South Korea’s Automotive Labour Regime, Hyundai Motors’ Global Production Network and Trade-Based Integration with the European Union

This article explores the interrelationship between global production networks (GPNs) and free trade agreements (FTAs) in the South Korean auto industry and its employment relations. It focuses on the production network of the Hyundai Motor Group (HMG) — the third biggest automobile manufacturer in the world — and the FTA between the EU and South Korea. This was the first of the EU’s ‘new generation’ FTAs, which among other things contained provisions designed to protect and promote labour standards. The article’s argument is twofold. First, thatHMG’s production network andKorea’s political economy (of which HMG is a crucial part) limited the possibilities for the FTA’s labour provisions to take effect. Second, that the commercial provisions in this same FTA simultaneously eroded HMG’s domestic market and corporate profitability, leading to adverse consequences for auto workers in the more insecure and low-paid jobs. Inmaking this argument, the article advances amultiscalar conceptualization of the labour regime as an analytical intermediary between GPNs and FTAs. It also provides one of the first empirical studies of the EU–South Korea FTA in terms of employment relations, drawing on 105 interviews with trade unions, employer associations, automobile companies and state officials across both parties.


Introduction
A major driver and outcome of trade-based integration in the global economy has been the enhanced capacity of capital to organize highly calibrated Liam Campling and  networks of production across various jurisdictions. Trade-based integration refers to the liberalization of import tariffs and quotas, the standardization of products and practices across national regulatory systems and the constitutionalization of political processes governing economic activity. This has functioned to integrate economic space on the basis of broadly consistent rules, contributing to the creation, stabilization or intensification of capital accumulation (Azmeh 2015;Bair 2006;Campling 2016;Curran et al. 2019;Smith 2015a). Researchers deploying global production network (GPN) and global value chain frameworks have drawn attention to the ways in which this cross-border accumulation has been coordinated by transnational capital, in particular those organizations dubbed 'lead firms'. This coordination has been premised on putting suppliers in sharper competition with one another such that cost-cutting cascades through the network or chain of firms, directly and indirectly squeezing workers and the social reproduction of labouring classes (Mezzadri 2016;Pattenden 2016;Pickles and Smith 2016;Selwyn 2015).
Societal responses to this phenomenon, led by social movements and trade unions, have included a series of campaigns against consumer-facing lead firms, with brand-name companies in apparel, food and electronics among the most prominent targets. This has led to new forms of nonstate labour governance, such as internal codes of conduct and external certification schemes against which supply chain practices have been audited (see Locke 2013;Jackson et al. 2018). Among the state-regulatory responses has been a revival of policy interest in the 'trade-labour linkage'. Following the failed attempt to include a social clause in the formation of the World Trade Organization in the 1990s, a linkage to labour standards has reappeared in a large number of free trade agreements (FTAs). One-third of all trade agreements now contain labour provisions, with the largest proportion accounted for by the EU (ILO 2016: 22). Compared to the extensive literature assessing non-state or 'private' governance, research on the internationalization of state-based labour governance remains relatively thin.
The 2011 EU-South Korea FTA (EUKOR) was the EU's first 'new generation' trade agreement; so-called because it contained a range of provisions that went far beyond what had been agreed -and what could plausibly be included -at the multilateral level (see European Commission 2019b). Among these were provisions on labour and environmental standards, brought together in a Trade and Sustainable Development (TSD) chapter of the agreement. TSD chapters have since become a standard fixture in EU trade policy, featuring in FTAs signed with 17 other states as of 2018, including those with major national economies such as Canada and Japan. Research that has sought to do more than identify changes in labour law -a common approach in analysing trade-labour linkages -is now emerging. This has traced the impacts of TSD chapters on labour governance and employment relations in EU trade partners, yet focused on the labour provisions somewhat in isolation from the rest of the agreement (Marx et al. 2016;Orbie et al. 2017;Harrison et al. 2019). By bracketing out the wider ramifications of FTAs for workers, this approach risks siloing the study of trade-related labour issues to a discreet set of provisions.
As noted above, GPN research takes an inter-firm approach to the study of employment relations, highlighting the differentiated consequences for wages, working conditions and collective action in lead firms and suppliers at various tiers (see Jenkins and Blyton 2017;Lakhani et al. 2013;Newsome et al. 2015;Reinecke et al. 2018). In this regard, it has done much to illuminate the effects of the globalization of production on working lives and the changing nature of labour governance, both at the firm-level and within the domestic institutional arrangements in which GPNs are embedded. Yet it has largely downplayed the influence of international trade agreements on work, despite the fact that these inter-state agreements have provided the very opportunity for GPNs to proliferate (for exceptions, see Anner et al. 2013;Tran et al. 2017;Smith et al. 2018). This is especially problematic given the increasing incorporation of labour provisions in FTAs, which have received scant attention to date.
This article seeks to bring together these parallel literatures to take into account the interrelationship between FTAs and GPNs. Its central argument is twofold. First, that Hyundai Motor Group's (HMG) production networkand the Korean political economy of which it is a central part -limited the possibilities for the FTA's labour provisions to take effect. Second, that the commercial provisions of this same FTA shifted the competitive conditions in the Korean auto market and contributed to eroding HMG profits, with adverse impacts on workers in the more insecure and low-paid jobs, especially those located in the lower tiers of the production network. In making this argument, the article advances a multi-scalar conceptualization of the 'labour regime' as an analytical intermediary between GPNs and FTAs, allowing their interrelated impacts on employment relations to be better understood (see also Pattenden 2016;Smith et al. 2018). Alongside this analytical contribution, the article also provides one of the first empirical studies of the effects of EUKOR on employment relations; an agreement which remains politically significant as the precedent and public exemplar of the EU's TSD chapters and supposed commitment 'to ensuring that trade policy is also about promoting values such as the protection of human rights [and] labour rights' (European Commission 2019a: 8). Our case study on the automotive industry is especially salient as it was an area of major concern to both sides in the negotiation and implementation of EUKOR. This was because of the commercial significance of the trade in automobiles and auto parts, as well as the high-profile struggles between the militant labour movement and politically powerful conglomerates (chaebols) in Korea that would serve as a formidable test for the EU's stated trade policy intentions.
