Dualism or solidarity? Conditions for union success in regulating precarious work

Coding contextual factors

We coded each article across a variety of dimensions (see Online Appendix for abridged coding scheme). First, we coded contextual factors for each article and its cases, including country, level of analysis, sector and whether migrant workers were present. We identified a bias for Centre/Western Europe, which represented 48 percent of the European cases. The modal country is Germany, representing roughly a quarter of total cases (see Figure 1). Between 15 to 17 percent of cases are based in the Nordic countries, UK and Ireland, and the Mediterranean. Only 4 percent of cases focused on Central and Eastern Europe. An overt emphasis on Centre/Western Europe may be problematic insofar as country-level factors (such as institutional configurations) are a strong predictor of union success regulating precarity. In addition, the frequent inclusion of German cases in comparative studies may bias results. For example, these cases may be more likely than non-German cases to focus on the dynamics of dualism that are considered more ‘typical’ in Germany. We discuss this further in our conclusions, as it may influence the generalizability of findings.

OPEN IN VIEWER

Figure 1.

Distribution of cases by country.

Most articles compare or analyse cases at either the workplace/company (38 percent) or sector level (38 percent). In terms of industry focus, the largest number of cases (49 percent) was in the service sector, followed by 37 percent in manufacturing and 10 percent in construction. An additional 4 percent of cases did not reference a specific sector but instead focused on outcomes occurring cross-sectorally at a local, regional or national level.

To facilitate our efforts to ‘map’ the relationships between case-level variables and observed outcomes, we also coded each article across a variety of potentially influential factors, including institutional environment, union and employer strategies (inclusive vs exclusive), strength of collective worker identity and structural conditions, among others. We present an abridged version of this coding scheme in the Online Appendix (full coding scheme available upon request).

Coding outcomes

Coding outcomes proved to be the most challenging aspect of our process, for both theoretical and practical reasons. From a theoretical perspective, there is a lack of scholarly consensus regarding what constitutes precarity (Chhachhi, 2014). In the absence of a universal definition, identifying ‘success’ or ‘failure’ in regulating precarity becomes a subjective endeavour. For the purposes of our analysis, we consider precarious work to be employment that is ‘characterised by a high degree of worker insecurity and instability in an employment relationship or labour market’ (Doellgast et al., 2018: 1). In turn, precarity reduction can take many forms, including but not limited to increased pay, job security and regularity. Unions can engender these improvements for core/standard workers, for example, through renegotiating provisions of an existing contract, or for peripheral/non-standard workers, by extending the collective agreement to these groups or otherwise operating on their behalf. Furthermore, given the increasing threat to standard work posed by institutional erosion and market liberalization, we also consider unions to have succeeded in regulating precarious work when they maintained a lower level of precarity; for example, through resisting employer demands for concessions.

With these parameters in mind, we established the following coding scheme for recording each article’s outcomes in terms of precarity for both peripheral workers and core workers:

The comparative case studies we reviewed often show different combinations of these ‘precarity reduction’ outcomes. For example, Larsen and Mailand (2018) show that the risk of precarious work across three Danish industries was lowest in hospitals and highest in cleaning, due to hospitals’ higher union density and stronger collective agreement coverage. They found successful initiatives by social partners to reduce atypical employment and improve conditions across their three cases; however, there were more of these initiatives where precariousness was highest (the cleaning industry). All three outcomes were found in 18 percent of the articles, while 4 percent found improvements only for internal workers and 13 percent found no reduction in precarity across cases. Thus, most studies find some evidence of union success in reducing precarious conditions, even if this is based on trade-offs. Taken together, these findings contradict the rather blanket pessimism in some analyses concerning possibilities for regulating precarity through winning improvements or fighting to defend past gains.

