Temporary Agency Work and Trade Unions in Comparative Perspective

The article examines local union approaches toward temporary agency work in Belgium and Germany. Heterogeneous plant-level use of temporary work is explained by differences in collective bargaining, together with representation structures and rights for agency workers. Specifically, within a context providing effective rights for representation, the Belgian unions responded to firms’ economic difficulties by improving the working conditions of agency workers through negotiating plant-level agreements that contributed to fostering equality between agency workers and regular workers. In contrast, agency work became instrumental in safeguarding the core workforce’s employment in the German workplaces, where the representation rights were absent. As a result, the status of agency workers remained vulnerable.


Introduction
Temporary agency work (TAW) is considered to be one of the most vulnerable forms of employment (Fudge, 2011). The use of such work arrangements, where a worker is hired by an agency but performs work in a user firm, has grown rapidly in Europe in recent years (Håkansson & Isidorsson, 2012). According to the European Statistical Office, about 15% of workers were employed on temporary contracts (fixed term and agency) in the Euro-zone in 2012 (Eurostat, 2013).
Although TAW represents a challenge for trade unions, research has hardly begun to examine the policies and practices negotiated by employee representatives at the workplace level. Yet such practices are likely to influence the job stability and working conditions of agency workers. According to Olsen (2005), local unions face a dilemma when coping with TAW: should they utilize it to increase the regular workforce's job security or should they represent a plant's whole workforce regardless of contractual status? Recent labor market developments point to the second alternative, as the increasing use of flexible work arrangements has contributed to the evolution of a "fragmented landscape of labor relations" (Holst, 2014, p. 3), with traditional coreperiphery divides becoming weaker and blurring the borders between the different contractual groups. This is likely to put pressure on the core workforce. Specifically, Peck and Theodore (2007) demonstrate how agency workers are used to discipline the permanent staff in the U.S. labor market. Eichhorst and Marx (2011) report similar findings for the German context. So why do trade unions not oppose TAW, if it threatens the working conditions of the permanent workforce?
The article addresses the gap in the research by studying the policies and practices negotiated by local employee representatives within four similar subsidiaries of two U.S.-American multinationals in Belgium and Germany. Drawing on the analysis of Locke, Piore, and Kochan (1995), which explains heterogeneity in union responses through differences in the national institutional system, this study further illustrates the point that systems of employee representation and collective bargaining come together to shape diversity in local unions' strategic choices with regard to TAW.
Germany and Belgium both feature a union-dominated works council system, and agency workers are entitled to vote in the social elections (company-level elections of employee representatives) in the user firm after having worked there for at least three months. As practically all agency workers in the two Belgian workplaces could exercise this right, local employee 575633S GOXXX10.1177/2158244015575633SAGE OpenDoerflinger andPulignano research-article2015 1 Catholic University Leuven, Belgium representatives attempted to gain their votes by putting TAWrelated issues on the bargaining agenda. As a result, workplace-level bargaining in the Belgian plants was more effective in limiting the use of TAW and at the same time obtaining increased job stability and enhanced working conditions for agency workers. The Belgian unions also incorporated agency work into central (sector) agreements while locally negotiating improvements in agency workers' employment prospects. The situation was different in Germany, where agency workers mostly stayed for only short periods of time in a specific workplace. Consequently, they were not entitled to vote in the user firms' social elections, which reduced the works councillors' interest in protecting them. Moreover, disorganized bargaining decentralization and the growing use of opening clauses (clauses in sector-level collective agreements that allow company-level deviations from the standards set at the sector level under the condition that both works councillors and management at the company level agree) led to a power shift from the sector level to the workplace level in Germany. Under the constant threat of plant closure, job losses, and restructuring, the works councillors' major objective was to safeguard the employment of the core workforce. As the acceptance of relatively high levels of TAW contributed to this aim, works councillors engaged in concessions to secure the employment and working conditions of the core workforce.
In theoretical terms, this article suggests a heuristic view of the industrial relations and labor market institutions as influencing trade union attitudes toward TAW. Agency workers not only face differences in pay and working conditions due to national regulatory contexts, but they also face differences in regard to representation rights. Those should be taken more fully into consideration when researchers are studying trade unions' approaches toward agency work.