The rest of the article proceeds in the following sections. Section 2 sets out our approach to understand the articulations of trade-based integration, GPNs and their employment relations consequences through the analytical prism of the labour regime. Section 3 provides an overview of the automotive production network and labour regime in South Korea with particular emphasis on network governance, the distribution of value among firms and workers and employment relations dynamics. This sets the context within which the trade agreement's labour provisions would be operationalized. Section 4 shows how the EUKOR provisions around automotive trade and labour standards were negotiated and what their implications have been for the Korean automotive labour regime. Section 5 concludes.

Labour regimes, GPNs and trade-based integration
Labour regimes can be defined as 'the interrelation of (segmented) labour markets and recruitment, condition of employment and labour processes and forms of enterprise authority and control, when they coalesce in sociologically well-defined clusters with their own discernible "logic" and "effects"' (Bernstein 2007: 7). Heeding Thompson and Smith's (2009) call for labour process theory to move beyond distinct workplaces, the approach we adopt here is one that recognizes the variegated scales at which a labour regime is constituted. Distinct formations have been explored at the local scale (e.g. Burawoy 1985;Jonas 1996;Smith and Pun 2006), the national (e.g. Anner 2015), the macro-regional  and in the relationship between them (Baglioni 2018;Taylor and Rioux 2018). As Pattenden (2016) argues, the labour regime is a useful mediating category between the day-to-day labour processes of a particular workplace with its diverse 'forms of exploitation' and the more abstract 'general forms of domination' under capitalism. It is this multi-scalar quality that is particularly helpful for bringing GPNs and FTAs together, since the former governs labour at the level of the firm according to a transnational spatial formation, and the latter at the level of the state according to an international spatial formation.
Our starting point is thus to conceptualize a labour regime as composed of nested scales, institutionalized formally in organizations with codified rules but also informally in relational norms and habits ; see also Baglioni 2018). The first scale is the workplace, which includes the labour process and its 'dynamics of control, consent, and resistance' and at which point the organization of labour in the production process provides the basis for the creation and appropriation of surplus value (Thompson and Smith 2009; see also Cumbers et al. 2008). This is the scale at which popular campaigns tend to politicize working conditions and scrutinize the effectiveness of trade-related labour provisions, as seen in the response to the Rana Plaza collapse that implicated a number of EU-based clothing companies in the deaths of garment factory workers in Bangladesh. The second scale is the national, comprising the state regulation of labour standards and the industrial relations of nationally-organized labour and capital. This is, in theory, where trade-based integration ought to have most effect on employment relations, especially where states are required to adopt or follow certain rules around labour standards as a matter of law. The third scale is the GPNs through which the configuration of power relationsnetwork governance -among suppliers and lead firms is deployed. This allows for the recognition of distinct forms of exploitation, including those shaped by varieties of subcontracting relationships. This practice can subvert labour law and reduce or even avoid formal worker negotiation (Kelly 2001;Mezzadri 2016), as happened in Korea as we show below in the distinction between 'regular' and 'irregular' workers; the latter category referring to those on shorter contracts with fewer benefits and greater insecurity. Trade-based integration can also play a role at this scale in reconfiguring relationships among firms and the coordination of GPNs. For example, trade agreements can be used to spread the adoption of a particular country's product safety standards, which gives its home firms commercial benefits in terms of scale economies and the ability to procure from a wider range of suppliers; something again that occurred in the case we consider (Campling 2008;Smith 2015b). The ways in which these three scales interact to shape employment relations is an empirical question. For example, relations at the workplace scale cannot be read off from those at the national scale (and vice versa) because of the relative autonomy that actors have at different scales. . It is for this reason that our methodology relied on an extensive range of interviews with representatives located in a relevant position of authority within our nested scales. These were asked to point to the particular pressures affecting labour governance and employment relations in their area of responsibility and to reflect on the extent to which these were tied up with the EUKOR agreement. In total 105 interviews were conducted as set out in Table 1. Of these 67 were in South Korea during January 2016 and March 2017 with individuals representing trade unions, employer associations, auto firms and the Korean government. The remainder were in the EU, mainly Brussels, during May/June 2016 and March 2017 European Commission officials and representatives of industry associations and trade unions. To protect their identity, interviewees are referenced anonymously according to an alphanumeric system (K23, e.g., E12). The letter represents the location of the interview -K for South Korea, E for EU -and the number a unique interviewee. Where a dash separates two or more individual numbers (e.g. K61-K62), it indicates that they were interviewed together. In order to minimize the dangers of interviewee bias affecting key conclusions, interviews were triangulated across multiple interviewee types and/or by verifying with documentary evidence, including Korean-language documents translated on behalf of the authors. Any significant differences in opinions between key informants are explicitly identified.

Hyundai's GPN and Korea's automotive labour regime
In this section, we map the GPN of the HMG and examine its differential consequences for employment relations across the varied firms in the sector, that is, the interaction between the workplace and global scales of Korea's automotive labour regime. It is onto this map that we subsequently detail the consequences of the trade-based integration of EUKOR in Section 4.