Next, we sought to evaluate the extent to which unions successfully engaged with peripheral workers in the context of a dualized or dualizing labour market. Analysing the conditions that stymie union efforts in this regard provides additional insight into the growth of dual labour markets, even in historically labour-friendly institutional contexts. While some of the articles in our database also contain cases in which precarity reduction efforts targeted members of the core workforce, for purposes of this analysis, we focused on cases where there was a clear division between core and peripheral workers. We coded each of these cases as showing evidence of dualism, solidarity or ‘failed’ solidarity, which we defined as follows:

A further 16 percent of cases were coded as showing none of these three outcomes: most commonly because there was no or weak evidence of union action, usually in comparison to matched cases where unions did act with success or failure, or because the case did not assess conditions for peripheral work or contracts.

The most common form of dualism was based on actions or campaigns in which the union negotiated to reduce the use of outsourcing or agency work to secure internal jobs – often involving concessions (Ahlstrand, 2015; Doellgast, 2008; Håkansson et al., 2020) and sometimes accompanied by direct termination of agency worker contracts (Doerflinger and Pulignano, 2018). In other cases, the union is reported as ‘prioritizing’ the core workforce or neglecting precarious worker groups (Kornelakis and Voskeritsian, 2018). These studies often report reduced precarity based on some measures, such as job security, but with trade-offs such as increased pay precarity – so while we label cases ‘dualism’ where the core workforce was prioritized, outcomes are often ambivalent for these workers in the longer term. In some articles, the dualization of bargaining outcomes in a sector was the broader focus of the analysis, as in Kahancová and Szabó’s (2015) study of the Hungarian and Slovakian health care industries.

We took an inductive approach to the coding process described above, based on an initial summary of findings in different categories paired with simple codes, and then developing further coded variables in the database iteratively as we uncovered patterns that we had not a priori anticipated. For example, our initial coding scheme allowed for only two possible outcomes relating to union strategies towards peripheral work: dualism and solidarity. However, after conducting a first read of the articles, we determined that this categorization masked the potential disconnect between union strategy and outcome, and we chose to add the third possible outcome of ‘failed’ solidarity.

We used a similar strategy to identify the constellations of factors or ‘paths’ associated with dualism, solidarity and failed solidarity. Once we had established our final coding, we examined patterns among cases with similar outcomes through mapping relationships between coded variables – akin to the method of ‘grounded theory articulation’ described by Gioia et al. (2012). This exercise was by necessity inductive; our intention was not simply to reiterate the findings of this body of research, but rather to articulate novel patterns that were made visible once the relevant studies had been aggregated and coded.

Two paths to dualism

As described above, 26 percent of the cases in our database represented dualistic outcomes. Based on a comparison of the findings across those cases, we identified two broad paths to dualism. The first combines fragmented institutions and lack of solidarity across unions. The second combines partnership-oriented union identities with unfavourable structural conditions. We see different combinations of these four factors across the dualism cases, but authors tended to focus on one of these two paths in explaining outcomes.

The first path is most consistent with a ‘vicious circle’ in which institutional fragmentation undermines solidarity in the labour movement, allowing employers to exploit divisions or disagreements between unions and/or works councils to negotiate concessions (Doellgast et al., 2018). In some articles, the focus is primarily on employers, who exit formerly inclusive institutions or exploit existing loopholes (Jaehrling and Méhaut, 2013). However, many of the studies describe divisions in the labour movement as a central factor in exacerbating dynamics of expanding precarity. Helfen et al. (2020) link dualizing employment relations in one German airport to ongoing disagreements between worker representatives, in the context of fragmented collective agreements. In other cases, the presence of competing unions in a given workplace or sector undermined the negotiation of comprehensive collective agreements. In their study of the telecommunications sector in Germany, Denmark, Austria and Sweden, Benassi et al. (2016) show increased dualization in the first two countries, an outcome they attribute to both a lack of coordination among competitor unions and exit options specific to sectoral institutions. An important source of division is between native and migrant workers, precluding the development of a cohesive representation structure. In some cases, unions failed to ensure democratic representation structures and access to leadership opportunities for migrants, or even actively fronted opposition to independent organizing action by migrant workers (Lillie and Sippola, 2011).