The article is structured as follows. The first part sheds light on theoretical perspectives around union approaches to TAW. Having explained the regulatory settings of TAW in Belgium and Germany, the article turns to the national/sectoral union policy toward agency work in the two countries. Before focusing on the empirical evidence regarding the four workplaces, the research design and methodology of the study are presented. After a discussion of the results, the article ends by drawing conclusions and indicating the implications of the study.

Theoretical Perspectives on Agency Work: The Role of Trade Unions
Changing legal regulations have encouraged companies to use flexible forms of work (Antoni & Jahn, 2009;De Cuyper et al., 2008;Mitlacher, 2007). Unions-which need to find responses to this-may tend to oppose TAW because of the difficulty in organizing and representing such workers (Heery & Abbott, 2000;Pfeffer & Baron, 1988). Moreover, the division of workforces into a core, consisting of regular, high-skilled, and unionized workers with career opportunities, and a periphery, comprising workers who are easily disposable and usually nonunionized, leads to a fracturing of collectivism, which was a traditional characteristic of unions (Bacon & Storey, 1996). In addition, TAW is mostly considered as insecure and unstable work. Thus, unions would exclude agency workers to keep their numbers low, positioning themselves as supportive solely of forms of work with decent levels of job security (Gallagher & Sverke, 2005;Gumbrell-McCormick, 2011). As Houseman (2001) illustrates, this should lead to a negative relation between unionization and the use of agency work. However, Böheim and Zweimüller (2003) find that under specific circumstances, a high level of unionization does not automatically imply a low level of use of temporary work. Cahuc and Zylberberg (2004) present a possible explanation because unions may utilize TAW as a buffer to protect regular workers from fluctuations in labor demands. Thus, the aforementioned research accounts use polarized conceptualizations of trade unions: They are either in favor of TAW or against it (Lindbeck & Snower, 1988). However, recent contributions have provided alternative approaches. Accordingly, by referring to Hyman's (2001) typology of trade union identity, Heery (2004) shows that unions' reactions toward TAW are situated between "accepting" and "rejecting." Specifically, five types of union responses to TAW are identified, which refer to different types of union identities: exclusion, servicing, partnership, social dialogue, and mobilization (Heery & Abbott, 2000). "Exclusion" stands for the trade union's rejection of agency workers, as they could threaten not only the permanent employees' security but also the power of the union. "Servicing" refers to unions offering individual services to agency workers. Thus, it is based on the acceptance of TAW, although unions could still tend to defend the interests of their members. "Partnership" reflects union attempts to join forces with management and push for agreements, for instance, on job security, but such agreements are likely to create benefits for permanent employees only. "Dialogue" reflects approaches to influence government policy in a way to secure the members' jobs and reduce the vulnerability of agency workers. Finally, "mobilization" refers to unions as social movements, mobilizing its members and the wider public to fight against injustice at work.
However, using only such research accounts to explain union responses to TAW might be shortsighted, as tensions could arise between different union strategies and within union policy. Bergström and Styhre (2010) stress that unions could coalign between two contrasting logics: accepting the use of TAW in practice while resisting it in principle. In particular, trade unions may develop distinctive practices at the plant level that may influence the working conditions of agency workers. Hardly any study so far has scrutinized the extent to which national/sectoral trade union policy is reflected in the practices unions adopt at the workplace level.
Furthermore, there is a lack of consideration of the dynamics explaining such union policies and practices. Pfeffer and Cohen (1984) highlight the fact that national and sectoral regulation and external labor markets in particular, but also management strategies, could constrain local unions' reactions, stating that they cannot "freely" make decisions when negotiating about TAW.
However, it remains unclear how locally developed union practices affect the rights and status of agency workers. Are unions able to turn vulnerable work into secure work, improving workers' rights and working conditions in the workplace? Or do local practices leave agency workers' vulnerability unchanged? How can we explain diverse union approaches toward agency work? To what extent can local union practices contribute to protecting agency workers and their rights in the workplace? If trade unions are not free from pressures coming from situational constraints, how can we explain the differences in the local practices they negotiate?