The automobile industry constitutes around 7 per cent of GDP and may account for up to 15 per cent of manufacturing employment in South Korea (Song and White 2018). 1 Almost 65 per cent of Korea's auto production is exported making the industry sensitive to shifts in trading dynamics such as those as generated by FTAs. Over 75 per cent of all finished production is accounted for by HMG, which includes Hyundai Motor Company and Kia Motors, acquired in 1998(KAMA 2014. As shown in Figure 1, HMG coordinates a domestic production network of around 400 first-tier and 2,000 to 3,000 second-tier suppliers. Many of the latter also supply the other lead firm manufacturers in Korea, which, in declining order of production units, are General Motors, Renault-Samsung and SsangYong Motors (KAMA 2014(KAMA , 2017. HMG's production network is commercially 'captive' in the sense that its suppliers are transactionally dependent on HMG (Kim 2015). HMG dominates South Korean auto manufacturing, including through its outsourcing subsidiary Hyundai Mobis. By virtue of its unparalleled dominance as the system integrator, HMG coordinates and controls relations in the production network to capture surplus value and outsource cost and risk to first-and second-tier suppliers (K37-39, K54, K61-62). HMG's strategy of squeezing suppliers appears to have intensified: between 2006 and 2014 the average cost of a finished car increased by 34 per cent while the cost of parts increased by only 2 per cent (PSPD and KMWU 2016: 44). This is reflected in the stratification of profit rates in the supplier network ( Figure 1). Smaller profit rates in the lower tiers have translated into poorer wages, worse working conditions and reduced capacities of labour to organize (K37-39, K48, K49, K50-51, K55, K56). Companies in these tiers 'are trying to increase irregular workers and outsource to address pressures of competition' (K53). Migrant workers from Southeast Asia and Mongolia in the second tier are often doubly disadvantaged because of their migrant status, especially where they are not registered or are on short-stay working visas (K61-62).
Within HMG, employment relations have been impacted by two managerial strategies: the segmentation of the workforce and 'circular shareholding'. Segmentation can be traced back to the auto workers who were among the vanguard of South Korea's democratization movement in the 1980s after decades of dictatorship, justified and enabled by the Cold War context in Northeast Asia (Kwon and O'Donnell 1999;Eder 1997;Glassman and Choi 2014). Auto workers' self-confidence and distrust of management translated into militant trade unionism, which produced a number of workplace victories in negotiating wide-ranging employment gains through much of the 1990s (Koo 2001;Lansbury et al. 2006). In response to the 1997 economic crisis, the Korean state and the IMF began to erode these gains through labour law reform, including the encouragement of subcontracting, albeit for a maximum of two years O'Donnell 1999, Lee andLee 2003;K53). After years of dealing with militant unionism, Hyundai executives used the opportunity to quickly employ a greater number of subcontracted 'irregular' workers (K61-K62). Regular and irregular workers were played off against each other through the 2000s and into the 2010s (Lee and Frenkel 2004;Yun 2016). HMG employed a greater number of younger, precarious, irregular workers at lower pay (at Hyundai Motors monthly wages were 60-70 per cent of regular workers) and with far reduced working conditions (e.g. no paid holidays, supervised toilet breaks), often to work on the same production line as regular workers (Yun 2016;K54, K55). This in turn served to discipline regular workers through the threat of wage cuts and job insecurity (Lee and Frenkel 2004;Kim 2015;K53). Circular shareholding is the practice used by many family-controlled conglomerates in South Korea (chaebols) to transfer value between legally separate publicly limited entities to smaller, family-controlled ones in ways that mask financial performance and facilitate continued familial control, including succession. It is widely recognized that HMG subsidiaries such as Glovis and Hyundai Mobis (see Figure 1) use intra-group transactions to redistribute value away from suppliers and control away from external shareholders (K37-39, K41, K45-46, K52, K54; Kim and Jin 2017;KMWU 2017;PSPD and KMWU 2016). For example, Mobis is the entity where Chung Mong-Koo (HMG's CEO) has his second largest shareholding and is thought to be the de facto controlling company of HMG (K35, K37-39, K47; Jung 2017). Mobis was formed in 2000 to coordinate and produce module systems and parts supply for after-service; emerging to be the world's number seven auto parts supplier in 2017 (K54, K57-59; Wad 2008; Automotive News 2018). One explanation for the rise of Mobis is its use of its strategic position in the HMG production network of concentrating parts supply to capture a portion of value from suppliers 'like a commission' (K35). For example, its annual profit rate did not drop below 8 per cent between 2008 and 2015, even during 2008-09 when the global economic crisis saw losses in two competing parts suppliers, Bosch and Denso (KMWU 2017: 13). Mobis's profit rate is consistently 50 per cent greater than the Korean average for first-tier auto parts suppliers. A similar dynamic is apparent for HMG 'specially related companies' that provide marketing and IT services within the group; again Chung Mong-Koo's share ownership is proportionately much higher in these entities and they had profit rates above the HMG average (KMWU 2017). One implication of circular shareholding for employment relations is that profit is captured from the production network and paid out as executive wealth rather than being redistributed in the form of wages, especially among lower tier workers, and/or R&D, thus potentially undermining future security of employment. For example, Hyundai Motor Company's spending on R&D between 2011 and 2015 was around half that of German rival Volkswagen (PSPD and KMWU 2016: 45).
Formal labour representation in Korea's auto production network is uneven, with ß70-80 per cent unionization in branded-manufacturers and in first-tier auto parts and module firms, but only ß20-30 per cent in secondtier suppliers (K47). This still contrasts positively with a national union density of around 10 per cent (K5, K22, K23), which halved from over 20 per cent in the late 1980s (Koo 2001). Two competing unions organize auto workers across these thousands of firms. The Korean Metal Workers' Union (KMWU) predominates in HMG and other lead firms' workplaces and is formally organized as an industrial union. The Federation of Korean Metal Workers' Trade Unions (FKMTU) tends to be more active in supplier firms and organizes around branches at individual factories. 2 This seemingly corresponds to the conventional characterization of labour organizing in Korea, with 'solidarity-oriented industrial unionism' on the one hand and 'micro-corporatistic enterprise unionism' on the other (Lee 1998). However, in practice the distinction blurs. Even though KMWU is organized at the national level as an industrial union, the only collective bargaining agreement it negotiates across the industry as whole concerns the minimum wage (K47; K50-51). As such, the majority of industrial relations activity in Korea's automotive labour regime is characterized by 'an enterprise agreement legacy, which is a cultural history . . . the core contents on working conditions and labour standards is in enterprise agreements' (K47; also, K36; K47; K49; K50-51). The long-standing culture of enterprise unionism is reproduced by the fact that workers in leading factories in the production network, especially those owned by HMG, are able to negotiate better pay and conditions at the enterprise level (K50-51; K54; K61-62; Lee 2011a). The leaderships of KMWU and the Korean Confederation of Trade Unions (KCTU), of which it is a leading member, are clear on this tension and are actively building solidarity (especially with irregular workers), engaging in ongoing campaigns to increase regularization, and developing social movement unionism (K5, K26-28, K36, K38, K41, K47, K50-51; Fleckenstein and Lee 2019). Efforts to promulgate industrial unionism are further undermined at the scale of the state through the design of national labour law (Lee 2011a) and non-ratification of ILO conventions 87 and 98 on Freedom of Association and on Collective Bargaining.
