Subcontracting and Labour Standards: Reassessing the Potential of International Framework Agreements

Attempts to regulate labour standards in multinational companies face clear difficulties, not least because companies themselves may not have the executive power to enforce terms throughout complex and fragmented subcontracting structures. In the case of international framework agreements (IFAs), this might suggest a fundamental weakness. Taking our example from the South African construction industry, this article presents an IFA in the context of both employer and union strategy. We demonstrate that a two‐track approach exists: highly interventionist approach to quality‐critical issues compared with labour‐related issues. On this basis, we suggest that, far from being over‐hyped, IFAs have yet to be taken seriously enough.


Introduction
Rather than treating multinational companies (MNCs) only as a threat, over the past decade there has been a growing appreciation among unions internationally that MNCs also afford considerable opportunities, not only in terms of the pay and working conditions of their own employees but also via their impact on labour markets in their host countries. Thus, MNCs have become the focus of debates about labour standards in which the lead company is seen as the linchpin in promoting 'decent work', not only in national bases and subsidiaries but also throughout the commodity/value chain. However, because MNC employment comprises only a small component of the total labour market in developing countries, such an outcome is not automatic.
In the absence of effective public regulation of wage and labour standards, either internationally or, in many cases, nationally, a great deal of attention has been paid in recent years to the possibility of voluntary regulation, that is to say the agreement and implementation of standards within companies that go beyond any external-imposed regulatory requirements that might apply. Such voluntary arrangements include internal company codes of conduct, typically included in corporate social responsibility policies, standards developed in conjunction with NGOs or other third parties ('private' social standards; Riisgaard and Hammer 2011) and also -the focus of this articleagreements resulting from some form of collective bargaining with unions, such as International Framework Agreements (IFAs).
The scope for union intervention in this way -and the ability of an MNC itself to influence the working lives of those who contribute to the production process -is clearly affected by the type of product market within which the company operates and also the way in which the production process is either integrated within one firm or dispersed between many. And therefore, in order to understand the extent to which standards or procedures, such as those embodied in IFAs, are capable of being implemented and enforced, some account needs to be taken of the nature of the relationship between the firms that carry out production and the relative power of the various organizations involved.
Much use has been made recently of the approach to value chain analysis developed by Gereffi et al. (2005) as a way of modelling these dimensions. Gereffi et al. identify five types of value chain governance, which can be seen as forming a continuum between external market relationships and internal bureaucratic control:  (Dore 1983;MacNeil 1985;Powell 1990), this categorization serves to disaggregate the notion of 'hybrid' forms of organization. Crucially for Gereffi et al., the continuum between hierarchy and market is one of power asymmetry, which determines the level of direct managerial control that can be exerted. And this asymmetry can be understood as influenced by specific objective factors. In all five types of relationships, Gereffi et al. see the nature of the linkage as 'associated with predictable combinations of three distinct variables: the complexity of information to be exchanged between value chain tasks; the codifiability of that information; and the capabilities resident in the supply base' (Sturgeon et al. 2008: 307). The structural features of the relationship can be seen to predispose organizations to particular forms of governance. 'Relational' contracts, for example, are characterized by a relatively symmetrical power relationship between parties and, therefore, a need for a high degree of trust and discretion on the part of the contractor. 'Captive' links arise where prohibitive switching costs mean that contractors are effectively dependent on large clients. Typically, governance in these cases is reliant on extensive monitoring and the detailed control of 'less competent suppliers' by the lead firm (Sturgeon et al. 2008: 307).
However, the concept of 'governance' is a broad and rather elusive one, only partially captured by the hierarchy-market continuum. While some degree of simplification is required in order to model the salient features of inter-firm relationships, the world in which unions operate is self-evidently more complex than this. For this reason, most attempts to apply value chain concepts to labour standards have focussed on relatively simple, linear chains. Equally, while the approach is, in principle, capable of moving beyond static, descriptive accounts to tackle both the dynamics of organizational restructuring and the mechanisms of control that make particular arrangements possible (e.g. Flecker and Meil 2010), the application of the concept to labour standards has often treated variants as the outcomes of relatively fixed inter-organizational relationships (Quan 2008). Riisgaard and Hammer (2011) focus on the extent to which value chains are 'driven' by one of the constituent firms, and show how this power to impose conditions within the chain is a precondition for effective union influence. This power, though, has more than one dimension, and the varied forms of contract relationships mean that we cannot assume that the 'judicial' governance function (in our case, for example, the monitoring of standards) coincides with the ability to exercise 'executive' governance (Kaplinsky and Morris 2001: 30). In other words, it may be possible for clients to implement, but not enforce, rules.
Yet there is a danger of overstating the limits to organizational agency. To understand the potential for IFAs, and for union intervention in general, it is necessary to understand something, not only about how the work process is organized but also why it is organized in this way. This amounts to a change of analytical focus, with the central problem being the way in which specific structures result from employer strategy rather than the way in which strategy is determined by these organizational characteristics. This focus makes it possible to explain outcomes in terms of the range of competing priorities faced by employers. Companies' 'make or buy' decisions are influenced by a number of factors, of which the cost of provision is only one. Companies may need to protect against risk of market fluctuation; they may intend to move the cost of particular assets or functions 'off the books'; or they may choose to distance themselves from responsibility for a particular section of the workforce. It is this 'fragmentation strategy' (Shutt and Whittington 1987), rather than its resulting structure, that is our starting point here.