The Regulation of Agency Work: Belgium and Germany Compared
National-level legislation and sector-level regulation define the general framework for TAW (De Cuyper et al., 2008). Since the mid-1990s, both the Belgian and German systems of employment relations have been exposed to change. However, the resulting decentralization pressures have been handled differently in the two countries. Although the Belgian system has seen a growing number of companylevel agreements, intersector and sector-level agreements are still of crucial importance. Belgium's collective bargaining coverage rate of 96% is among the highest in Europe; and bargaining coordination and centralization scores are comparably high as well (European Commission, 2012). One cause could be the labor legislation of 1968, prohibiting the possibility of deviations from intersector and sector-level collective agreements, that is, deviations that go below the set standards (Pulignano, 2012). Only a limited number of "opening clauses" allowing deviations exists, and their use is highly regulated (Keune, 2011).
Generally, in Belgium, deviations from minimum wages or the equality principle between different groups of workers are impossible. Equality (i.e., in terms of pay and working conditions) is set by law. Regarding representation in the user firm, all agency workers who have worked there for three months or more are entitled to vote in the social elections. They are not, however, allowed to run as candidates themselves, although they are included in the user firm's headcount for the purpose of calculating thresholds relating to worker representation.
In contrast, in Germany, opening clauses have allowed companies to deviate from collectively agreed wages and working conditions to ensure firms' competitiveness (Bispinck & Schulten, 2002). Hence, the shift of power from the sector to the company level has increased the works councils' discretion on local issues.
According to data from the German IAB panel (Institutfür Arbeitmarkt-und Berufsforschung) for 2011, about 35% of the companies in manufacturing included opening clauses in their collective agreements. About 64% of these companies used them to adjust wages, and about 74% were related to working time (Ellguth, 2013). Opening clauses, however, may have contributed to the growth of precarity, hitting especially those workers at the bottom of the labor market (Deppe, 2012). A company's use of agency workers is not contingent upon the presence of an opening clause, but deviation from the legal principle of equal treatment is possible if a valid sector-level collective agreement is in place. This is the case in Germany, but these agreements are far below the standards of other sectors. Although a sectoral minimum wage for agency workers in Germany was introduced in January 2012, it is too low to close the pay gap relative to regular staff. As in Belgium, agency workers in Germany who have been working for the same user firm for more than three months are entitled to vote during the social elections, although they are not authorized to become candidates. So far, agency workers have not been included in the user firms' headcounts to calculate representation thresholds, but this issue is currently under discussion by the German labor court.

Sector-Level Trade Union Policy Toward TAW in Belgium and Germany
From the 1970s, when TAW was legalized, until the end of the 1990s, German unions viewed it as a threat to human dignity and wanted to forbid it (Wölfle, 2008). However, facing high unemployment and job losses, unions were under pressure to make the labor market more flexible, and therefore Germany's main trade union confederation, Deutscher Gewerkschaftsbund (DGB), withdrew its request for a legal prohibition of TAW (Meyer, 2013). The unions could not stop the subsequent deregulation of TAW in Germany, which led to a boom in the whole agency sector: The number of agency workers climbed from 300,000 in 2004 to more than 900,000 in 2012, employed by any one of 18,000 temporary work agencies (Bundesagentur für Arbeit, 2013). However, the gap in wages and working conditions between regular and agency workers has persisted ever since. In the past decade, three principal fields of union activity can be identified (Wölfle, 2008). First, Germany's metalworker union, Industriegewerkschaft (IG) Metall, has been seeking to regulate TAW in collective agreements. The 2012 agreement set pay premiums for agency workers depending on assignment duration in the same establishment as well as the obligation to offer fixed-term employment after two years of working continuously for the same user firm. Second, IG Metall and the service sector union Verdi have been following a strategy of representative participation by building up structures to organize agency workers. Finally, campaigns have become an instrument for mobilization in favor of agency workers' rights.
The Belgian unions' approaches to TAW are also inclusive, but they reveal more stability over time. When the law on TAW came into effect in 1987, trade unions started to engage in creating an appropriate regulative framework (Algemeen Christelijk Vakverbond, 2010). Consequently, agency workers have the same rights and entitlements as regular workers-this is guaranteed by law and collective agreements. Furthermore, unions have increased awareness among agency workers, for example, by holding events to educate them about their rights, particularly in the metalworking sector, where unions have been successful in organizing agency workers (Perin, 2008). Belgian unions in general try to limit the deployment of agency workers in firms while providing equal working conditions, thereby enhancing workplace rights. Thus, they aim at creating a strong basis for equality and job stability for the whole workforce.
