in Eswatini—became a key issue when the country was expelled from the AGOA. When the SFL and the SFTU formed TUCOSWA in 2012, both the two former federations were still formally registered. The state now argued that the law did not allow it the scope to register the new federation. In response to the state’s clampdown on workers’ organisation, the USA expelled Eswatini’s from the AGOA, effective from January 2015. In turn, Eswatini suspended not only labour federations, but also employers’ associations and unions (notably the newly formed ATUSWA). Garment exports to the US from Eswatini declined by US$50 million from 2014 to 2015 and the value of this was only made up by 2019, but with increased exports to South Africa (Pasquali & Godfrey, 2022, p. 482).

In order to be re-admitted to the benefits of AGOA, the US administration required Eswatini to amend labour laws to allow for registration of trade union and employer federations, to amend repressive security legislation, to remove the civil and criminal liability to union leaders during protest actions, and to implement a code of good practice on protest and industrial action (see Khumalo, 2014). At first, the government publicly blamed unions for the loss of AGOA benefits. Clampdowns on protests and union activism continued, but amendments to laws were nevertheless made. This was done without consultation with stakeholders and the outcomes were a far cry from what was really required. The American Federation of Labour-Congress of Industrial Organizations (AFL-CIO) initially opposed Eswatini’s re-admission to the AGOA, but ATUSWA as their counterparts in Eswatini, supported by their Global Union IndustriaALL’s representatives, argued in favour of re-admittance in the hope that the industry would take off again and throw a ‘lifeline for young women who constitute over 90% of the workers in this industry’ (ATUSWA TUCOSWA IndustriAll, 2017, p. 2; Interview 21). In mid-December 2017, the USA’s Trump administration announced Eswatini’s re-admission to the AGOA. While clearly not ideal, the outcome of state’s response to the expulsion was to allow the newly formed union and the union federation to register formally and to operate legally. The unions succeeded in using transnational pressure to secure a certain level of national institutional power.

Discussion and Conclusion

In our comparative case study of the garment industries of Lesotho and Eswatini, we have identified the sources of power that workers and unions draw on in their attempts to combat the violation of their rights. We also considered how CoCs, despite their inherent limitations, enabled workers to strengthen their bargaining position in GPNs. Our analysis linked our understanding of sources of workers’ power to how GPNs operate at both national and transnational scales. Given the precarious nature of the inclusion of production operation in the GPNs and RPNs, it is noteworthy how persistent the workers and unions have been in their contestations and what they have been able to achieve. That workers have agency is not in question—this is clear from the number of campaigns and strikes that happened in both countries. Nevertheless, agency is constrained by structural locations (see Coe & Jordhus-Lier, 2011, 2023) and for unions contingent on strategic choices around the mobilisation of campaign resources (see Table 3).

Table 3 Summary of main findings
Full size table

The two case studies highlight the nexus between various sources of workers’ power and emphasise how they could be used in combination. The two cases illustrate two divergent paths. In Lesotho, local organising took second place in the context of a primary focus on the symbolic leverage made available to them as part of CoCs of US-based firms. In contrast, unions in Eswatini focused primarily on local organising and collective bargaining at the expense of global campaigns. The analysis presented here points to both possibilities and limitations of how unions in the two countries made strategic campaign choices.

We would be reluctant to describe the one case as a success and the other as a failure. Rather, we are interested in what could be learned when the two case studies of resistance against conditions of exploitation are contrasted. As mentioned, a key point for us is the need to identify and contextualise potential sources of power, and to then consider the nexus between the available sources of power. In both countries garment workers lack structural power. Regarding workplace bargaining power this is due to the buyer-driven nature of the GPN and RPNs. They also lack marketplace bargaining power due to high levels of unemployment and the availability to skilled garment workers in the wake of the downturn in the industry. Unions in Eswatini use logistical power but directed at campaigns for democratisation rather than at conditions in the garment industry.

In this context, workers and unions in both countries have primarily focused on associational power. Unions put in significant effort to mobilise workers, to increase unionisation, and to achieve increased associational power vis-à-vis garment manufacturers. Nevertheless, unions in the two countries took different approaches. In Lesotho, union officials tended to focus on the power that CoCs and international scrutiny gave them in their day-to day raising of worker grievances. In the local parlance, they ‘faxed LA’. Also, unions remain divided, despite a significant attempt by the regional structure of their Global Union to bring about worker unity. Lesotho, then, illustrates how divisions among workers and a lack of union initiatives can lead to CoCs standing in for organising, instead of making it possible. Regarding the power resources nexus, symbolic leverage stood in for collective mobilisation, rather than reinforcing it. One reading of the situation is that the unions are essentially there to provide political career paths and access to international resources for officials, rather than members. Very much like the state in Lesotho, unions run the risk of being seen to draw their legitimacy from international donor connections and funding, rather than any real organisation. Hence, unions themselves may reinforce the tendency for CoCs to deal with factory conditions rather than the right to representation.