In this way, the disintegration of production presupposes a mechanism -be it via the market or direct managerial control -of reintegrating the overall process. In cases where a product or output can be specified in sufficient detail and monitored sufficiently closely, then it may be that the market power of the client alone is enough to avoid post-contract opportunism (Perrow 2009). However, given that the client needs to control for both cost and quality, relationships are rarely as 'hands-off' as might be suggested by the formal terms of the contract (MacKenzie 2008;Rubery 2007). The dilemma posed by such principal/agent problems has prompted attempts to find 'third way', or network arrangements -arrangements that ostensibly occupy a position between those of the open market or closed bureaucracy (Grimshaw et al. 2005). The characteristics of such a relationship may, conceivably, be facilitated by the contractual terms but are never fully specified. Hunter et al. (1996), for example, see this as an evolutionary process, whereby a low-trust demands model, where outputs are closely specified and economies extracted by competition between contractors, may be developed into a supplier development model, focused on inputs, where the client takes an active part in contractor human resource (HR) management. The problem, though, is intractable because this high-trust 'relational' contracting may also involve high levels of scrutiny (Williams 2008).
For this reason, relationships may include both 'relational' and hierarchic elements. Muehlberger (2007) shows how dependency may be created among nominally independent contractors and how this enables the client to exercise direct managerial control while outsourcing entrepreneurial risk. Nevertheless, long-term 'relational' contracts might be expected to be more likely to facilitate the 'transfer' of labour standards (Fichter et al. 2011: 19). This assumes, of course, that the company sees such a transfer either as desirable or as their responsibility. And because contracting is typically justified in terms of cost reduction, there may be very little room for the parties to exercise discretion (Vincent and Grugulis 2009).

IFAs and union organization
IFAs arise from negotiation between MNCs and global union federations (GUFs) and, inter alia, set out labour rights that are held to apply to subsidiaries and, in some cases, to contractors. The first IFA was signed by the French food multinational, Danone and the GUF for the sector, the International Union of Food, Agricultural, Hotel, Restaurant, Catering, Tobacco and Allied Workers' Associations in 1988. About 80 agreements have been signed to date.
Research on IFAs has tended to focus on the content of the agreements themselves (Hammer 2005;Sobczak 2007), their inception and negotiation (Miller 2004;Schömann et al. 2008a), and the way in which the structure of value chains explains particular outcomes (Riisgaard and Hammer 2011). In fact, the substantive content of IFAs tends to be limited and generic, typically comprising a commitment to abide by the International Labour Organisation (ILO) 'core' labour standards. Since legal enforcement of these standards is, by definition, unlikely in countries that have not ratified the relevant ILO conventions, a clear weakness of IFAs as a strategic approach is that they would appear to work best where they are needed least. As Croucher and Cotton (2009: 68) point out, there is little point in creating agreements that trade unionists cannot use.
Notwithstanding these obvious practical difficulties, IFAs have been seen as a step towards the internationalization of collective bargaining (Bourque 2008;Schömann et al. 2008b;Telljohann et al. 2009). By framing labour rights as universal human rights, IFAs can be seen as a form of 'stateless' regulation (Seidman 2007), potentially capable of regulating supply chains that are 'insufficiently regulated' by national jurisdictions (Sobczak 2007). However, in the absence of supportive national legislation (Niforou 2011), the enforcement of agreements is largely dependent upon the existing strength of local union organization. For an agreement that itself includes the right to union organization and bargaining, this is a serious limitation: a 'chicken and egg' problem (Davies et al. 2011).
Building and Wood Workers International (BWI) has been one of the more active GUFs, having signed 16 IFAs (its first was with IKEA in 1998) (BWI 2011 On the other hand, Fichter et al. (2011: 6) 'define a successful initiation of IFA related practice transfer as the establishment of a viable conflict resolution mechanism'. These are clearly quite different conceptions and suggest two quite different strategic approaches. It can be argued, for example, that it is precisely the absence of organized conflict in the workplace that impedes implementation. IFAs, then, have a potential function in the top-down 'transfer' of practices from head office to local contractor (Fichter et al. 2011), and also as a tool for local organizing. Both functions are significant, but neither is straightforward. While the technical difficulties to be overcome in putting such standards into practice vary according to the form and complexity of the value chain, there is no insurmountable reason why fundamental rights at work cannot be promoted in this way. In the case of safety management, for example, the presumption of main contractor responsibility has been sufficient to prompt a relatively firm response: construction MNCs tend to take safety extremely seriously, with well-developed reporting arrangements, relatively active policing and the option of imposing sanctions on contractors in the case of infringement (Davies et al. 2011;James et al. 2008;van Tulder et al. 2009). The Lafarge case demonstrates the possibilities for labour and employment relations matters to be promulgated through the terms of outsourced contracts and for labour standards, in particular, to be monitored and enforced by the same mechanisms that are used so effectively to control quality-critical issues. We argue that this is true not only in those countries with poorly developed labour rights but also in those, like South Africa, with relatively strong unions and progressive labour legislation. For these reasons, there are grounds for suggesting that, rather than being overestimated, IFAs have yet to be treated sufficiently seriously.