In sum, both Belgian and German trade unions are inclusive toward TAW; however, there are differences in the meaning of "being inclusive." When the topic came up, Belgian unions became actively involved and pushed for the regulating of agency work by enhancing rights, whereas the German unions remained passive for a long time, simply demanding the legal prohibition of TAW. In Belgium, the union federations have primarily focused on achieving legal coverage, representation rights, and regulation to guarantee equality between permanent and temporary workers in the workplace, and they have been successful in this regard. In contrast, the German unions are still fighting for this kind of equality.

Research Design and Method
Based on a comparative in-depth case study analysis carried out in 2011 to 2012, the article investigates four workplaces belonging to two American multinationals operating in the automotive sector, both of them located in Belgium and Germany. The two companies, Car1 and Car2, were selected due to their similarity. Comparisons across production sites of the same firm are interesting because the company-wide policy toward TAW is likely to be the same. This likelihood is reinforced by the similarity in the production process, consisting of an automated assembly line in all four workplaces, with staff performing manual assembly functions along the line. The production process provided hardly any opportunities to separate the agency workers and the regular workers. In addition, all four workplaces underwent restructuring because of competition, cost pressure, and the economic crisis.
Car1 employs approximately 24,000 workers (mostly blue-collar) globally within the automotive supply sector. It produces low-tech products and components and depends heavily on its few main customers, which generate about 40% of the revenues. However, the emerging competition from East Asia is forcing Car1 to increase its ability to flexibly react to customer requirements. This creates permanent cost pressure and flexibility needs due to changing production volumes. Within Car1, two highly unionized production sites employing 1,500 to 1,600 workers are studied-Car1BE in Belgium and Car1DE in Germany (see Table 1).
Car2 is a conglomerate operating in various manufacturing subsectors, and its 130,000 mainly blue-collar workers are distributed globally. Car2 manufactures a variety of mostly low-tech components for the automotive industry. Although it is the leader in many of its markets, it faces high competition. As with Car1, cost competitiveness and flexibility have become crucial to coping with the growing Asian competition. The two Car2 plants under consideration-Car2BE in Belgium and Car2DE in Germany-are highly unionized and similar in size, one with about 2,500 workers, the other with about 1,700 workers (see Table 1). The comparison between Germany and Belgium is particularly interesting with regard to the subject under investigation because these countries feature different union traditions and systems of workplace representation. Furthermore, trade union membership in Germany is much lower than in Belgium, and this is likely to affect collective bargaining outcomes.
The data presented in this study are based on 35 semistructured interviews conducted in the respondents' native languages (German, Dutch, and French). Language issues did not emerge, as the researchers carrying out the interviews were either native speakers or had a near-native ability to speak the respondent's language. The interviews took 60 to 120 min and were recorded, transcribed, and translated into English. In both multinationals, we talked to the Europeanlevel human resources (HR) management to learn more about the companies' strategies, to plant managers to gain insight into the workplaces' particularities, and to several local HR managers to gain an understanding of policies and practices as well as the negotiations leading to their implementation. Furthermore, we interviewed works councillors and trade unionists (from all of the unions represented) in the workplaces to explore their positions and their perceptions of negotiation processes and specialists on TAW at the sector level. The rationale was to talk to all relevant interview partners at the level of the workplace (i.e., those preparing and/ or participating in workplace-level negotiations) to gain an understanding of the factors and processes leading to trade union responses to TAW as outcome. Furthermore, interviews at the sector level helped us to embed the workplacelevel findings in the wider context. Company-based documentary materials and collective agreements were used as secondary sources to complement the interview findings. NVivo was used to structure, code, and analyze the data. The final codebook evolved over time as newly collected data were regularly added, and this led to the development of new nodes or the adjustment of already existing nodes (i.e., merging or dividing nodes). Such adaptions were accompanied by regular discussions within the research team. As all of the members contributed to the coding process, the integration of several perspectives could be ensured. In a first round of data analysis, we focused on the workplace level to come to an understanding of the processes and factors leading to specific trade union responses to agency work. In a second step, we related those findings to the sectoral and national level to embed our findings in the relevant institutional context. Finally, we integrated the levels, as relevant processes and factors could be identified at the different levels. In other words, trade union responses to TAW can only be explained by looking at factors stemming from the workplace, the sectoral, and the national levels.