Regarding the use of associational power, Eswatini presents an interesting contrast. Here, unions were reluctant to draw on CoCs and connections to buyers to raise abuses of their members’ labour rights. Rather, they developed local structures and set up industry-wide collective bargaining for the first time in the country’s history. Here the nexus between power resources meant that unions were reluctant to enforce their very real attempts to open new institutional space for collective mobilisation by means of the symbolic leverage available through CoCs. Nevertheless, they stretched their achievement too far when they directed their demands at the state. On this front, they were vulnerable. Employers were able to collapse the collective bargaining arrangement in the absence of major international scrutiny as well as a lack of institutional power to back-up the collective bargaining process. Regarding the power resources nexus, this point illustrates how collective mobilisation as form of associational power stands in relation to both moral/symbolic power and institutional power. Where unions in Lesotho focused on moral/symbolic power at the expense of collective mobilisation, unions in Eswatini focused on collective mobilisation at the expense of the moral/symbolic power that communication and co-campaigning with the consumer movements in the main markets supplied by their employers could have provided.

In both Lesotho and Eswatini the state is hostile towards unions. This means that national institutional power is a terrain of contestation in both countries—unions must open up space for representational rights rather than just defending them. In both countries international institutional power provided opportunities for this. In Lesotho, the BWP had the potential to transform government’s approach to labour regulation. However, the programme bypassed both the state and unions—the latter due to negative union rivalry in the workplace. Rather, Better Work officials set up communication channels with workers that bypassed the unions. A unified labour movement could have strategically used the BWP to enhance the state’s capacity to enforce labour law and to open space for collective mobilisation. This also points to the relationality of sources of power and the nexus between them—without collective mobilisation the transnational dimension of multi-stakeholder initiatives or hybrid forms of labour regulation may prove unsustainable. In Eswatini the unions regrouped after their defeat in 2008 and set up a unified front. It is interesting that they decided to combine industrial sectors under one union to pool resources. This contrasts with Lesotho, where divisions and union rivalry remain. However, the state in Eswatini responded to this potential new force with further repression and by refusing to recognise and register the newly formed federation and trade union. Here the AGOA did have a certain impact and provided the unions with transnational institutional power to open the space for formal recognition. Where the AFL-CIO took a principled line, the Eswatini union and their Global Union took a more pragmatic approach. They appreciated the need for strong collective organisation as prerequisite for impact in all terrains of mobilisation.

Our analysis shows that unions primarily drew on associational power and to some extent on transnational institutional power to respond to economic exploitation and political oppression. Apart from the need for internal union democracy and worker unity, does the analysis here point to additional avenues that unions could potentially explore? Structurally speaking the RPNs directed at South Africa tend to have higher levels of investment in equipment and worker skills than the GPNs directed at the lower-end of the US market. This has implications for the marketplace bargaining power of workers who are employed at these factories. While South African retailers have a more limited commitment to CoCs due to the country’s underdeveloped consumer activist movement, the factories that supply them have more vested interests in their factories than their counterparts focused on the US market. As an organising strategy, unions could build their power in these factories. Another opportunity is to explore strategic use of the logistical points of pressure between the two countries and South Africa in association with South African unions in the way Swazi activists have been putting pressure on the political regime in Eswatini. There is the danger of overstating this point, since workers and unions in the two countries are acutely aware of their vulnerability within the power dynamics underlying production networks that precariously incorporate them.

Finally, we would like to make a theoretical observation linked to our empirical findings: Sources of workers’ power are analytically distinct but are relational and operate best when seen as mutually reinforcing. Union organising can potentially benefit from a careful examination of the nexus between sources of power, in addition to the relative strength and strategic targeting of one power resource rather than the other. We linked our analytical framework of sources of workers’ power to GPNs, CoCs, as well as the scales at which unions pitch their campaigns. We also benefitted from an historical analysis to show how these campaigns changed direction over time. Accordingly, we suggest that these refinements constitute our contribution to the analytical framework and the field as such.

The intervention is made in the spirit of the South and Southern African tradition of critical engagement (see Bezuidenhout et al., 2022) as an attempt to provide a perspective rooted in the Global South—both empirically and theoretically. Hence the decision to locate the study in two African countries that have seen significant investment in the garment industry. It is our hope that our synthesis of sources of workers’ power and a nexus approach to these sources of power might assist the international labour movement, NGOs and likeminded in efforts to strategise around the use of power, as well as create an awareness of how potential sources of power are relational and could be used in combination as a result. In addition, these opportunities are circumscribed and shaped by power dynamics within GPNs and RPNs, as well as how these dynamics are embedded in processes that are scaled from the national to the transnational (see Coe & Jordhus-Lier, 2023; Webster & Dor, 2023). Future research could advance the framework further through studies of the garment industry in other countries in the Global South, as well as other (global) industries.