The research
The focus of this article is the South African operation of Lafarge, a global producer of cement, aggregates and ready-mix concrete. We examine the impediments and the possibilities for translating the standards embodied in the Lafarge IFA into practice. The choice of company is important; the circumstances where implementation of IFAs is most problematic are reasonably clear (Davies et al. 2011), and this study sets out to explore the transfer of labour standards in more favourable conditions. Given that the challenges for IFA implementation are greatest in value chains that are only weakly 'driven' by the lead company (Riisgaard and Hammer 2011) and where a large number of suppliers compete for short-term contracts, we might expect the most fertile ground to be taken up by large, regionally dominant companies with relatively stable contractual relationships. Equally, South African employment legislation is compliant with relevant ILO conventions and the industrial relations machinery is, nominally at least, supportive of trade unions. In this study, we demonstrate how the outsourcing of a core activity can facilitate, on the one hand, a highly interventionist approach to quality-critical criteria -ostensibly a highly 'driven' relationship -and, on the other, a distancing from operational decisions that impact on labour standards.
Interviewees were selected with the aim of 'following' the IFA through the various levels of the company itself and its contractors. Key informants within Lafarge's South African subsidiary were interviewed in South Africa during October 2010. These included HR managers with a company-wide remit, as well as operational managers in business units. We also interviewed haulage subcontractors and recruitment consultants. These interviews were arranged with the assistance of Lafarge. Semi-structured interviews were conducted in the workplace and individuals were asked to describe the dayto-day operation of their unit, the interrelationship between client and contractors, as well as current labour relations from the perspective of their own work role.
In addition, officers from the BWI's Africa regional office in Johannesburg, full-time officials and elected stewards of the National Union of Mineworkers (NUM; representing in-house Lafarge employees) were interviewed, as well as officials of the transport union (South African Transport and Allied Workers Union) responsible for the collective bargaining agreement that covers the transport subcontractors. We also interviewed officials from the bargaining council for the freight industry, which is responsible for the collective bargaining agreement that covers the Lafarge haulage contractors discussed in this article. At total of 14 interviews were conducted during this part of the research.
The account presented here also draws on an earlier round of interviews in 2009, involving officials from the two key unions operating in the construction sector (NUM and the Building, Construction and Allied Workers Union [BCAWU]: seven interviews in total). Finally, the study utilizes a range of documentation from the company, the relevant unions, South African government bodies and South African industry organizations. This article forms part of a larger project that examines the impact of IFAs on the activities of three signatory construction MNCs in three countries (South Africa, Russia and India).

Lafarge in context
The South African construction industry employs over 1 million workers. Of these, 739,000 are in the formal sector and 293,000 are in the informal sector (Statistics South Africa 2011a). The market for construction products is dominated by three very large producers, with Lafarge accounting for 27 per cent of a market worth R5.4 billion in 2003R5.4 billion in (McCutcheon 2003. The construction industry as a whole is characterized by high levels of casual and informal labour, a poor safety record and very low rates of union membership. South Africa is no exception, with estimates of union membership ranging from 16 per cent (McCutcheon 2003: 33) to 12 per cent (Budlender 2009: 15) or as low as 9 per cent (Cottle 2010). Many of the reasons for this are familiar in other countries: for example, the large number of small businesses, the extensive use of casual work and the presence of a significant informal sector.
The Lafarge Group employs 76,000 people in 78 countries (Lafarge 2011: 24). Lafarge's wholly owned subsidiary, Lafarge South Africa Holdings Pty, was acquired in 1998. The parent company is Lafarge S.A. -a Limited Liability Company (Société Anonyme) incorporated in France under French law. Lafarge (2011: 24) describes itself as the world leader in the cement market, the second largest aggregates producer, the third largest concrete producer and the third largest gypsum wallboard manufacturer worldwide.
Lafarge regards sub-Saharan Africa as an important emerging market and South Africa as a key location for the company's expansionary policy. Within the country, it operates through the wholly owned Lafarge South Africa Holdings (Pty) Limited, which itself has four business units: aggregates, ready-mix concrete, cement and gypsum products such as plasterboard (Lafarge South Africa 2011).
The production of cement and ready-mix concrete differs from building site work in obvious respects. The nature of production means that the plant is relatively fixed and permanent and, related to this, the core workforce is in relatively stable employment. These businesses have very different profiles. For cement, there are high barriers to entry, competitors are large companies and the product is not perishable. The majority of workers are directly employed, augmented by agency workers and casual labour. This business unit is characterized by a bureaucratic/supervisory control regime. For the ready-mix concrete business, because of the perishable nature of the product and high transport costs, plants are dispersed around the country, based in 'micro markets'. Compared with cement, barriers to entry are low and there are many small competitors, adding to the volatility of these micro markets. So, plants may be dismantled and moved as market circumstances dictate. The core workforce comprises approximately 2,500 direct employees. These are almost all South African nationals, with less than 1 per cent migrant workers (interview, HR manager, October 2010). In addition to direct employees, there are around 450 contractors' staff in transport for ready-mix and 300 in transport for aggregates (interview, HR manager, October 2010). Each business unit has a separate human resources structure and collective bargaining agreement.
The fact that the construction process is, in essence, site bound means that it does not lend itself to 'spatial fixes' (Silver 2003) and is, therefore, inextricably part of local labour control regimes (Jonas 1996). The production of cement and concrete, for example, while demanding economies of scale in production, is nevertheless, because of the nature of the product and of the process, integrated within, and dependent upon, a defined geographical market.