Trade Union Responses to TAW
The four workplaces principally used TAW to cope with production volume fluctuations and therefore negotiated plantlevel agreements to agree on maximum percentages of agency workers relative to the plants' total headcounts. Specifically, Car1DE, in Germany, negotiated a threshold for blue-collar agency workers of 15%, and 40% during holiday periods. The works council initiated these negotiations to reduce its own workload-otherwise, it would have been legally obliged to individually agree to the use of every single agency worker. Overcoming this bureaucratic system also reduced HR management's workload, and therefore the negotiations went smoothly: Even as a member of the works council you cannot just say that agency work is bad, because you know that the company needs the flexibility to deal with large orders. Of course the situation of agency workers is not the best, but we definitely need an instrument to balance out the fluctuations in daily business. (Works councillor, Car1DE) By allowing a relatively high quota, Car1DE could also benefit from cost savings, as the price of TAW in Germany is relatively cheap compared with that of regular workers (i.e., lower wage costs, fewer and lower benefits, no dismissal protection). This contributes to the survival of plants like Car1DE that are facing cost and flexibility pressures to compete with emerging markets. Therefore, "liberalizing" TAW increased the core workforce's job security at the expense of the agency workers: Agency work was officially introduced to cope with large volumes of orders and to provide flexibility. But now we see that those workers are mainly hired in order to cut costs. (Works councillor,Car1DE) Car2DE, also in Germany, implemented a 20% agency work quota in 2004. Since 2002, the plant had had to cope with intensive restructuring and had lost almost 150 jobs. Declining order volumes and increasing cost pressures brought Car2DE close to bankruptcy. Therefore, the works council initiated the negotiation to trade off its agreement to the relatively high use of TAW against an employment guarantee for the regular workforce: Our labor agreement secures the permanent jobs at this plant. Flexibility is certainly a crucial factor for the survival of the plant. 15-20% of agency work seems to be necessary for the well-being of the company and the permanent staff. Of course, it is the agency workers that suffer from that situation. (Works councillor,Car2DE) As in Car1DE, the allowed percentage of agency workers also contributed to reducing labor costs, which proved to be beneficial to the plant in its difficult economic situation. Thus, the works councils in both Car1DE and Car2DE agreed to concessions. By increasing the use of the relatively cheap and highly flexible agency workers, they contributed to ensuring the core workforce's job security. Yet the core workforce had to make some sacrifices, too. Car2DE abolished its pay premium on weekend work and Car2DE reduced all premiums exceeding the sector-level agreement's provisions. For instance, this meant that the shift premium and various other voluntary premiums were cut back by 50%. While the use of TAW increased the core workforce's security, it contributed to reducing the agency workers' job stability and worsening their working conditions. This became evident during the economic crisis, when agency workers were made redundant as soon as production volumes dropped. As the user firm does not have a contractual relationship with the agency worker, it can flexibly send the worker back to the agency at any time. Moreover, there are no legally binding notice periods and costs such as severance payments. This made laying off agency workers quite easy.
Then we were in the middle of the economic crisis in 2009 and there was the fear that we had to develop a redundancy scheme. We reduced personnel costs by introducing short-time work and we minimized our agency workforce. We did this to secure all the permanent jobs in the location. We only managed that by sitting together arguing about the right way: How can the works council support the company? (Works councillor,Car1DE) Generally, TAW-related plant-level agreements in the German workplaces did not reduce the vulnerable status of agency workers. Apart from their lack of job stability, the agency workers' pay levels proved to be particularly problematic. Although works councils made an effort to introduce pay premiums, they were far too low to close the gap between core and agency workers. In 2004, an agreement entitling agency workers to a gross premium of 1 to 1.50 Euro per hour was concluded in Car2DE, but only applying to workers who had been with the company for at least six months. In practice, the agreement's success was limited, as most agency workers were hired for shorter periods. Furthermore, the premiums can be viewed as "stick and carrot" mechanisms to enhance the agency workers' performance as they would wish to qualify for longer assignments. In 2011, Car1DE introduced team performance pay premiums-also applying to agency workers-on a monthly basis. Every team member was entitled to the same amount of money depending on the whole team's performance, enabling coercive comparisons. This put pressure on the core workers in particular, as they had to enhance their productivity to compete with the agency workforce, because the two groups worked side by side within a team. This again emphasizes the absence of equal treatment in the two German workplaces.