Enterprise unionism is actively nurtured by HMG as a managerial strategy to encourage competition among workers and undermine industrial unionism. HMG management has widely suppressed organizers among irregular workers and discouraged union membership among the rank-andfile Doucette and Kang 2018). It also refuses to negotiate with the KMWU on an industrial scale, insisting that this must be done by 'independent' entities. HMG itself uses an internal division at the Group level -the Labour Policy Unit -to coordinate its employment policy and union negotiations across the Kia-Hyundai separation (K61-K62). HMG representatives were certainly very sensitive about labour, suddenly pushing an interview to close when the topic of 'human resources' was brought up (K45-K46). As another interviewee put it, 'The power of HMG is so great that it drags trade unions here and there' (K49; also, K54).
Wages paid to 'regular' workers employed by HMG are the highest across the entire industrial sector in Korea, and their salaries are estimated at two to three times the national average (K35). First-tier and second-tier workers are paid at, respectively, ß60-70 per cent and ß30-40 per cent of HMG wages (K37-39; K47; K54; K61-62; Cho 2006; Lee 2011a), leading some to argue that HMG workers gain at the direct expense of others in the production network (K35; K47; K54; K55; K66). Others argue that KMWU's regional branches -especially the Hyundai Motor Company's militant Ulsan branch, the world's largest car factory -are at the forefront of improvements in employment relations across the sector, setting the pattern for other KMWU branches and for FKMTU (K36; K50-51; K51; K54; K55; K57-59; K61-62). From this perspective, one of the reasons for the lower pay further down the supply chain is said to be the difficulty of labour organizing in smaller firms, where 'in many cases there are personal relations between owners and workers' (K47; also K55). The paradox of KMWU is that its members can be interpreted as both 'labour aristocracy' and 'vanguard'. KMWU is not entirely a 'pattern setter' as pay and conditions stem from each 'company's performance and are negotiated from there' (K53; also, K47; K57-59; K67), yet value is unevenly distributed through the production network and HMG's regular workers stridently defend their branch-level gains, which may be reduced if an industry-wide agreement were reached (K47; K51; K54; K55). At the same time, KWMU successes have transformed employment relations in a significant proportion of the wider automotive labour regime. These include an agreement with HMG in the early 2010s to regularize 6,000 workers by the end of 2017 and, perhaps most significantly, a reduction in working hours (K47, K54, K61-62).
Despite the context of HMG's divide-and-rule approach to employment relations and successive pro-business governments eroding labour standards and repressing worker resistance, most notably the Park Geun-hye administration's (2013-2017) attempt at sweeping labour law reforms and imprisonment of KCTU leaders (see below), KMWU's Hyundai Motors Ulsan branch negotiated a major change in the shift structure at the enterprise level which transformed shift patterns through much of Korea's automotive labour regime. Starting at two 12-hour shifts, Hyundai Motor Company transitioned to two ten-hour shifts by 2013 (ending late-night work) and to two eight-hour shifts by 2015. 3 This gain was won at the same pay, albeit with some intensification of the labour process via a 2-3 second speed up of the production line to 64 seconds per task (K37-39, K47, K51, K60, K61-62; IndustriALL 2013). The requirements of just-in-time production (i.e. lowfactory floor inventory, where parts and modules are supplied accordingly to highly calibrated time slots) meant that first-tier suppliers were compelled to mimic Hyundai Motor Company's new shift patterns, quickly diffusing a reduced working day through the production network and demonstrating one way in which the workplace and GPN scales interact (K37-39, K53, K55, K56, K57-59).

The EU-South Korea Free Trade Agreement and Korea's automotive labour regime
Having characterized the Korean automotive labour regime in terms of the workplace and GPN, we now bring in the national scale. As noted above, this scale serves as the best starting point to consider how the tradebased integration (liberalization, standardization and constitutionalization) augured by the EUKOR agreement has affected Korea's automotive labour regime. In this section, first we explore the negotiation of EUKOR to explain how the particular provisions governing automotive trade and labour standards came about and what they were intended to do. Then we examine the institutional form and implementation of the labour provisions within EUKOR and why they have been unable to gain much traction within the automotive labour regime. Finally, we take this a step further to examine the adverse effects of the commercial provisions within EUKOR on HMG and, in turn, the workers in its production network. As set out above, EUKOR has been repeatedly cited by the European Commission as a point of departure for its 'new generation' of FTAs and commitment to labour standards as set out in the TSD chapters; a claim acknowledged, albeit frequently challenged, by many interviewees in the EU (E10, E14, E16, E20, E21, E26). Within EUKOR, the TSD chapter contains both labour standards commitments and institutional mechanisms for their implementation, summarized in Figure 2. This became the model adopted and applied, with some variation, in all subsequent EU FTAs (Harrison et al. 2019). The substantive commitments refer to the eight ILO core conventions relating to freedom of association, non-discrimination at work, child labour and forced labour. The institutional mechanisms consist of an inter-governmental Committee on TSD, which is reported to by two Domestic Advisory Groups (DAGs) that include 'civil society' representation of economic, social and environmental interest groups in the EU and Korea.  Source: Cooper et al. (2011: 7, 9); EUKOR (2011).
Should a dispute emerge, and a satisfactory resolution is not forthcoming in consultations, either party can request independent adjudication in the form of a Panel of Experts. The focus of the EU's approach is thus one of aligning with the laws and norms of the ILO, promoting inter-state dialogue towards this end, and providing 'voice' to civil society actors to inform this process.