Some authors discuss change over time, as growing conflicts between unions allow further weakening of inclusive institutional frameworks. Benassi et al. (2019) show that inter-union conflict in the Italian metalworking sector enabled employers to negotiate a separate collective agreement de-regulating contract work with two of the more ‘moderate’ unions representing the workforce. In other cases, the institutional gaps precede growing divisions in the labour movement. Pulignano et al. (2017) demonstrate that opening clauses in collective agreements in the German manufacturing sector allowed employers to renegotiate certain terms at the company-level, hobbling the efficacy of sector-level bargaining and encouraging increased divisions across (and competition among) works councils. Collectively, these cases show the importance of coordination between different institutional domains as a key factor preventing dualization (Pulignano et al., 2016). It is particularly noteworthy that most of the dualism cases falling within this ‘path’ are based in Germany, which is more generally over-represented among dualism cases in the database.

A second path to dualistic outcomes combines partnership-oriented union identities (i.e. those predicated on maintaining cooperation with the employer/employers’ association) with unfavourable structural conditions in the country, region or sector. A partnership identity contributes to unions’ willingness to negotiate concessions with the employers, the impacts of which were often felt most acutely by already-precarious workers (Dorigatti, 2017; Gasparri et al., 2019). However, the impetus for concession bargaining that resulted in dualization often came from structural forces beyond the control of unions. Lillie (2012) shows that the conciliatory approach taken by Finnish unions in the shipbuilding industry stemmed largely from the increase in product market competition. Aware of the fact that production could be moved elsewhere, the union in this case chose to negotiate concessions that resulted in dualized employment relations rather than risk the employer shifting the site of production. Pulignano et al. (2015) similarly show that product market characteristics (level of competition and the relative skill of the work involved) were highly influential in unions’ willingness to engage in concession bargaining in Germany, Belgium, Italy and the UK, four countries with distinct institutional configurations. The slack labour market associated with the global recession beginning in 2008 also provided employers with the leverage needed to negotiate concessions from unions, resulting in dualistic collective agreements and increased precarity for peripheral workers, such as agency or part-time employees (Ahlstrand, 2015; Doerflinger and Pulignano, 2018). Public sector austerity was particularly important in these dynamics: for example, in the hospital sector in Hungary (Kahancová and Szabó, 2015). Similarly, EU-mandated market liberalization in some sectors, such as telecommunications, fundamentally reorganized industry structure and placed unions at a distinct negotiating disadvantage vis-à-vis employers (Sørensen and Weinkopf, 2009).

Two paths to solidarity

While some degree of dualism is present, the most frequently observed outcome across our 208 cases was solidarity, in which the union was able to improve working conditions for the peripheral workforce. We also identified two primary paths by which unions achieved this outcome: a conflict-based path and a social partnership path.

A conflict-based path appears more likely to occur either in the absence of institutional power or in contexts in which institutional power is in decline. This included countries where institutional protections for unions have been comparatively weak, such as the United Kingdom, but also countries and sectors in which institutional erosion has weakened unions’ traditional power sources, notably in Germany. In such cases, the union was more likely to engage its membership in combative action against the employer, organizing strikes or protests and even filing lawsuits or legal motions (Alberti, 2016; Dorigatti, 2017). These efforts succeeded to the extent that unions were able to mobilize associational power, both in terms of worker-to-worker identification and cohesion (Arnholtz and Refslund, 2019; Tapia and Turner, 2013), as well as in terms of the strength of coalitions with external groups.

In several success cases, unions sought to build alliances with community organizations and mount public awareness campaigns to exert pressure on employers (Alberti and Però, 2018; Benassi and Dorigatti, 2014). They also often contained an implicit or explicit focus on ‘building solidarity’ across different groups of workers and with the wider public. In one example, Simms and Dean (2015) attribute two successful cases of mobilizing contingent workers in UK higher education and entertainment to building solidarity among contingent workers, between contingent and core workers, and with audience members. This required overcoming resistance in some cases from the core workforce through union-led research and education that emphasized shared interests.