Introduction
Lafarge sees itself as having a positive relationship with unions and reported that in 2010, '67% of Group employees are represented by elected representatives or unions' (Lafarge 2011: 112) and 71 per cent of business units are covered by collective agreements (Lafarge 2011: 113). Direct employees in the cement and ready-mix business units (up to the deputy sales managers -a middle management role) are overwhelmingly members of the NUM, with a number of skilled workers in a smaller union, Solidarity. Trade unions in the construction industry have historically been weak (Goldman 2003). The NUM, which began as a mining union, started organizing in the construction industry after a COSATU (Congress of South African Trade Unions) resolution in 1997 recommended merger with the Construction and Allied Workers Union (CAWU). This was after several failed attempts to revitalize the construction sector affiliate.
The place of centralized collective bargaining in South Africa is underpinned by the 1995 Labour Relations Act (LRA), which sets out the formal status of Bargaining Councils: voluntary membership bodies that have the power to conclude and enforce collective agreements and -depending on membership density -to extend these to the rest of the industry. The law also provides a range of protections for union organization, including closed or 'agency' shop agreements. However, there is no bargaining council in the construction manufacturing industry and councils have generally been in decline, with a number of them collapsing in recent years (Godfrey et al. 2007). The recognition agreement in the Ready-Mix small business unit (SBU) creates an NUM 'agency shop' under Section 25 of the LRA. This aims to prevent freeloading by the automatic deduction of 'fees' from the pay of all employees who are not union members. The agreement also formalizes a set of organizing rights and trade union facilities, including union access to the employer's premises, a check-off system and time-off for union duties. Between disputes, at least, there appears to be a co-operative employer approach for Lafarge's own workers. Stewards' remits, for example, can extend between plants when representation is not available locally. Shop stewards are ordinarily employed on a full-time basis, paid by the employer. However, Lafarge had recently dismissed a steward following a dispute and no full-time shop steward was present at the Ready Mix SBU at the time of our interviews.
The transport division is contracted out and, as we demonstrate below, the way in which work is organized has enabled the client to maintain a high degree of detailed control while minimizing the opportunity for crosscompany union organization. Ready-mix truck drivers are not unionized and are not part of the bargaining unit. Transport subcontractors are, however, covered by the Bargaining Council for the freight sector, through extension of the collective agreement (interview, shop steward Ready Mix SBU, May 2011; NBCRFLI 2009). The agreements of this bargaining council are extended to non-parties in terms of Section 32(2) of the LRA, including the 'agency shop' agreement. However, these employees are not union members and SATAWU, 1 the main union in the transport and logistics sector -the subcontracted division of the SBU -does not actively attempt to organize the subcontractors.

The Lafarge IFA
The Lafarge IFA was signed in 2005 by the company, BWI and ICEM 2 (the GUF for mineworkers). The IFA commits Lafarge to the ILO's Declaration on Fundamental Principles and Rights at Work, the ILO Tripartite Declaration of Principles concerning Multinational Enterprises and Social Policy, the United Nations Global Compact and also the OECD's Guidelines for Multinational Enterprises. It also commits the company to several specific ILO conventions, including the 'core' labour standards. Of the conventions listed (29, 87, 98, 100, 105, 111, 135, 138, 155 and 182), South Africa has ratified all except 135 (Workers' Representatives).
Compared with some IFAs, the Lafarge agreement includes a relatively weakly worded commitment to extending provisions to contractors: Lafarge will seek to use the services of those trading partners, subcontractors and suppliers, which recognise and implement the principles listed below (Lafarge IFA). This wording can be contrasted with that of the IFA of Royal Bam, another construction MNC operating in South Africa: Royal BAM Group NV considers the respect for workers' rights to be a crucial element in sustainable development and will therefore refrain from using the services of those trading partners, subcontractors and suppliers which do not respect the criteria listed above (Royal Bam IFA).
Following pressure from the BWI, Lafarge reports that it . . . initiated an audit of sub-contracting activities and carried out surveys to investigate how fundamental labor rights are embodied in our labor practices (Lafarge 2010: 17). This was designed to 'ensure that fundamental rights are preserved in outsourcing contracts and share good practice' (Lafarge 2010: 17). BWI, while welcoming this, also noted that, throughout the company, outsourcing was increasing and that Trade unions expect a clear message from Lafarge to take on responsibility for human rights and supply chain management in its businesses in all countries (Lafarge 2010: 18).
The clause on monitoring is equally vague, with no explicit mention of enforcement: A reference group consisting of representatives of the Lafarge management and the signatory international federations shall meet at least once a year, or whenever necessary, to follow up and review the implementation of this agreement (Lafarge IFA).
According to the European Trade Union Confederation (ETUC 2010), there is also regular telephone and email contact and Regarding monitoring, there are regular meetings with the signatories. One or two people from ICEM and one from BWI meet twice a year or more if difficulties arise (ETUC 2010: 93).
The framework agreement itself is invisible in the South African subsidiary. This is not surprising given the absence of any mention of it in the parent company's annual report or the websites of the parent or South African subsidiary. A senior HR manager states that the IFA has 'no impact at all': The only way we are governed by international agreements would be for things like an insurance provider . . . But in terms of unions, labour relations, that is dealt with in SA with the NUM (interview, HR manager, October 2010).