Negotiating Labour Governance and Automotive Trade in EUKOR
Respective South Korean administrations have consistently promoted FTAs since 1997. For example, recognizing Korea's heavy trade-dependence, the Lee Myung-bak administration (2008-2013) and his pro-big business Grand National Party (now Saenuri Party) saw FTAs as the centrepiece in an ongoing export promotion strategy of achieving duty-free market access for 70 per cent of national exports, and attracting foreign investment (Office of the President 2009: 29). EUKOR was in part prioritized around liberalizing trade in the automotive sector, Korea's second largest export sector following electronics. The Korean automotive industry, dominated by HMG, was enthusiastic about EUKOR because it was assumed that it would enhance EU market access (K29, K30; Siles-Brügge 2012). In contrast, the European automobile manufacturers' association (ACEA) and the German auto industry initially opposed the negotiation of liberalization of the auto trade under EUKOR because it considered Korean imports as a threat and the restrictions imposed by Korea's 'non-tariff barriers' as a major obstacle (ACEA 2010; E33). It also saw that Korea was reluctant to accept the EU approach to measuring vehicle CO 2 emissions; somewhat ironic given that Volkswagen would later be fined by environmental regulators the world over, including in Korea, for cheating in emissions testing (European Parliament 2010: 12). The Italian industry too was opposed, principally due to concerns that HMG would compete directly with Fiat in the EU market for small cars; as were Czechia, Poland and Slovakia because of the threat to the competitiveness of foreign auto production investments in these countries (Dalton 2009). As shown in Table 2, the longer-phase out of tariffs on the import of smaller cars from Korea was an attempt to mitigate these concerns, but HMG's setting up of factories in Central Europe in 2007 and 2008 that specialize in smaller cars made this concession a moot point (E38, K35, K44, K45-46, K49; Pavlínek andŽenak 2011).
The non-tariff barriers identified in automotive standards were central to the negotiations, which was of particular significance given parallel FTA negotiations with the United States which was promoting its auto standards over EU ones (Cooper et al. 2011;Platzer 2010). The result was an annex on Motor Vehicles and Parts that bound South Korea to recognize the World Forum for Harmonisation of Vehicle Regulations as 'the relevant international standard-setting body' and to harmonize its standards accordingly (EUKOR 2011: Annex 2-C, Articles 2 and 3). While the Annex emphasizes 'regulatory convergence', the European Commission (2009) itself pointed out that 'Korea will essentially recognize as equivalent many European standards'. Importantly for our purposes, the Annex established a permanent Working Group on Motor Vehicles and Parts to meet annually to monitor implementation and for the purposes of consultation, much like the TSD chapter. But unlike the legally siloed Panel of Experts 'dispute' mechanism in the TSD chapter, automotive standards were covered by the 'hard law' dispute settlement process that regulates the implementation of EUKOR as a whole, but with modifications that heighten the urgency of any dispute and speed up the arbitration process (Annex 2-C, Article 10). Put plainly, the constitutional dimension of EUKOR explicitly placed the regulation of auto standards well over those of labour.
The commercial interests at stake in the negotiations over automobiles and other high-value sectors such as machinery and appliances played a major role in mitigating the EU's ambition for the TSD chapter in the FTA with South Korea. The notion of a TSD chapter and the labour provisions contained within them were not seen as a natural part of trade agreements by the Korean side; they were proposed and driven through by EU negotiators (K23, K25, K26). The relative size of the Korean market meant that negotiations were less asymmetrical than other negotiations the EU was then conducting -EUKOR was touted as the second largest FTA in history at the time, following the North American Free Trade Agreement (Lee 2010). This made it more difficult for the EU to impose a trade-labour linkage in EUKOR negotiations as it had done when insisting on the inclusion of core labour standards in its 2008 FTA with Caribbean countries (Harrison et al. 2019). As mentioned above, the EU's rapid advancing of FTA negotiations with Korea was spurred by parallel FTA negotiations between Korea and the United States, with both seeking a first mover advantage (Cooper et al. 2011;Platzer 2010). Negotiators from the European Commission had included a TSD chapter primarily as a result of pressure from the European Parliament, supported by organized civil society groups and European-level trade unions, and as such saw the inclusion of labour standards as a narrow textual requirement and thus did not pursue an aggressive pro-labour agenda (E7, E9, E12, E17; Barbu et al. 2018).
These factors contributed to enabling Korean negotiators to successfully demand fewer references to international labour standards and the removal of any immediate obligation to ratify all fundamental ILO conventions (K25). Korea has ratified only four of the eight ILO core conventions, and not the four relating to freedom of association and forced labour. Thus, while EUKOR contains a commitment by the parties to 'respecting, promoting and realizing in law and practice' the eight fundamental ILO conventions, in contrast to all of the TSD chapters negotiated since, EUKOR contains a rider that: The Parties reaffirm the commitment to effectively implementing the ILO conventions that Korea and the Member States of the European Union have ratified, respectively. The Parties will make continued and sustained efforts towards ratifying the fundamental ILO conventions as well as the other conventions that are classified as 'up-to-date' by the ILO. (EUKOR 2011: ch. 13, Art. 13.4, para. 3, added emphasis)

EUKOR's TSD Chapter and National Labour Governance
We turn our attention next to the question of whether the labour provisions within the TSD chapter had any discernible effects on Korea's automotive labour regime. In this respect, early efforts following the launch of EUKOR focused on Korea's failure to make progress on ratifying ILO fundamental conventions, especially relating to freedom of association. After European and Korean trade unions had highlighted this issue in several joint meetings between DAGs, the EU DAG eventually sent a letter to the Commission requesting that formal consultations be initiated -a precursor to convening the Panel of Experts. The basis of the EU DAG letter was that insufficient efforts had been made towards ratifying key conventions and at the same time, widespread violations of labour rights, particularly freedom of association, were allegedly taking place, including in the auto industry (Jenkins 2014). But the Trade Commissioner at the time, Karel De Gucht, rejected the request and instead promised to pursue the matter through inter-state dialogue (Vogt 2015). Perhaps reflecting the shift in policy emphasis of the Commissioner since 2014 with Cecilia Malmström, the Commission did start to make more public statements subsequent to this letter. These included highlighting 'serious concerns in particular in the area of trade and sustainable development, i.e. on insufficient progress on the ratification and implementation of ILO conventions and on protection of labour rights in Korea' (European Commission 2017: 11; see also European Commission 2016: 12).