Union agency was particularly important in adopting more inclusive (or exclusive) strategies for migrant workers, where institutional power is typically lower due to their exclusion from national laws and bargaining structures. In several cases, unions adopted a stronger social justice orientation and sought to increase migrant worker voice within union structures, with some success. Alberti et al. (2013) describe the UK union Unite’s success in organizing migrant workers as part of the Justice for Cleaners campaign, a result the authors attribute partially to the union’s establishment of a separate structure to address migrant workers’ unique needs. Connolly et al. (2017) also seek to establish a link between democratic representation for migrant workers and levels of migrant worker organization. In other cases, weaker institutional power may actually encourage stronger ‘bottom-up’ organizing among migrants. Marino (2015) argues that the more decentralized and fragmented union structures in Italy (compared to centralized structures in the Netherlands) better supported the development of migrant worker-driven initiatives.

By comparison, the social partnership path was more often utilized by unions where they retained some semblance of institutional strength. This path allowed unions to leverage existing institutions to widen their jurisdiction and codify new institutional protections for peripheral workers (Håkansson et al., 2020; O’Brady, 2019; Shire et al., 2009; Van Jaarsveld et al., 2009). In particular, where sectoral-level bargaining remained strong, unions were able to negotiate collective agreements that extended protections to workers that have historically been excluded from coverage, including agency, part-time and subcontracted workers (Benassi et al., 2016; Holst, 2008). Sector-level bargaining constrained employers’ ability to exploit exit options or seek concessions and was therefore a key source of leverage for unions seeking to reduce precarity in a given industry (Doellgast et al., 2016).

At the same time, the articles show that associational power often continued to be important to union success in regulating precarious work under conditions of institutional strength. Unions typically did not use their associational power to engage in the kind of direct confrontation with the employer used in the conflict-based path. However, they still mobilized heterogeneous local power resources to sustain or activate institutional power (Pulignano and Keune, 2015; Pulignano and Signoretti, 2016). For example, Benassi (2017) shows that conditions for temporary workers in German auto plants did not uniformly improve following strengthened national legislation and sectorally bargained rules. Instead, different bargaining outcomes at plant level were explained by the local socio-economic context and works councils’ plant-level associational power resources.

We also observe that in both the conflict-based path and the social partnership path, unions often appealed to the state to re-regulate or legislate protections for precarious workers. These appeals could take place from a position of institutional strength, where unions, as social partners, were given some voice in government decision-making (Arnholtz and Refslund, 2019; Durazzi et al., 2018; Rathgeb, 2018). Unions also launched these appeals in the absence of institutional strength, instead relying on grassroots mobilization to get the attention of local, regional and national legislators (Doellgast et al., 2009; Marino et al., 2019; Tapia and Turner, 2013). In one example, Jaehrling et al. (2018) compare three city-level cases where unions succeeded in encouraging socially responsible procurement. They find that these successes could be attributed to both pragmatic alliances among local progressive politicians, unions and employers, as well as trade union mobilization and a favourable public discourse. However, the clauses led to most significant improvements in pay and benefits (and reduced precarity) in Denmark and least in the UK, due to differences in each country’s existing institutions that enforce minimum standards.

‘Re-regulation’ when applied to migrant workers most directly involved the ability to extend collective agreements to these groups. In some countries, bargaining extension has allowed unions to prevent wage dumping and exploitation of migrants (Alsos and Eldring, 2008; Hardy et al., 2012; Refslund, 2018). Connolly et al.’s (2014) comparative study of migrant worker organizing in the Netherlands, Spain and the UK illustrates most comprehensively the interaction between institutional setting and union identity or orientation – which they argue takes on different configurations of race-based, social rights-based and class-based logics in each country. In line with our other ‘solidarity’ cases, the authors find that institutional power is associated with more social partnership approaches towards solidarity, and that appeals to the state are central but also take different forms across countries.

Failed solidarity

Solidaristic efforts by unions do not always translate to solidaristic outcomes. Why are unions unable to reduce precarity for peripheral workers, even if this is their stated aim?