Similarly, a transport supervisor, who has not heard of the IFA, states: Labour relations and bargaining units is dealt with nationally and we never get information of the European unions or American unions about how they negotiate or what kind of contracts they sign. We don't have knowledge of that (interview, transport manager, October 2010). This lack of awareness extends to the trade union. The NUM organizer for Lafarge claimed to have heard of the IFA but was not familiar with any relevant details. He suggested that shop steward councils are provided with all agreements, including IFAs. However, a shop steward from the Ready Mix business unit commented: I have never heard of that agreement. I guess they are handled by our national executive, but it has not yet filtered down to us on the shop floor. . . . I deal with day-to-day shop floor issues affecting our members and certainly do not in any way involve such global agreements (interview, October 2010).
The ETUC (2010: 94) reports that 'local management and unions are involved in implementation . . .'. It is difficult to see how this could be the case, given the lack of knowledge of the IFA that we found, and the ETUC (2010: 94) does concede that there is 'poor information on the agreement at the local level'. This is in the context of an otherwise highly centralized and directive management approach. This distinction is highlighted by the comparison between the core, in-house employees and the peripheral, contingent workforce.

Managing the Core Workforce
In terms of the number of activities involved, cement manufacture is an uncomplicated industry, making it possible to define organizational boundaries sharply. Lafarge's own plants have a core of directly employed workers, while activities defined as non-core or temporary are bought in. For example, Lafarge outsources mining, blasting and excavation, although these are clearly integral parts of the process. Concentration of ownership, coupled with the cost of transport, means that some of these firms are heavily dependent on Lafarge contracts. The main gypsum processing plant, for example, which is operated by a third party, was constructed near to Lafarge's Lichtenburg site.
Several respondents described Lafarge as a 'multi-local' organization in which 'we don't necessarily wait for France to make a call on us -we do make local decisions as well' (interview, site manager, October 2010). On some issues however, such as occupational health and safety, head office control is very direct and specific, with standards defined and applied to all contractors worldwide (also see ETUC 2010). A similar approach is now taken to ensuring the 'sustainability' credentials of suppliers and contractors. A plant supervisor explained how instructions are cascaded down from the MNC HQ: They have a global safety manager and that will have an African counterpart, an Asian counterpart . . . and then it starts becoming a web until it comes to a South African safety manager and a regional safety manager . . . It would be the same set of rules . . . if I went to Lafarge Kenya I would expect to find the same rules, the same safety walkways, the same signage, the same paperwork . . . (interview, plant supervisor, October 2010).
Health and safety and sustainability are not the only areas subject to global control. Pay and grading structures and principles relating to remuneration and incentives are also set by the company headquarters: The job grading system is decided in Paris. You have no choice. You will use the Hay job grading system. You are not allowed to look at [other factors]. That is it -that's the 'law'. Performance bonus structure for execs is decided in Paris. They decide what will incentivise people . . . Your internal auditors are decided in Paris . . . Expat management -how you manage them and how their packages are structured -decided in Paris (interview, HR manager, October 2010).
The company thus makes a distinction between terms, such as those of the IFA, and other requirements that are considered critical to quality. As the next section demonstrates, the latter are enforceable even in complex and fluid outsourcing arrangements.

Managing Agency and Casual Labour
The construction industry can be seen as a barometer of economic activity. In South Africa, a peak during the run up to the World Cup in 2010 has been followed by a pronounced slow-down. Lafarge claims to use temporary labour specifically to cope with peaks in workload and to act as a buffer for core jobs. As is the case throughout the industry, temporary labour is provided via third-party contractors. There are a series of distinctions made between types of operators. Recruitment 'consultants' are used for more specific posts that require some selection, whereas temporary manual labour is provided by labour brokers and labour-only subcontractors. Labour brokers (also called temporary employment services) provide a specified number of workers with particular skills to a client for a fixed period of time. The broker remains the employer, but the worker is under the control and direction of the client, who dictates tasks, methods and timing of the work to be done. By contrast, a labour-only subcontractor may be considered a straightforward outsourcing operation, hired by the client to carry out a specific task over a set period. The subcontractor has a certain amount of autonomy in carrying out the task and employs their own workers who are not directly controlled by the client (Bamu and Godfrey 2009). Larger and more reputable brokers distinguish themselves from the 'bakkie brigade' (Bamu and Godfrey 2009). A recruitment consultant explains: [They] arrive with a truck in the morning. The employees are gathered somewhere. They have never seen them. They say 'We need 20 people to dig roads'. These people don't know what a road is, never mind how to dig it (interview, October 2010).
Workers recruited through registered temporary employment services are legally employed by the contractor but are integrated into the host company as any other member of staff.
In theory we are the boss, but . . . their contract that they receive from us says that they will abide by Lafarge's procedures. We even say who they will be reporting to.
Even though we are the ultimate employer and we do the payroll, they will report to [a Lafarge supervisor]. We don't interfere in that relationship. She is our employee but we play more of a HR role . . . they abide by the company's rules. The company decides what are their hours, what is the lunchtime, what they want from the person and the wage rate (interview, recruitment consultant, October 2010).