Trade union and other civil society representatives in Europe, however, were sceptical that inter-state dialogue would produce tangible results for labour standards in Korea (E15, E19, E26, E29). These concerns are reinforced by the failure of European Commission officials to raise concerns about the imprisonment of leading figures in the KCTU, despite their imprisonment being widely condemned by relevant international bodies (ITUC 2016;FIDH 2018). The largest affiliate of the KCTU is the KMWU and as such auto workers play a major role in financing and driving politically KCTU's activities (K5, K49, K61-62). European Commission officials emphasize the difficulty of the labour situation in South Korea and that action taken under the TSD chapter should not frustrate the overall objectives of the agreement (E10, K24). One Commission official told us that: It is important to have a positive forward looking agenda. Confrontation would lead to a backlash on behalf of Korea. We want to add investment protection into the agreement. If we took action under this chapter, we might lose benefits elsewhere. So we do need to think about the bigger context (E14).
This impasse appeared to have been removed with the impeachment of President Park Geun-hye on corruption charges in March 2017 and subsequent sentence to 25 years in prison. Presidential elections in May 2017 brought Moon Jae-in to power (his Democratic Party already had the largest minority in the National Assembly) and the pro-business, repressive labour reforms advanced by the Saenuri Party appeared shelved. Indeed, Moon's manifesto had committed to both labour reforms and to reform the circular shareholding model of the chaebol. The minimum wage was increased, the working week was capped at 52 hours, and the current and former KCTU presidents released from prison. Further, HMG itself has been targeted in an antitrust investigation by the Fair Trade Commission, which spurred HMG to announce corporate governance reform by spinning off Mobis's module and after-service auto parts provision to Glovis (Jin and Lee 2018). However, by mid-2018, Moon's commitments were being squeezed by an economic slowdown, the chaebols' political power, and the emboldening of the Right within the National Assembly. The KCTU responded with a mass strike in November 2018, including KMWU workers, against Moon's roll-back from chaebol and working hours' reforms (Harris and Song 2018).
In seeming response to the retrenchment of Moon's electoral commitment to reform labour governance, the European Commission sent a letter to the Korean government in December 2018 requesting consultations under the terms of the TSD chapter (EU 2018; also, Malmström 2019). The letter highlights several specific concerns in the Korean Trade Union Act and especially Korea's failure to 'to make continued and sustained efforts towards ratifying the fundamental ILO Conventions' as per EUKOR Article 13.4 (3). In July 2019, the European Commission triggered the Panel of Experts mechanism (European Commission 2019a), which is an institutional first across the network of EU FTAs with TSD chapters. Nevertheless, the effect of such a move remains unclear because, as noted earlier, Panel opinions are only advisory and the TSD has no recourse to EUKOR's dispute settlement process.
Another case demonstrating apparent weaknesses in the TSD chapter involves the exploitation of 'dispatched workers', a category of irregular workers. Under Korean labour law it is illegal to dispatch manufacturing employees from one firm to undertake work for another firm. HMG had sought to get round this by renaming their practice as one of 'in-house' or 'on-site' subcontracting' but was found guilty of dispatching workers to suppliers because supplier firms were effectively controlled by HMG as they have few or no alternative buyers for their product and Hyundai managers were actively involved in supervision and control of these workers' activities (Lee and Chang 2015;K27;K28;K47). The issue of dispatched workers was at no point raised by the DAG members within EUKOR. Thus, even where Korean industry was found to have systematically contravened a domestic law with clear implications for the 'trade-related aspects of labour' (EUKOR 2011, Article 13.2.1, emphasis added), the TSD process was entirely silent.
That issues faced by dispatched workers in the auto industry identified above were not forcefully pursued by trade union and civil society representatives certainly raises questions over whether this institutional mechanism, as currently constituted, is sufficiently representative of all workers' interests, or whether it privileges the positions of formally organized and represented workers. On the European side the focus of trade union and civil society concern, as well as from Members of the European Parliament, has been about the degree of independence from the Korean government of academics and other professional researchers on the Korean Domestic Advisory Group (E6, E17, E30). But there is also a question about the degree to which the two trade union federations -the Federation of Korean Trade Unions (FKTU) and KCTU who are the worker representatives on the DAG -represent the interests of all workers. Among the most important characteristics of the Korean labour market is the segmentation of regular and irregular work, a central issue in Korea's automotive labour regime and the domestic economy as a whole. As highlighted in Section 3, irregular workers receive substantially lower pay and working conditions and face major hurdles in formal self-organization, which tends to include proportionately more women and unrepresented migrant workers (K23, Chang 2009; Lee 2015; Lee and Kang 2012;Kim 2015). While both national trade union federations work hard to represent the interests of irregular workers -including sometimes even against the interests of their own rank-and-file members -predilections still persist (K5, K21-23, K31, K37-39, K47, K49, K61-62, K63).
Various observers and even some union officials point out that the relative emphasis in negotiation practices still favours the interests of regular workers (K49, K55, K56, K61-62, see also Chang 2009). When irregular workers selforganize and take industrial action they are particularly vulnerable to attacks by employers (K36, K37-39), including state-sanctioned practices of civil suits for damages for the obstruction of business which allows for individuals to be held liable (Doucette and Kang 2018). As of March 2015, a total of US$148 million in damage claims had been made in 17 cases, including against three irregular worker branches at Hyundai Motors and several auto parts companies. Individuals have taken their lives under the pressure of damage claims, including two auto workers (Doucette and Kang 2018). Given that nationally only 1-2 per cent of all irregular workers are members of a trade union (K23), it may not be appropriate that the FKTU and KCTU are the only worker representatives on the Korean DAG.