A key theme across the cases involving failed solidarity is union inability to create worker-to-worker identification and collaboration across age groups (Kretsos, 2011), occupations (Murphy and Turner, 2016) and even countries. For instance, Greer et al. (2013) study the collapsed European Migrant Workers Union, an initiative spearheaded by German union IG BAU and whose ultimate failure the authors attribute to the lack of buy-in from affiliated national unions. In this case, the failure to build solidarity stems from the myopia of individual unions, rather than union members, but the result is the same: unchanged levels of precarity despite union efforts to combat it.

This said, the failure to build associational power cannot be attributed solely to the actions of unions themselves, which often face conditions highly unfavourable to organizing. Even in cases where unions are able to build cross-worker identification, the structural or institutional context may prevent them from leveraging this associational power to make meaningful gains for workers. This is most evident when looking at the cases that involve migrant workers, who tend to be at extreme structural disadvantage compared to native workers. Of the cases that show failed solidarity, 56 percent involve migrant workers. Several themes emerge from our database of articles that provide a basis for understanding the prevalence of migrant workers in our failed solidarity cases.

First, and perhaps foremost, the lack of institutional protections for migrant workers and their concentration in non-standard or informal employment arrangements poses distinct challenges to organizing, even in countries with a strong regulatory framework. This is particularly the case for posted work (e.g. Larsen and Mailand, 2018; Wagner and Refslund, 2016). Berntsen and Lillie (2016), for instance, show that in spite of a supportive institutional environment, unions in the Netherlands were largely unsuccessful at recruiting and protecting migrant workers at two focal construction sites. Although the collective agreement ostensibly extended to cover migrant workers in their case, the complexity and opacity of subcontracting chains at the site meant little practical enforcement was possible. Similarly, in their study of the CGT’s participation in the sans papiers movement in France, Barron et al. (2016) show that the union was unable to effectively represent migrant workers for whom striking was not a legally protected action. These studies show a multiplicative effect of migrant status and non-standard status on precarity.

A second limiting factor on union efficacy is the structural dependency of migrant workers on their employer. Several articles in our database demonstrate employers’ capacity to leverage this dependence to create a climate of fear, which makes organizing migrant workers more challenging for unions (Lillie and Sippola, 2011; Wagner, 2015). In a similar vein, ‘hyper-mobile’ migrant workers may see little benefit to organization, given the shorter duration of their employment contract (Berntsen and Lillie, 2016). Unions interested in representing temporary migrant workers may therefore find themselves primarily in a servicing role, for instance, offering legal counselling to migrant workers seeking back pay or unpaid wages from their employer (Alberti and Però, 2018; Meardi et al., 2012). While this produces tangible gains for immigrant workers in some cases, servicing without concomitant organizing can be financially unsustainable, ultimately dooming the union’s initiative (Greer et al., 2013).

A third limiting factor may be union strategies themselves, that is, a focus on ‘top-down’ inclusion, rather than ‘bottom-up’ solidarity actions and political education. There is some evidence from the case studies that worker-to-worker solidarity is a central precondition for successful organizing, both between native and migrant workers (Arnholtz and Refslund, 2019; Keune and Pedaci, 2020) and within migrant communities (Jiang and Korczynski, 2016; Marino, 2015). However, research findings suggest that union campaigns, even when focused on precarious migrant workers, often fail to build solidarity between natives and migrants – thus limiting access to ‘associational power’ from coalitions (Alberti, 2016; MacKenzie, 2010). These cases suggest that racism or prejudice among existing union members, or divisions within migrant communities, may play a central role in the high rate of union failure in reversing a ‘vicious circle’ among these groups. Conversely, political education focused on actively overcoming racism and anti-migrant prejudice is central to the project of reducing precarity. These worker-level dynamics receive surprisingly little attention in the CER research in our database. This indicates that additional research is needed to unpack the dynamics of worker identity and identification, as both outcome and input into union and employer strategies.