The contractor's role extends to disciplinary management: 'They say "Come and sort this person out". All the labour problems -all the nasties -we handle' (interview, recruitment consultant, October 2010). In these cases, though, there is close co-ordination between contractor and client and the contractor's ability to intervene rests on the client's power to vary the terms of the contract. So, although temporary employees are entitled to union representation, the disciplinary process in such cases tends to be curtailed: 'Generally, the client would shorten the term, because the term isn't definite' (interview, recruitment consultant, October 2010).
Trade unions elsewhere have struggled to organize temporary and casual workers, but this is particularly challenging in an economy with high and growing unemployment. The fragmentation of the workforce has drawn new patterns of social exclusion and imposed constraints on the labour movement (Webster and Von Holdt 2005). For example, the NUM succeeded in including new provisions for 'limited duration contracts' in the 2009 agreement for the civil engineering sector, giving workers with 18 months of continuous employment equal rights with permanent employees. However, perhaps predictably, the employer response is often to terminate employment before this point (interview, NUM national officer, October 2010).

Managing Contractors
From the early 1980s, Lafarge SA began to outsource the transport of ready-mix concrete. This was explicitly driven by the potential for cost savings. Initially, this work was transferred to a large number of ownerdrivers, each operating a single truck. This maximized Lafarge's market power with respect to its contractors, but it also meant that the company was reliant on individual drivers for the delivery of a perishable product. For this reason, Lafarge encouraged contractors to expand, allowing up to six trucks per owner (Lafarge South Africa 2010). One consequence of this increase in contractor size is that individual drivers are now employees of the contractor and thus covered by labour relations legislation. The outsourcing of this responsibility is an integral part of strategy: 'Drivers would triple the payroll. And strikes -there is quite a lot of baggage to carry all these trucks' (interview, plant supervisor, October 2010).
These contractors are integrated closely into Lafarge's operation and are subject to detailed control. Schedules, sent from the national head office, set out truck movements and loads, so that 'The driver has nothing more to do other than collect his delivery schedule and concentrate on providing a quality service' (Lafarge South Africa 2010). As deliveries of ready-mix cover the whole country, Lafarge needs its fleet to be positioned to match demand in the various regions. Contractors may be required to move trucks (and drivers) at short notice (interview, transport contractor, October 2010). When orders decreased in 2010, for example, following the completion of the various World Cup projects, contractors with trucks at plants in the north of the country were told to move these to the south (interview, sales manager, October 2010). This re-siting of trucks, along with day-to-day movement, is closely monitored by Lafarge's truck scheduling programme, which enables the client not only to keep track of truck availability but also to calculate the contractors' performance in terms of profit, load and mileage per truck.
The owners have signed a service level agreement saying that 95 or 98% the truck is available for business. We would like all our trucks to be 100% utilised . . . if there is no income, we are going to have to move them into the next level to say You are going to have to move to this section to make money for yourself (interview, sales manager, October 2010).
This degree of control is partly a function of the labour market during recession. But it is also the result of the nature of the contract relationship. Contractors begin from a position of weakness and are then locked in to a long-term commitment that makes them entirely dependent on the continued allocation of work.
Trucks are bought on loan with Lafarge surety (loans of R38 million up to 2009). Although drivers were given a 'free hand' in their choice of truck, they all use the same model (Mercedes Benz South Africa 2010). Every contractor also opted to take up the approved finance and insurance package. Some who began as owner-drivers now operate more than 20 trucks. Having tied up this capital, though, contractors have little control over their investment. Although they own (or are paying for) the truck cab, the mixer unit remains the property of Lafarge. This is clearly not something that can be replaced quickly, and particularly because this carries the Lafarge insignia, the contractor's own cab cannot be put to any other use. The contractors' investment effectively ties them to the current client.
Contractors' single use investment in trucks makes them entirely dependent on Lafarge. In this sense, contractors are akin to franchisees. The other side of this is that the investment requires Lafarge to act as guarantor for contractors' debt. The company therefore has an incentive to maximize the use of the asset for the duration of the debt, giving some measure of security of employment. It is when the debt is paid off that contracts are reassessed.
We had a meeting and they said 'All the trucks that's been paid up, the contracts are now finished. We are not going to renew the contracts'. . . . Five of my trucks were taken out of the fleet and the mixers were taken off. . . . I did not retrench anyone -I just offered them another position in Jo'burg in trucks that I still owe on. So they did do it. There were a lot of trucks taken out of the fleet. Say there were 200 in the fleet, but there was no work for that amount of trucks . . . all the trucks that's been paid were taken out of the fleet (interview, transport contractor, October 2010). This is clearly an extremely unequal, 'captive' relationship, in which the client is able to exercise considerable power. This is evident in the company's approach to health and safety management. [Contractors] have to adhere to Lafarge's safety standards. We train our contractor on safety. PPE [personal protective equipment] -that is Lafarge standards. The contractor has to purchase PPE from Lafarge to ensure that it is up to our standards. Working hours -overtime regulations -we guide them on that but we obviously can't enforce it (interview, HR manager, October 2010).
When necessary, then, Lafarge is able to impose conditions methodically along the value chain. This contrasts with the treatment of the company's commitments to labour standards and makes the lack of penetration of the IFA particularly interesting.

Trade union strategies and prospects
Some aspects of the challenge posed by outsourcing to union organization and strategy can be seen as universal and predictable. The break-up of larger membership and bargaining units increases the call on resources while at the same time limiting the union's ability to respond. As the viability of the union as a whole ultimately depends on the viability of individual units (Willman 2001), unions face fundamental decisions concerning the allocation of resources and the prioritization of organizing activity. For this reason, outsourcing calls into question the uneasy tension between workplace democracy and managed centralism that is a feature of many unions.