HMG thus appears to have been able to ignore national labour law and a juridical ruling reaffirming it, and these workers were directly involved in producing cars, many of which would have been exported to the EU market. This issue was not formally raised in the context of the TSD chapter, despite this containing an obligation that the parties 'shall not weaken or reduce . . . labour protections afforded in its laws to encourage trade or investment' (EUKOR 2011, Article 13.7). Further, the Park Geun-hye administration sought to push a number of repressive reforms to domestic labour law (K10, K22, K23). This included a proposed law allowing dispatched workers in manufacturing, which was known among some observers as 'Chung Mong-Koo's law', the CEO of HMG, because it would have allowed for the continuation of his firm's existing practices.
In summary, despite the expectations around governing international labour standards in the context of major trade agreements, the institutions of the TSD chapter have been slow in making progress on the serious and contested labour issues which have been raised by DAG representatives.
It remains unclear what material effect the Panel of Experts advisory recommendations will play in driving through improvements given the overall architecture of the TSD chapter and its separation from more legally binding dispute settlement areas of the agreement, which for the latter include more meaningful recourse to trade remedy mechanisms, compensation and even the possibility of suspension of the trade agreement. Equally, other issues of particular importance to the auto industry have not been raised at all via the DAG process, perhaps in part because of questions about the extent to which the DAGs are representative of the most vulnerable workers. Consequently, the key mechanism seeking to regulate international labour standards via trade agreements remains significantly constrained in its ability to intervene in a meaningful manner. In other words, at the national and workplace scale, the automotive labour regime appears to be largely unaffected by the labour provisions of EUKOR. Where EUKOR has changed employment relations, though not for the better, is via the commercial pressures on the production network of HMG, discussed in the following section.

EUKOR, Automotive Trade and the HMG Production Network
Here we argue that the implementation of EUKOR -especially liberalized market access for EU automobile imports and harmonized auto standards -is putting new commercial pressures on HMG, threatening the current configuration of the labour regime and eroding employment relations. The commercial effects of EUKOR on the parties' respective auto markets and industries were the converse of expectations. European industry had predicted a loss to the EU auto industry because of an assumed increase in exports from Korea (see Siles-Brügge 2012). Instead, as shown in Figure   Indeed, these two auto firms constituted 47 per cent of Korea's 2016 imports in number of automobiles from the EU, Japan and the United States combined. The dominance of high-end, large vehicles is reflected by EU exports to Korea being double the price per unit of volume compared to Korean exports to the EU (author calculations using Eurostat 2018).
The EUKOR agreement has arguably been central to this turnaround in the EU-Korea auto-trade surplus. 4 Certainly that was the position of the Korean auto industry, which saw a once relatively protected market suddenly flooded with German cars following the three-year phase out of the 8 per cent tariff on large cars (Table 2; K44, K63-65). Given that large cars are priced more highly, Korea's previous import tariff had a proportionately more protective effect in making European cars more expensive compared to domestically produced ones. European Commission officials mirror this argument by taking credit that their negotiation of EUKOR benefitted EU auto producers (E33). Alongside tariff liberalization has been the importance of standardization in allowing scale economies in auto distribution and post-sales servicing. Before the expansion of Korean imports of German cars, the costs of maintaining sales and service centres were higher, but with the rise in imports coinciding with EUKOR liberalization and the reduced per unit cost of parts with standards harmonization, increased market share meant that overhead costs were reduced and the retail price of German cars and parts were more competitive (K35, K45-46). These scale economies are partly illustrated in an almost doubling in monetary value of auto parts exports from the EU to Korea between 2008 and 2018, from 557 million to 923 million (author calculations using Eurostat 2018). As a whole, this further reduced the consumer costs of running a German car in Korea, thereby contributing to the erosion of HMG's long-term domestic market advantages.
The upshot of these various dynamics is that the implementation of EUKOR has contributed directly to Korea's growing import of large German cars which has put pressure on HMG's profitability with implications for domestic workplace relations. A major issue here, quite simply, is that larger cars are more profitable. While some parts are of a higher quality, many are the same or similar, and most of all the labour process of assembling even 'luxury' cars is largely equivalent, with the result that the cost of labour power per unit is commensurate to a small car. HMG responded to competition by launching in 2017 its Genesis brand with its 'luxurious image [designed] to push our range as a whole' (K45-46), but it is yet to displace the status of German brands in Korea or in export markets (K35, K54, K60, K63-65). Yet profit margins at both Hyundai Motors and Kia dropped continuously since 2014 (KMWU 2017;Bloomberg 2018a,b). The decline in profitability is the result of a number of factors, including executive capture through circular shareholding, declining exports to China and the United States, and the failure to identify market trends in product types, but the influx of competing German products in the high-margin large vehicle market has certainly played a major role (K37-39). The combined Hyundai-Kia share of the Korea auto market dropped from around 80 per cent in the 2000s to around 65 per cent in 2016 (K35, K45-46, K52, K54, K60). While in parallel HMG's global sales increased, the loss of domestic market share is of particular commercial significance because HMG's long-term strategy was to use the relative protection of Korea's trade regime (tariff and non-tariff barriers) to charge a higher price for cars sold domestically compared to export and transplant markets (K29, K30, K41, K44). This strategy of capturing greater profits in Korea allows HMG to cross-subsidize and 'overcome financial losses at the initial stage in foreign markets' (K54, also K41). As a result, Korean industry contested the win-win arguments advanced by the European Commission in relation to EUKOR (K45-46); European industry, in contrast, was said to be 'happy' with the outcome (E33).