One major reason for this was that core and agency workers were covered by different sector-level collective agreements. Thus, the agreement for the TAW sector was far less beneficial than the one covering the regular workers in the metal sector, as it contained fewer pay premiums for weekend and night work and only marginal Christmas and holiday allowances. Furthermore, the sectoral minimum wage for TAW was still substantially lower than the average hourly wage of regular workers in the metal sector: Employees from agencies only receive about 8 Euro per hour and they do not get any extra payments. You can imagine how difficult it is to motivate someone to do a good job if he sees that all his colleagues get far more money for exactly the same work. (Works councillor,Car1DE) The unequal treatment of core and agency workers was also pronounced in terms of training. While core employees were entitled to regular training classes, agency workers received only on-the-job training in the tasks they should perform and a health and safety briefing at the beginning of their assignment: An agency worker is not part of [Car1DE] but belongs to a temporary employment agency. Therefore, there are no training measures for agency workers apart from showing them how to work on a certain spot in production. (Works councillor,Car1DE) The lack of training provision gave agency workers no opportunity to develop their skills and increase their employability. However, this might be intentional on the part of management, as the chances of being offered fixed term or permanent employment by the user company are rather low. Furthermore, it makes clear that works councillors do not consider agency workers as being part of their workplace and therefore as being part of the workforce they represent. Works councils seemed to be more concerned about the core workforce, rather than the flexible layer of agency workers around them. A possible cause for this could be the specifics of the agency workers' right of participation in the user firm. They are entitled to vote in the social elections after having spent at least three months in the same user firm. Given the fact that agency workers' assignment durations tended to be shorter than this in Car1DE and Car2DE, the possibility of voting was in practice rare. This could explain the works councillors' reluctance to represent them effectively, as agency workers were not potential voters. Overall, the behavior of German works councillors can be associated with what Heery (2004) calls "replacement." They accept agency workers as labor market actors and potential union members but treat them as a buffer that helps to secure the core workforce. Besides, it is obvious that the actions of local works councillors hardly reflect the unions' inclusive sectoral policy approaches.
The situation looked different in both of the Belgian workplaces, as equal pay and working conditions were guaranteed by law and sectoral collective agreements. Local unionists agreed on a TAW threshold in both Belgian workplaces, but they could utilize the favorable regulatory framework to avoid high levels of TAW.
We have to accept TAW because Belgian law permits it. For me, personally, nobody has to work in this place with an agency contract. Because, for the person in question, it is the most miserable situation someone can be in. You have no security of income whatsoever. That's why we need to do something for these people. (Works councillor, Car1BE) Therefore, Belgian unionists and works councillors in the user firms became involved in negotiations to improve the agency workers' working conditions. Specifically, Car1BE implemented a 15% flexibility quota in 2004, consisting of 10% fixed term and 5% agency work. The local unions' major aim was to keep the use of TAW at a low level to avoid its abuse.
Following the agreement, we need to have 85% permanent contracts, 10% temporary, and 5% agency workers. If the number of agency or fixed-term workers is higher based on the yearly average, management has to offer permanent or fixedterm contracts to a certain share of workers. . . . It is based on our system of entering the company, applying to blue-collar workers: you start as an agency worker, then become a fixed-term worker, and finally after two years, you get a permanent contract. (Local unionist,Car1BE) The employment path agreed upon in Car1BE represents an extended probation period for the employer and gives clear prospects of permanent employment to agency workers. Hence, both parties could profit from the agreement.
In 2007, local unionists in Car2BE successfully pushed for the banning of TAW. However, they had to agree to a 20% quota of fixed-term contracts in exchange.