However, beyond this level of generality, the impact of restructuring is a product of time and place: in this case, moulded by the way in which the South African economy and industrial relations have evolved since 1994. To see the specifics of union action primarily as responses to employer initiatives is to understate the interconnectedness involved in such a process of evolution; nowhere is this more apparent than in South Africa. The changing contours of national labour control regimes present both employers and unions with new opportunities, constraints and priorities. Not only union responses, but also company outsourcing strategies themselves need to be understood in this context. By most objective measures, the history of South African trade unions since the fall of the apartheid regime has been one of remarkable success. COSATU's revolutionary role pre-1994 and its formal place with the African National Congress (ANC) and the South African Communist Party (SACP) in the 'triple alliance' made it possible to secure constitutional status for labour rights and to build a legislative and regulatory framework that is supportive of collective bargaining and gives unions a significant voice via social partnership institutions at a national level. Early membership growth has been checked by the neoliberal turn of government economic policy, by economic slow-down, as well as by the type of restructuring discussed in this article, but nevertheless, COSATU currently claims 1.8 million paid-up members and total membership stands at 3.2 million, or 25 per cent of the workforce (Budlender 2009;COSATU 2011). This history has embedded trade unions, as institutions, within national life and established principles of collective organization that are yet to be seriously challenged. The declared aim of the 1994 LRA, for example, was to promote economic development, social justice, 'labour peace' and democracy in the workplace. On the other hand, it can be argued that unions' influence and their success as institutions has been at the expense of a broader, independent agenda. Buhlungu (2010), for example, argues that the centralization of power within COSATU and the refocusing of union priorities on a narrow, economistic agenda can be seen as a 'paradox of victory' in which 'the fruits of [union] victories continue to elude them as the processes of liberalisation that they champion almost always result in the organisational weakening of union structures' (Buhlungu 2010: 17).
The challenge posed by an increasingly fragmented workforce in a fragmented labour market is quite different from that of a relatively stable, unskilled workforce in mining and manufacturing (Webster and Von Holdt 2005). Organizing this more contingent and precarious workforce is inherently difficult, but on the other hand, retrenchment around an organizable 'core' is itself unsustainable, as this core is progressively eroded. This debate is not unique to South Africa or to the construction industry, and it is clear that organizing subcontractors is not impossible but does have broader implications. MacKenzie (2009MacKenzie ( , 2010 shows how successful union organizing in the Irish telecommunications sector was made possible by a shift of strategy, from the exclusion to the active targeting of contingent workers. Such union engagement may, however, have the effect of 'de-stigmatizing' the use of contractors (MacKenzie 2009: 558). Given South African trade unions' long-standing campaign to ban labour brokers, this fear of legitimization goes some way to explain the NUM's stance, not only on brokers, but contingent labour in general.
As noted in the case of Lafarge, the law sets out employer responsibilities with respect to trade union organization and formalizes the right to strike. Bargaining councils also administer legally binding closed shop, or 'agency shop' agreements for member organizations, which, when enforced, tackle the problem of free-riders. Crucially, however, the law does not impose a duty on employers to bargain.
In the case of the bargaining council for the road freight industry, the agreement specifically applies to subcontractors: An employer who subcontracts work falling within the Council's registered scope shall be jointly and severally liable, together with the subcontractor, for the subcontractor's compliance with the provisions of this Agreement (NBCRFLI 2009: 38.2).
It is worth noting, however, that this refers to subcontracting by affiliated employers; the separation of such bargaining council jurisdictions means that the main contractor -which is not itself a freight operator -is not covered.
The bargaining council structure also sets union demarcations, meaning in this case that the NUM is not able to recruit drivers. Section 12 of the LRA gives the clear right to trade union representation, as well as the right to hold union meetings on employer's premises. Outsourcing, though, has provided the pretext for undermining such rights because the client is not technically the 'employer'. Nevertheless, South African law remains relatively consistent, clear, and supportive and might reasonably be used as the platform for organizing.
Given this last point, the unions' track record is, perhaps, surprising. Subcontracted drivers working for Lafarge are not union members. In other words, their 'agency shop' fee is deducted, and they are, as individuals, covered by the terms of the collective agreement, but they are not represented. The difficulties in organizing temporary and agency labour are understandably much more serious. The NUM claims to attempt to recruit limited-duration workers, for example, but sees this as an 'impossible' task (interview, NUM officer, October 2010.) These constitute 60 per cent of the total construction workforce according to the NUM and most of these workers have, effectively, been categorized as non-organizable. Even in large civil engineering projects, unions have taken the view that it is not financially viable to target employers with fewer than 50 workers (interviews, NUM and BCAWU National Coordinators, August 2009).