HMG's export-oriented production in Korea remains crucial to its business model and by extension, employment relations, as evidenced in the active positions adopted by trade unions. Even though HMG has a GPN of own-factories and captive suppliers in its major markets (e.g. the EU and United States), exports from Korea continue to make-up a major component in its product mix. At the same time, HMG does not import cars into Korea from its overseas factories (e.g. in China) because 'the trade unions will block it -and the government supports this'. This means that unlike competitors such as General Motors, HMG is not able to fully capitalize on its GPN and Korea's network of FTAs in order to plan its 'footprint portfolio' according to pure cost considerations (K63-65, also Reed 2012). Further, KMWU branches actively negotiate with HMG to produce domestically cars that command higher margins (e.g. larger cars and the Genesis brand) and those using new 'eco-friendly' technologies such as hybrid, electronic and hydrogen cars (K57-58, K61-62). This effectively blocks HMG from producing higher-margin cars in its overseas factories. An analysis by KMWU confirms this dynamic by suggesting that the profitability of HMG's overseas plants is low and that most profit comes from Korean plants (K41, also E36-37, K61-62).
It is important to emphasize that KMWU auto workers actively shape the contours of the labour regime, even in so far as demanding high-profit production in Korea because they know that this gives them greater leverage to extract higher wages and better working conditions. This strategy stems in large part from the high level of profitability of HMG's domestic production and Korean auto workers' effective articulation of demands around the ongoing centrality of Korea in HMG's GPN. Yet, the main long-term effect that EUKOR is having is to erode HMG's profitability with the rapid rise in imported European cars, which represents an erosion of the elemental basis for high salaries and good working conditions for HMG's workers. Given the importance of HMG workers to shaping employment relations in Korea's automotive labour regime as a whole, such an outcome does not bode well for sector-wide working conditions. Indeed, the Korea auto industry's negative view of the effects of EUKOR was being borne-out by late 2018 in a growing crisis in Korea's automotive industry. GM Korea closed one of its plants in 2018 due to poor sales (including domestically), and auto parts firms were to receive government supports of over US$ 3.1 billion (Choi and Jin 2018;Song and White 2018).

Conclusions
It is well established that both the GPNs of transnational capital and the international trade agreements of states matter for employment relations, but there is less research showing how they interrelate. Our contention in this article has been that these two literatures need to be brought into closer conversation, particularly in relation to the labour standards provisions that have proliferated within the FTAs constituting the leading edge of trade-based integration. Among these is the 2011 EU-South Korea FTA, noteworthy as the EU's first 'new generation' FTA that proposed to protect and promote core labour standards and ensure that international trade would not lead to erosion in domestic labour law -a commonly cited reason why globalization supposedly leads to a 'race to the bottom' in labour rights and working conditions.
Our theoretical contribution has been that a critical analysis of such FTAs must consider how they articulate with sector-specific labour regimes, conceptualized here at the scales of the workplace, the nation-state and the GPN. In our particular case on the Korean auto industry, we have emphasized the relational processes that exist between these scales, showing, for instance, how Korean automotive workers sought to influence the patterns of HMG production and export, even going so far as demanding the domestic production of high-profit vehicles because they know that this gives them greater leverage to extract higher wages and better working conditions. But by the same token, HMG has organized its production network in such a way as to create and sustain differential outcomes in employment relations, such that significant inequalities exist within and across the workplaces of the Korean automotive industry between regular and irregular workers.
In terms of the impact of the EUKOR at the scale of the Korean nationstate, we have shown that while the institutions of the TSD chapter have only recently (and after years of sustained pressure from trade union groups) found ways of making progress on the serious and contested labour issues which have been raised by DAG representatives such that the Panel of Experts process has been convened, the non-binding nature of any recommendations constrains meaningful change. Meanwhile other issues that are particularly important in the auto industry have not been raised at all, perhaps in part because DAGs are not representative of all workers. Consequently, there has been no discernible change in labour standards in the automotive sector as a result of the TSD mechanisms. Indeed, entire segments of the workforce do not have a voice in the TSD process, despite being among the most highly unionized and militant sectors in Korea. This indicates an important mismatch in the design of labour standards in EU FTAs which tend to privilege formally organized workers in strategic positions of production networks over those in more precarious positions, and thus in most potential need of the protections that the TSD chapter purports to offer. Finally, we moved beyond the conventional approach to studying labour provisions in FTAs by considering them alongside the commercial provisions. Here, we argued that the tariff liberalization and standards harmonization under EUKOR is both directly and indirectly eroding the profitability of major employers in Korea's automotive industry by shifting the competitive conditions in the Korean auto market, with adverse employment relations consequences for workers, especially in the more insecure and low-paid jobs. In these ways, Korea's automotive labour regime shaped aspects of the negotiation and implementation of the FTA; and over time, the FTA has materially shaped the labour regime. Lead auto firm and first-tier suppliers' employees alone make up 9 per cent of total manufacturing employment, at 125,754 and293,146, respectively, in 2016 (KAMA 2017: 4, 21). Korea's Ministry of Commerce reportedly puts the figure at 7 per cent of employment (Song and White 2018) but based on the KAMA data this is a gross underestimate. 2. It is estimated that around 90 per cent of auto workers in Korea are male (K47), but some HMG factories employ a higher proportion of women (e.g. 35 per cent in one). In this latter factory, workers are members of FKMTU at enterprise level and representatives claimed to have 'a high level of maternity protection in the factory', an outcome of 'gradual change over last 10 years' that is 'ahead of the law', with the effect that the resignation rate among women is under 1 per cent (K57-59). This effect was also supported by the end of night shifts (see below). 3. This constituted a reduction in hours worked from between 2,500 and 2,700 hours a year to around 2,000 hours, although one interviewee pointed out the prior peak of 2,700 hours was itself a violation of Korea labour law (K56). In comparative context, South Korea is third only to Mexico and Costa Rica within the OECD in terms of the longest number of hours actually worked per worker between 2008 and 2017, with Korea being 15 per cent above the 2016 OECD annual average of 1,765 hours (OECD 2018b). 4. In contrast, representatives of EU industry emphasized that the trend was already an upward one from 2009 because of consumer preference -'Korean's like our cars' -and the sharp depreciation of the euro to the Korean won from a 2009 peak (E33, E36-379).