As a trade union we made clear we did not want agency workers. The company does not have a choice because it already has plenty of flexibility with the workers with fixed-term contracts and we decided that we could not agree to everything either. As we know the company's mentality, we simply refused to accept agency workers. If you don't set any limits, they [management] push even further. (Works councillor,Car2BE) As with TAW, fixed-term contracts could be seen as a way to try workers out before offering them permanent employment. Hence, unionists assumed that such arrangements would not threaten the job stability of those workers. Like Car1BE, Car2BE also concluded an agreement on employment paths, stating that white-collar employees would be offered permanent employment after one year of fixed-term work, and blue-collar workers no later than after three years. In both Belgian workplaces, contracts were upgraded over time based on performance indicators and training activities. Union representatives and management locally agreed to provide training for all employees regardless of contractual status. Treating all workers equally was a matter of fact: We would not accept any differences between the situation of workers with a fixed-term contract and those with a permanent contract. (Works councillor,Car2BE) However, the Belgian workplaces' TAW strategy was severely challenged by the financial crisis. When production volumes reached their minimum, both Car1BE and Car2BE reduced their number of agency and fixed-term workers greatly. On one hand, this contributed to making both plants secure but, on the other hand, it made clear that the workplaces' approaches toward TAW were contingent upon growth. A crisis was an unforeseen scenario when the TAWrelated agreements were concluded. As a consequence, the decision to let temporary workers leave was probably overhasty, because it led to the loss of skills in both workplaces. When production volumes rose again, those skills were scarce on the labor market, and therefore Car1BE especially faced problems in finding suitable staff. Car2BE, however, was able to rehire many of its former fixed-term workers: There were many workers with fixed-term contracts-we saw a strong increase in workload. But then we were hit by the crisis and we didn't know what to do. During the crisis they [fixedterm workers] left. Fortunately, many of these workers are back to replace the retirees. (Works councillor,Car2BE) Due to the problem of losing skills, management and works councils in both workplaces are currently rediscussing their strategy toward TAW. Overall, the position of the local unions in Belgium can be associated with what Heery (2004) calls "regulation." Local unions became involved in negotiating plant-level agreements guaranteeing training provision, employment paths, and decent overall working conditions for agency workers. As agency workers are mostly employed for more than three months according to the collectively negotiated employment paths, they are entitled to participate in the user firms' social elections. Consequently, works councils have an interest in representing and improving the working conditions of their potential voters. Thus, they actively engaged in turning vulnerable work into stable and secure work because local unions accepted agency workers as equal. This also makes it clear that local unionists followed the national union policy.

Discussion
The four workplaces' exposure to fluctuations in production volume led to the use of TAW as an instrument to ensure flexibility. Thus, local unionists and works councillors engaged in plantlevel collective bargaining to conclude specific agreements on agency work. However, this also influenced coordination between the national/sectoral trade union policy on the use of TAW and local-level approaches to it. It is remarkable that the differences between the countries are more pronounced than those within the countries. We explain this by taking a heuristic view on the institutional context influencing trade union responses to agency work. Aside from the regulatory setting, collective bargaining, structures of representation, and rights of agency workers are considered.
In Germany, the growing use of opening clauses shifted power from the sector level to the plant level with its dual system of representation based on works councils' codetermination rights (Whittal, 2005). Hence, tensions arose between the sectoral union strategy and the agreements concluded by works councils. As agency workers could rarely use their right to vote in user firms' social elections, it was easy for works councils to accept relatively high levels of TAW. This increased the workplaces' adaptability and ensured the core workforce's jobs and working conditions. The unfavorable regulation of TAW with the absence of the equality principle hampered the works councillors. Hence, the workforce's core-periphery divide grew more pronounced, with the consequent erosion of rights and deterioration of working conditions for temporary workers in comparison with the permanent workforce. Moreover, the disorganized nature of bargaining decentralization in the German context augmented the tension between the unions' sectoral approach and the works councils' local approach toward TAW. IG Metall changed its sectoral strategy on TAW from exclusion to inclusion. At the workplace level, works councillors gained more power, but at the same time, they had to cope with the deregulation of the German labor market. In this respect, in particular, the firms' growing use of temporary contracts and the threat of opting out of collective agreements created pressure. This pressure may explain why the sectoral union's approach was not reflected in the workplaces under investigation. As a result, facing competition and cost pressure, works councillors agreed to concessions contradicting the sectoral union's general policy and following the logic of productivity-coalitions (Heery & Abbott, 2000). However, those concessions benefited only the core workforce and did not reduce the agency workers' vulnerability. Hence, agency workers in both of the German workplaces were unable to benefit from the inclusive sectoral union policy.