In order to understand this, we need to recognize that the way in which restructuring and fragmentation affect the ways in which unions operate is more complex than a simple response to employer initiatives. Unions have been forced to prioritize resources and to adopt a more 'strategic', managed approach to organizing. Thus, while militancy can be seen as a response to economic circumstances (Kraus 2007), it also needs to be seen in the context of a 'social movement' unionism that has proved unsustainable (Bramble 2003). Von Holdt (2002: 10) argues that under apartheid, rather than agents for the 'negotiation of order' (Hyman 1975: 11) in the workplace, as in the industrial democracies, unions were agents for the 'generation of disorder' as part of the struggle against white minority rule. The model of industrial unionism that remains appears incapable of responding to the scale of restructuring under neoliberal reconstruction (Buhlungu 2010). As South Africa's unions have become more institutionally embedded, it is arguable that they are beginning to face some of the problems of their sister unions in the northern economies (Baccaro et al. 2003), in that they rely more and more on their institutional position rather than membership strength and, in so doing, risk further demobilization.

Conclusion
The focus on freedom of organization and representation signifies a key difference between IFAs and corporate codes of conduct. Rather than attempting to 'rachet up' standards from afar, IFAs hold out the possibility of building sustainable local union capacity. It is this aim alone that avoids the charge made of voluntary corporate arrangements, that by casting workers as victims, rather than active agents, they risk further disempowering labour (Seidman 2007).
However, logical problems become immediately apparent. On the one hand, a commitment to core labour standards is of most obvious relevance to workers in countries where those standards are not part of domestic law. On the other hand, in the absence of an international jurisdiction, it is in these very labour-repressive countries where effective enforcement of 'private' standards is most difficult. More generally, the prospect of enforcing freedom of organization where there is no extant union and where no such right actually exists in law remains a conundrum. For this reason, there is a growing awareness of a point that might appear self-evident in national studies of industrial relations: negotiated agreements are only the starting point, not ends in themselves.
In this respect, South Africa represents an interesting test case: a country with progressive labour laws, a supportive industrial relations framework and a union movement that remains strong, both industrially and politically. The rights outlined in the agreement are already protected and enforceable in SA labour law. And although, like many emerging and developing economies, it has a high proportion of workers in the informal sector -around 30 per cent (Statistics South Africa 2011b) -its industrial relations system in the formal sector is highly developed, with collective bargaining embedded in key industries and services. The IFA was presumably not negotiated with countries such as South Africa in mind.
Our case study demonstrates that this profile, while accurate at the aggregate, national level, masks important factors. We might expect Lafarge in South Africa to be a 'textbook' example of how negotiated items are enforced with the full force of global headquarters' authority. In fact, the company's fragmentation strategy has combined with the strict jurisdictional rules on union coverage to weaken the position of workers. In the case of ready-mix delivery, outsourcing means that what would once have been considered a 'core' activity has been converted to an ostensibly straightforward market relationship. Nevertheless, the form of this relationship allows the client to exercise exactly the same level of control of the work process as would be the case with direct employees. Yet, on the other hand, contractors can be treated as independent third parties who are not subject to the client's HR policies or broader commitments on labour standards or industrial relations.
A combination of low market power and high asset-specificity means that contractors are effectively captive. This has enabled the client to demand a degree of flexibility and responsiveness that appears to be feasible only because the cost of this flexibility has been shifted to the contractor. Operational control also includes a close specification of contractors' obligations with respect to quality-critical factors and areas, such as safety, where liability is not transferable. On the other hand, personnel management issues, particularly those associated with labour standards, are distanced from the main company's remit. Thus, a partial deregulation of the employment relationship is accompanied by a reregulation of the production process. While some of the bureaucracy associated with managing the employment relationship has migrated from client to contractor, the client's approach involves highly formalized performance monitoring, effectively reintegrating the work process (MacKenzie 2000(MacKenzie , 2002. Trust, in this case, requires vigilance.
It is possible to see the IFA as part of this process of reregulation. The IFA is significant in that it aims to cover the whole value chain, as opposed to the partial coverage of the various bargaining agreements. The extent to which IFA commitments impact on local practice is clearly influenced by the way in which production is fragmented along the value chain. However, a focus on structure is only a starting point because particular client-contractor arrangements may either facilitate or obstruct the application of standards or practices. In order to understand how the disintegration of ownership can coexist with close operational integration, we argue that an awareness of the power asymmetries embedded in value chain links needs to be accompanied by a closer focus on the nature of the work process.
If union organization is seen as the keystone of any campaign for labour rights, then this calls for strategy at all levels to mesh with local priorities. We have shown how Lafarge is prepared to impose standards and procedures throughout its value chain in certain areas -health and safety, sustainability, and pay and grading systems. There is no reason why this same approach could not be extended to labour standards and labour rights. Equally, this requires unions to readopt a strategy of organizing workers, rather than organizing employers, in order to overcome jurisdictional problems in which employees fall between two or more areas of union responsibility. GUFs have a useful function in this regard also.
The proliferating literature on the subject has sometimes set high expectations of IFAs, yet evidence of concrete outcomes remains patchy. This might suggest that the IFA is a solution in search of a problem. If the problem is essentially a transnational one -the attempt to win concessions on behalf of a global workforce by influencing MNC policy -then perhaps unions should be wary of being 'trapped in localism' (Burawoy 2010: 306). But if the intended gains are local, and if we accept that such battles are necessarily won or lost at the local level, then the challenge is for international bargaining to go with the grain of workplace organizing. On this basis, it appears that both the limitations and the untapped potential of IFAs have been underestimated.
Final version accepted on 25 October 2012. by CRIMT, the Inter-University Centre for Globalisation and Work, and we would especially like to thank Reynald Bourque for his help and encouragement.