Agency work is highly regulated in Belgium, both by law and by collective agreements, without the possibility of deviating from the principle of equality between regular and temporary workers. Local unionists complied with the national union policy by concluding agreements to limit the use of TAW and to improve the job stability and working conditions of the flexible staff. Thus, local trade unions actively engaged in turning vulnerable work into secure work and in avoiding a possible core-periphery division of the workforce. This may explain why Belgian agency workers are not less satisfied with their jobs than permanent workers, according to several studies (Claes, 2005;De Witte & Näswall, 2003). Local unionists engaged in minimizing the use of TAW to avoid a further "liberalization" of its use. This can be explained by the strong union presence at the plant level and the ability of corresponding structures like the union-dominated works councils to foster dialogue. Furthermore, and certainly equally important, the fact that the majority of agency workers were entitled to vote in the user firms could be seen as an extra incentive for works councillors to represent them to gain their votes during the next social elections. In other words, being entitled to vote may have helped agency workers to obtain "representation security" (Standing, 2011) in the user firm.

Conclusion
The growth of TAW has forced trade unions to develop strategies to cope with it both at the national/sectoral and at the workplace level. Even though this article acknowledges the influence of the national/sectoral context, it emphasizes that the practices adopted at the workplace level are pivotal. They shed light on the extent to which and the conditions under which local employee representatives can contribute to improving the working conditions of the overall workforce and to guaranteeing equality between different groups of workers by turning vulnerable work into secure work.
The heterogeneous trade union responses that were observed can be explained by differences in the national regulatory setting, industrial relations structures, systems of employee representation, and bargaining and representation rights for agency workers. Those variables interact with the firm's socioeconomic situation, generating specific dynamics that influence not only plant-level negotiations but also the permanent and the temporary workforce. The article has indicated that national/sectoral union policies toward TAW are not necessarily reflected in workplace-level policies and practices. Membership interests as well as segmented social structures are only two examples of factors potentially affecting TAW-related plant-level collective bargaining.
Agency workers mostly suffer from a lack of job security; however, the evidence presented reveals differences. The Belgian cases especially demonstrate that plant-level agreements can remarkably improve the agency workers' status. Certainly, the membership strength of the Belgian unions contributed to guaranteeing equal pay and working conditions. Furthermore, the high level of coordination between national/sector-and plant-level bargaining ensured that no deviation was possible from sector-level agreements. In addition, the agency workers' right to vote in the user firms arguably helped them to obtain representation by works councils in local bargaining. As a result, unions used their resources to reduce the vulnerability of agency workers. The situation was different in Germany, where the growing use of opening clauses in the context of disorganized bargaining decentralization weakened the works councillors at the plant level. Furthermore, management's threats of job cuts made it very difficult for the councillors to argue for guarantees for the agency workforce. The works councillors' primary aim was to safeguard the core workforce's employment. They hardly represented the agency workers in the user firm due to the short duration of these workers' assignments. Furthermore, the absence of the equality principle weakened the agency workers' position and therefore their rights remained limited and their status remained one of vulnerability.
The article contributes to our understanding of the effects of workplace-level bargaining and structures of representation on the job stability and working conditions of agency workers. It also demonstrates nuances in employee representatives' local-level approaches and shows that some agency workers are more or less vulnerable than others. Besides the regulative context, this strongly depends on local agreements that potentially govern matters like training provision, pay levels, or employment transitions. Furthermore, strong collective representation rights for agency workers in the user firm contribute to the improvement of their situation as well. Thus, not only do the national legal framework and sectorlevel regulation play a role in this respect but so do the practices local employee representatives negotiate at the workplace level. Finally, to understand these practices, it is also relevant to explore the extent of collective rights for agency workers in the user firms.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.